90-Day Mathematics Examination Preparation Plan

A 90-day plan is not three months of doing more Mathematics. It is three months of changing what the student can reliably produce.

That distinction changes the entire revision strategy.

If a student begins ninety days out with unstable algebra, weak retrieval and poor paper control, simply adding paper volume can make those weaknesses more visible without making them smaller.

A better plan changes emphasis as the examination approaches.

The 90-day architecture

1. 90 to 61 days: Repair. Find the highest-cost weaknesses and rebuild them.
2. 60 to 31 days: Integrate. Make repaired knowledge retrievable across mixed questions.
3. 30 days to examination: Perform. Convert capability into reliable paper execution.

The phases overlap slightly. A student will still do some papers early and may still need repair late.

But the dominant job should change.

Before Day 90: establish the real starting point

Do not build the plan from memory alone.

Collect evidence:

• recent school tests;
• weighted assessments;
• preliminary papers where available;
• homework;
• corrections;
• timed practice;
• questions the student could not begin independently.

Then separate lost marks into useful categories.

• Concept: the mathematical idea is not understood.
• Prerequisite: an earlier skill is blocking present work.
• Recognition: the student knows the method but does not see when it belongs.
• Retrieval: the method is unavailable after time has passed.
• Execution: the route is correct but algebra, arithmetic or notation breaks.
• Transfer: the student succeeds only when the question resembles the example.
• Verification: unreasonable answers survive.
• Time: capability exists but too little of the paper is completed.

For a deeper version of this step, read How Mathematics Diagnosis Works.

Days 90–81: stop the largest leaks first

The first ten days should not attempt to fix everything.

Look for the weaknesses with the widest effect.

In Secondary Mathematics, this is often algebra.

A sign-control problem can damage equations, graphs, coordinate geometry and trigonometry. Weak fraction manipulation can spread through ratios, formulas and A-Math. Poor problem representation can affect many chapters that look unrelated.

The first repair target should therefore be chosen for leverage.

Do not ask, “Which chapter is weakest?” Ask, “Which weakness is costing marks in the largest number of places?”

Days 80–71: rebuild and prove the repair

A repair is not complete when the student understands the explanation.

It is complete when the student can use the repaired idea independently after the explanation is gone.

Use a short loop:

Explain → attempt → correct → retry → wait → retrieve → vary the question.

The delayed return matters.

A method that works only while the correction is fresh has not yet become dependable examination knowledge.

Days 70–61: reconnect the repair to the current syllabus

Foundation repair can fail if it remains isolated.

A student may spend a week revising algebra drills and still not recognise when algebra is needed inside geometry or trigonometry.

So the repair must travel.

• Use repaired algebra inside graphs.
• Use repaired fraction control inside percentage or trigonometry.
• Use repaired representation inside unfamiliar word problems.
• Use repaired checking habits inside timed sections.

The first month ends when the student has fewer structural weaknesses and stronger links between older and current Mathematics.

At 60 days: the job changes

With roughly two months remaining, the centre of gravity should move away from pure repair and towards availability.

The student now needs to retrieve Mathematics without chapter labels, worked examples or immediate reminders.

This is where mixed practice becomes increasingly important.

Days 60–51: mix topics deliberately

Topical practice answers one question before the student begins: which method family is likely to matter?

Mixed practice removes that clue.

The student must now:

• identify the structure;
• retrieve a candidate method;
• decide whether it fits;
• begin without a prompt;
• change route if the first attempt fails.

This is closer to the real examination job.

Days 50–41: introduce examination sections, not only full papers

Full papers are useful, but they are not the only way to train examination performance.

Shorter timed sections can target specific behaviours:

• ten questions with strict timing;
• mixed algebra and geometry;
• a difficult-question block;
• a checking-only exercise;
• a section where the student must decide when to skip and return.

This is useful because it allows one examination behaviour to be trained without the noise of an entire paper.

Days 40–31: begin full-paper cycles

By this stage, the student should be doing enough full-paper work to reveal realistic performance patterns.

But each paper should feed the next action.

Paper → classify lost marks → choose one or two repair priorities → targeted work → retest.

A paper without a post-mortem is often only a score-producing event.

A paper with a good post-mortem becomes a teaching instrument.

For the wider paper-performance layer, read Mathematics Examination Craft.

At 30 days: stop trying to rebuild the whole student

With one month remaining, revision needs discipline.

There may still be weaknesses, but time is now a scarce resource.

The student should prioritise:

• high-frequency weak areas;
• high-cost recurring errors;
• paper completion;
• checking;
• difficult-question recovery;
• stable sleep and routine.

Trying to master every obscure corner of the syllabus can damage the strong majority of the paper.

Days 30–21: protect reliable marks

Students often focus so heavily on their hardest questions that they neglect the marks they should already own.

The final month should make accessible marks highly dependable.

Ask:

• Which questions should this student almost never lose?
• Which careless patterns still leak easy marks?
• Which common formulas are still retrieved too slowly?
• Which algebra transformations still create unnecessary risk?

Reliability before heroics.

Days 20–11: train paper management

At this stage, the student should practise how to behave when the paper does not go perfectly.

A strong examination student is not the student who never gets stuck.

It is the student who can get stuck without allowing one question to damage the rest of the paper.

• recognise when a question is consuming disproportionate time;
• leave a visible marker;
• move on cleanly;
• return later;
• recover without emotional collapse;
• check high-risk answers first.

Days 10–4: reduce novelty

The final ten days are usually a poor time to redesign the entire study system.

Use familiar resources.

Return to known error families.

Use selected papers and short retrieval.

Keep sleep stable.

Confidence should come from evidence that the student can execute the system they have already built.

Days 3–1: taper rather than panic

The last days should protect clarity.

Heavy last-minute volume can create fatigue without adding much durable capability.

A calm final routine might include:

• short retrieval;
• selected familiar questions;
• formula and error-family review;
• paper strategy reminder;
• normal meals and sleep;
• stopping at a sensible time.

The purpose is not to feel finished.

No student ever feels they have exhausted Mathematics.

The purpose is to arrive able to use what has been built.

How many papers should a student do?

There is no useful universal number.

The better question is whether each paper is producing change.

If the same three mistakes appear across six papers, the student has not completed six useful cycles.

The student has repeated the same unresolved state six times.

Paper volume should rise only when the student is extracting enough information from each paper to improve the next one.

What parents should track across the 90 days

• Are repeated errors decreasing?
• Is the student starting mixed questions more independently?
• Is older Mathematics staying available?
• Is paper completion improving?
• Are difficult questions causing less disruption?
• Is checking becoming more selective and useful?
• Is the difference between supported and independent work shrinking?
• Is the total workload sustainable?

Marks matter, but these behaviours tell you whether the mark is becoming more likely to hold.

When tuition helps during the 90-day runway

• the student cannot identify why marks are being lost;
• one weak prerequisite is affecting several topics;
• the student is doing many papers without improvement;
• timing is poor but the source of slowness is unclear;
• the child understands corrections but repeats the same mistakes;
• mixed questions are much weaker than topical work;
• the student needs an external structure for paper post-mortems and retesting.

When more tuition may not help

If the student is already overloaded, another class can remove the sleep, recovery and independent practice needed for consolidation.

If the student knows exactly what to do and is executing independently, more instruction may add little value.

The final phase of preparation should include release.

Frequently Asked Questions

Is 90 days enough to improve Mathematics?

Yes, meaningful improvement is possible, especially when the plan targets active weaknesses rather than attempting to relearn everything equally.

What should change at 60 days?

The emphasis should shift from mainly repairing weak foundations towards mixed retrieval, method selection and examination-style application.

What should change at 30 days?

The emphasis should shift again towards reliable marks, full-paper execution, timing, checking, error reduction and protecting confidence and sleep.

Should the student still learn new things in the final month?

Yes where necessary, but new learning should be selective. High-value gaps deserve attention; low-probability novelty should not destabilise secure parts of the syllabus.

Should we do one paper every day in the final month?

Not automatically. Papers should be balanced with post-mortem analysis, targeted repair, retrieval and recovery. More papers are useful only when they improve future papers.

What this 90-day page owns

This page owns the 90-day Mathematics examination preparation timeline: when the preparation job should change, what evidence should trigger the change, and how repair, retrieval, mixed practice, papers, checking, recovery and tapering fit together. BTT already has deeper owners for past-year papers, mock examinations, paper post-mortems, getting unstuck and poor-prelim recovery.

The countdown is therefore an operating system, not a pile of revision tips. It should tell the learner and tutor what the next block is for, how to recognise when a repair is stable enough to leave, which activities are now highest-value, and when the final task is simply to protect a functioning mathematical system.

Repair → retrieve → mix → perform → protect.

The 2026–2027 Singapore examination context

Students graduating in 2026 remain on the relevant existing GCE routes. From the 2027 graduating cohort, the Singapore-Cambridge Secondary Education Certificate replaces the separate GCE N(T), N(A) and O-Level certificates under Full Subject-Based Banding. SEAB states that students sit subjects at the respective G1, G2 or G3 level, while the certificate continues to be jointly examined and awarded by SEAB, MOE and Cambridge International Education.

For 2027 school candidates, SEAB lists Mathematics as K110 at G1, K210 at G2 and K310 at G3. Additional Mathematics is listed as K232 at G2 and K341 at G3. Any 90-day plan must therefore begin by confirming the learner’s actual subject level, subject combination and cohort. Use the official SEC overview, G1, G2, G3 syllabus lists and MOE’s Full SBB / SEC announcement.

Days 90–61: Repair and reconnect

Dominant job: Use the first month to reduce structural leakage before paper volume rises.

Start from several pieces of evidence, rank weaknesses by transfer value and downstream cost, and select only a few active repair targets.

Repair at the smallest useful scale, then reconnect the repair to current or examination-style Mathematics immediately.

Delay and vary verification. A repair is not secure merely because the student succeeds while the explanation is fresh.

Move on when the largest structural gaps are repaired or contained and mixed work begins to expose selection or performance issues rather than obvious missing content.

These phases overlap. A strong student may use papers earlier and a late high-value gap may still need repair. The value of the architecture is that the dominant training problem stays explicit.

Days 60–31: Retrieve, mix and integrate

Dominant job: Make known Mathematics available without chapter labels, worked-example cues or tutor prompts.

Increase delayed retrieval and controlled interleaving. Ask students to classify the problem and justify a candidate method before calculation.

Use examination-style sections to isolate reading, method selection, execution, checking and recovery before relying heavily on complete papers.

Begin serious full-paper cycles only when the paper can reveal integrated performance rather than simply a list of unlearned chapters.

Move on when the learner can retrieve most high-frequency methods from mixed conditions and the error map has narrowed.

These phases overlap. A strong student may use papers earlier and a late high-value gap may still need repair. The value of the architecture is that the dominant training problem stays explicit.

Days 30–1: Perform and protect

Dominant job: Convert capability into stable examination execution while reducing unnecessary novelty.

Move secure topics to maintenance, keep fragile topics alive through short retrieval and repair only high-value remaining gaps.

Use papers inside a closed loop: diagnose, repair, verify and test again. Train pacing, leave-return, checking and recovery.

As the final ten days approach, reduce novelty, protect sleep and keep the routines familiar.

The final target is a learner who can read, choose, execute, check and recover with a stable routine.

These phases overlap. A strong student may use papers earlier and a late high-value gap may still need repair. The value of the architecture is that the dominant training problem stays explicit.

Thirty-two operating systems inside the 90-day plan

Baseline and error map

Purpose. The role of this system is to build a precise starting state instead of a vague sense of weakness. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. collect several scripts, homework and one mixed diagnostic; classify concept, prerequisite, recognition, retrieval, execution, transfer, verification and time. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. use the map to decide which mechanisms still need active repair and which can move into retrieval or mixed work. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. compress the map into a personal risk register for paper performance. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the active list should narrow over time even if new one-off errors appear. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. keeping every mistake forever creates noise and weakens prioritisation. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving baseline and error map. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For baseline and error map, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Prerequisite repair

Purpose. The role of this system is to fix earlier Mathematics only when it blocks current learning or affects several later topics. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. identify high-leverage dependencies such as fractions, negative numbers, ratio, algebra, equations, graphs and diagram reading. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. reconnect repaired dependencies to mixed current questions so the student learns where the skill belongs. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. late repair should be selective and judged by likely mark return and transfer value. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the skill should survive delay, a changed surface and reduced prompting. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. re-teaching the whole earlier syllabus can create permanent lag. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving prerequisite repair. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For prerequisite repair, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Algebraic reliability

Purpose. The role of this system is to treat algebra as infrastructure rather than a single chapter. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. separate sign control, expansion, factorisation, substitution, equations and symbolic organisation. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. embed algebra inside graphs, geometry, trigonometry or A-Math so it becomes usable under integration. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. protect fluent algebra under time because paper performance often depends on accurate symbolic execution. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. error recurrence should fall across several topics, not only algebra worksheets. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. topic drills can hide cue dependence when the learner never decides when algebra is needed. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving algebraic reliability. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For algebraic reliability, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Number, fractions and proportion

Purpose. The role of this system is to stabilise multiplicative reasoning across percentage, rate, scale, formulas and applied problems. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. test relationship understanding rather than separate memorised procedures; repair fraction and ratio control where necessary. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. mix proportion with non-proportion situations so the learner must classify before calculating. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. use estimation and reasonableness checks under paper conditions. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the learner should recognise proportional structure without a chapter label. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. formula memory without relationship sense can produce correct-looking work on the wrong model. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving number, fractions and proportion. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For number, fractions and proportion, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Graphs and representations

Purpose. The role of this system is to make the learner move between words, tables, algebra, coordinates and visual graphs. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. diagnose whether plotting, interpretation, axes, intercepts, gradient or variable meaning is the weak link. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. mix representations so the learner must translate rather than treat each form as a separate topic. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. use graph consistency as a checking tool in papers where appropriate. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the learner should explain what a point, intercept or change means in context. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. students can draw accurately while still misunderstanding the represented relationship. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving graphs and representations. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For graphs and representations, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Geometry, trigonometry and diagrams

Purpose. The role of this system is to teach diagrams as representations of constraints rather than pictures to trust. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. repair property knowledge, diagram reading, units and translation from words to labelled structure. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. integrate algebraic and trigonometric reasoning with geometry so method choice becomes explicit. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. train efficient annotation and verification under time. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the learner should distinguish given information from inferred relationships. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. rushing to a formula before reading conditions creates errors that later calculation cannot repair. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving geometry, trigonometry and diagrams. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For geometry, trigonometry and diagrams, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Statistics, probability and data

Purpose. The role of this system is to protect interpretation as well as calculation. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. diagnose whether errors arise from language, sample/population ideas, averages, spread, probability structure or computation. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. mix data questions with other topics so careful reading remains necessary. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. preserve interpretation and units under time while using calculator functions accurately where relevant. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the learner should explain what a statistic means, not merely produce it. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. calculator accuracy can hide a wrong interpretation of the question. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving statistics, probability and data. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For statistics, probability and data, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Retrieval

Purpose. The role of this system is to make Mathematics available after time has passed and without immediate cues. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. schedule short returns to repaired skills after increasing gaps. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. use retrieval across mixed topics so memory and method selection begin to work together. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. keep secure topics alive cheaply while scarce time moves to fragile mechanisms. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the learner should reproduce a method after delay and explain why it applies. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. re-teaching every forgotten method wastes time when the weakness is retrieval. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving retrieval. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For retrieval, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Spacing

Purpose. The role of this system is to distribute learning so the student repeatedly reconstructs knowledge. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. set deliberate future return dates for repaired mechanisms instead of trusting same-day fluency. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. lengthen intervals for stable skills and shorten them for fragile skills. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. use spacing to protect a broad syllabus without revising every topic every day. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. fewer complete resets should be needed as the countdown progresses. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. massed practice creates temporary fluency that may disappear when the worksheet closes. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving spacing. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For spacing, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Interleaving

Purpose. The role of this system is to train the decision about which method belongs. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. begin with two or three related methods once procedures are stable. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. widen the mixed set so the learner recognises structure rather than surface. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. use full-paper order as the final form of interleaving when ready. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. method-selection errors should fall and first moves should become more independent. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. mixing everything too early can create noise instead of useful discrimination. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving interleaving. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For interleaving, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Worked examples

Purpose. The role of this system is to use models to form correct schemas without creating permanent dependence. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. show complete examples only where genuinely new, then fade steps and ask the learner to reconstruct. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. compare multiple valid routes and changed examples so principles become visible. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. reserve full modelling late for high-value repair or route comparison. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the learner should solve related work with the model closed and later retrieve independently. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. open worked solutions throughout practice turn problem solving into recognition. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving worked examples. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For worked examples, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Prompt burden

Purpose. The role of this system is to measure how much of the route still belongs to the tutor. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. track whether the tutor names the topic, supplies the method, gives the first step or signals an error. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. protect independent first attempts and use graduated hints only after observing the learner. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. complete serious timed work and full papers without live intervention. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the minimum prompt level for familiar tasks should fall. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. supported performance can look strong while examination independence remains weak. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving prompt burden. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For prompt burden, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Method selection

Purpose. The role of this system is to make the first mathematical decision an explicit training target. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. ask for target, known information, constraints and candidate representation before calculation. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. use paired questions with similar surfaces but different structures and vice versa. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. under paper conditions, train fast but justified selection rather than keyword guessing. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. more unfamiliar questions should be started independently. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. students can know every procedure yet remain unable to choose among them. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving method selection. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For method selection, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Question reading

Purpose. The role of this system is to treat reading as mathematical work. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. identify command words, quantities, relationships, conditions and irrelevant information before calculation. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. compare near-similar wordings whose structures differ so keyword matching becomes less attractive. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. train an efficient routine under time: target, data, constraints, representation, route. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. misread-question losses should fall and the learner should paraphrase the task accurately. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. fast calculation cannot rescue a model that was wrong from the first sentence. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving question reading. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For question reading, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Working and notation

Purpose. The role of this system is to use written work to externalise thinking, preserve logic and make checking possible. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. build line-by-line organisation, notation, labels and units while solutions are still relatively short. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. reduce unnecessary narration but preserve structure in multi-stage work. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. make working concise and auditable so errors can be located and repaired quickly. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. transcription and sign errors should decline and long problems should be easier to manage. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. over-compressed working creates avoidable mistakes that look careless. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving working and notation. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For working and notation, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Calculator use

Purpose. The role of this system is to make technology reduce mechanical load without replacing interpretation. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. diagnose premature calculator use, entry errors, weak estimation and dependence on button sequences. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. combine symbolic reasoning with calculator work so the learner knows what should precede numerical evaluation. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. train efficient entry, precision decisions and reasonableness checks under paper conditions. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. outputs should be interpreted against expected magnitude or structure. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. accurate key presses cannot repair an incorrect model. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving calculator use. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For calculator use, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Speed and pacing

Purpose. The role of this system is to locate where time is actually lost. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. measure reading, selection, execution, calculator, checking and stall time on short tasks. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. use timed sections with known Mathematics to separate fluency from decision problems. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. train leave-return and selective checking once full-paper timing becomes dominant. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. completion should improve without a disproportionate fall in accuracy. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. telling a student to work faster is not an intervention. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving speed and pacing. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For speed and pacing, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Checking

Purpose. The role of this system is to turn verification into a personal high-yield system. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. teach substitution, estimation, sign, unit, condition, graph and alternative-method checks where relevant. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. run checking-only drills with limited time and prioritise completed questions. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. under papers, target likely errors rather than rereading everything. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. checking should catch more preventable errors per minute. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. generic instructions to check are too vague to change behaviour. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving checking. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For checking, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Recovery after getting stuck

Purpose. The role of this system is to reduce the cost of difficult questions. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. teach an entry routine and distinguish productive struggle from helpless persistence. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. practise changing representation, trying a simpler case, leaving cleanly and preserving useful work. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. train sensible leave-return thresholds and calm re-entry. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. one hard question should cause less downstream incompletion. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. strong students are not those who never get stuck but those who recover. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving recovery after getting stuck. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For recovery after getting stuck, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Paper cadence

Purpose. The role of this system is to make every full paper part of a feedback loop. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. use papers sparingly at first unless the learner is already syllabus-secure. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. increase cadence when mixed retrieval is strong enough for papers to reveal integrated patterns. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. use realistic papers only when there is time for post-mortem and repair. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. later papers should show lower recurrence of earlier mechanisms. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. paper counts become vanity metrics when the learner’s state does not change. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving paper cadence. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For paper cadence, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Past-year papers

Purpose. The role of this system is to use authentic resources for a defined purpose. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. extract selected questions that test specific repair mechanisms rather than consuming full papers too early. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. use mixed sections for retrieval, method selection and timing. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. use complete papers for realistic performance when appropriate. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. each selected paper or section should answer a diagnostic question. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. randomly consuming every paper can waste high-quality resources. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving past-year papers. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For past-year papers, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Mock examinations

Purpose. The role of this system is to test the whole operating system under controlled realism. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. avoid high-stakes simulation before the result can be interpreted. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. use partial simulations first when one paper behaviour needs isolated training. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. run full mocks when realistic sequencing, fatigue and decision pressure are useful. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. mock results should generate specific changes in pacing, checking, repair or recovery. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. a stressful mock that produces only a score is theatre rather than training. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving mock examinations. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For mock examinations, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Paper post-mortem

Purpose. The role of this system is to convert completed papers into future decisions. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. classify lost marks and identify the first wrong decision rather than copy corrections. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. use patterns across papers to separate recurring mechanisms from one-offs. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. compress late post-mortems into a small action list so analysis does not consume all practice time. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the active error map should narrow and later papers should show fewer repeats. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. re-solving every wrong question equally can bury the few causes that matter most. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving paper post-mortem. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For paper post-mortem, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Poor prelim recovery

Purpose. The role of this system is to use a damaging result as evidence rather than a verdict. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. separate widespread gaps from paper-specific difficulty and choose two or three highest-return repairs. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. rebuild mixed and paper cycles after targeted repair rather than immediately attempting another full paper. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. protect confidence by showing which losses are repairable and which areas remain secure. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the next comparable paper should improve on the mechanisms chosen for repair. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. panic-driven schedule changes can waste the unusually rich evidence in a prelim script. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving poor prelim recovery. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For poor prelim recovery, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Mathematics plus A-Math

Purpose. The role of this system is to coordinate two subjects without doubling every activity. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. build separate error maps plus a shared algebra layer so common dependencies are repaired once. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. alternate subject-specific retrieval and papers while protecting total weekly workload. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. prevent one subject from consuming all time simply because its hardest questions feel more urgent. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. shared errors should decline in both subjects and neither subject should decay. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. treating both subjects as completely separate worlds creates unnecessary duplication. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving mathematics plus a-math. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For mathematics plus a-math, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

G1 route

Purpose. The role of this system is to keep preparation faithful to the actual subject level while strengthening practical application and reliable fundamentals. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. repair high-leverage number, proportion, measurement, algebra and interpretation weaknesses with level-appropriate questions. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. mix topics and real-world contexts so interpretation matters as well as calculation. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. use the correct G1 resources and focus on stable accessible marks, pacing and checking. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. independence should improve on G1 demand itself. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. using harder questions merely to create intensity can reduce readiness. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving g1 route. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For g1 route, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

G2 route

Purpose. The role of this system is to balance conceptual control, algebraic fluency, application and independent method selection. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. repair algebra, ratio, graph, geometry or data dependencies with broad downstream cost. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. widen mixed practice and use exam sections to train reading, selection and working. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. use correct G2 papers and coordinate G2 A-Math separately where relevant. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. retrieval, selection and completion should stabilise across varied work. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. generic Secondary resources can miss the actual route. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving g2 route. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For g2 route, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

G3 route

Purpose. The role of this system is to manage breadth, algebraic load, problem solving and paper efficiency at the learner’s actual G3 demand. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. target high-transfer algebraic or representational weaknesses rather than reopening secure chapters. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. remove topic cues aggressively and use unfamiliar surfaces to train structural recognition. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. focus final paper cycles on stability, should-own marks, timing, checking and recovery. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. strong performance should repeat across different papers. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. high topical marks can coexist with poor full-paper stability. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving g3 route. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For g3 route, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

G2 Additional Mathematics

Purpose. The role of this system is to separate A-Math-specific weaknesses from shared Mathematics foundations. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. repair shared algebra once and verify it in both subjects; identify genuine A-Math concept gaps. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. interleave within A-Math while coordinating total homework and retrieval. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. use correct G2 A-Math resources and keep diagnosis subject-specific. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. shared algebra should strengthen across both subjects while A-Math-specific error families narrow. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. doubling worksheet volume because two Mathematics subjects exist can overload the learner. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving g2 additional mathematics. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For g2 additional mathematics, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

G3 Additional Mathematics

Purpose. The role of this system is to protect long-chain symbolic reliability while integrating advanced topics. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. stabilise high-leverage algebra and major conceptual bottlenecks before relying heavily on papers. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. mix functions, trigonometry, calculus and other relevant domains so route selection becomes independent. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. train working clarity, pace, checking and recovery while coordinating with G3 Mathematics. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. long-solution accuracy and independent selection should remain stable under time. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. treating every difficult A-Math question as a whole-subject weakness prevents precise repair. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving g3 additional mathematics. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For g3 additional mathematics, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Sleep, workload and recovery

Purpose. The role of this system is to keep the learner cognitively available across the entire countdown. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. set a sustainable baseline alongside school, CCA, other subjects, meals, travel and sleep. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. increase task quality rather than volume when the timetable becomes crowded. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. taper late-night work as the examination approaches. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. attention quality, completion and error rate should not deteriorate because the plan itself is exhausting. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. a plan that improves Mathematics by damaging every other subject and sleep is inefficient. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving sleep, workload and recovery. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For sleep, workload and recovery, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Parent communication

Purpose. The role of this system is to keep the family focused on mechanisms and progress rather than panic metrics. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. agree on first repair priorities and what evidence will count as improvement. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. report retrieval, mixed-work independence and paper behaviours rather than only hours studied. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. track stability, sleep and recurring error families without turning every paper into an emotional referendum. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. parents should know what is active, fragile, repaired and maintained. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. paper counts and peer comparison create noise without improving decisions. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving parent communication. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For parent communication, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

Tutor decision loop

Purpose. The role of this system is to make the plan adaptive rather than calendar-driven. It matters because examination preparation becomes inefficient when visible symptoms are trained while the mechanism underneath remains unchanged. The countdown should therefore state what this system is trying to improve and what later evidence will count as success.

Days 90–61. each week answer what the highest-value weakness is, what evidence supports it and what intervention is being used. Work at the smallest useful scale, then reconnect the improvement to realistic Mathematics. Early progress is credible when the student can reproduce the change without having the explanation immediately available.

Days 60–31. change the balance of repair, retrieval and paper work when verification shows the learner’s state has changed. This is where chapter-specific success must become usable Mathematics. The student should face less cueing, more delayed retrieval and more decisions about which route belongs.

Days 30–1. become more selective and spend tutor attention on diagnosis after independent performance rather than constant live help. Late training should become selective. Secure material does not need equal time; fragile high-value mechanisms and paper-control behaviours deserve the remaining attention.

Evidence. the plan should respond to evidence faster than to anxiety. Verification should use independent work, a delay where appropriate and enough variation that the learner must recognise the underlying structure rather than repeat a memorised surface.

Failure mode. a rigid schedule can continue an activity long after it has stopped being high-value. If this appears, do not automatically increase volume. Revisit the diagnosis, change the training condition and test whether later independent performance improves.

Tutor decision. The tutor should state the smallest next intervention capable of improving tutor decision loop. That might mean a short explanation, a narrower repair set, a delayed retrieval probe, a mixed selection task or a timed section. The decision should be specific enough that the following session can test whether it worked instead of simply continuing the same activity by habit.

Parent and student evidence. A useful update describes what became more independent, not merely what was completed. For tutor decision loop, the student should be able to point to a concrete change in working, retrieval, error recurrence, method choice, paper completion or checking. This keeps the 90-day programme anchored to observable capability rather than study hours alone.

The explicit 13-block countdown

Days 90–84: Evidence baseline

Collect several scripts and one mixed diagnostic. Choose two or three active mechanisms. Establish prompt burden, retrieval and paper-completion baselines. Do not turn the first week into a full-paper contest.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 83–77: High-leverage repair

Repair dependencies affecting the largest number of topics. Use reduced complexity where necessary, then reconnect to current Mathematics. Keep one short exam-style section to test travel.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 76–70: Delayed verification

Close the notes, wait, then retrieve. Change the surface. A repair that works only while the explanation is fresh remains fragile.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 69–63: Reconnect to syllabus

Embed repaired number, algebra, graph or representation skills inside current school and exam-style questions. Make the learner explain where the dependency appears.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 62–56: Mixed retrieval

Begin controlled interleaving. Remove some chapter labels. Accept temporary hesitation while the learner practises method discrimination.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 55–49: Method selection

Train the first move explicitly. State target, data, constraints, representation and candidate route before calculation. Compare similar surfaces with different structures.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 48–42: Timed sections

Use short sections to isolate pacing, checking, recovery or accurate execution. Change one performance variable at a time so the result is interpretable.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 41–35: Full-paper cycles

Run papers when enough syllabus control exists. Use the cycle paper → classify → repair → verify → next paper. Track recurring mechanisms, not paper count.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 34–28: Should-own marks

Build reliability on common and accessible structures. Separate routine losses from genuinely difficult losses. Raise the floor before chasing exceptional questions.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 27–21: Pacing and recovery

Measure where time is lost. Rehearse leave-return and clean re-entry. One hard question should not create several later blanks.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 20–14: Selective checking

Use a personal checking hierarchy linked to error history. Train checks under limited time and measure yield.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 13–7: Stabilise routine

Reduce novelty, maintain fragile mechanisms, use familiar resources and protect sleep. Simulations should reinforce a known system rather than introduce chaos.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Days 6–1: Taper

Use short retrieval, representative questions and personal error review. Stop while attention is good. Protect logistics and cognitive freshness.

End the block by recording one independent capability that improved, one mechanism that remains fragile and one activity that can now be reduced. This keeps the countdown adaptive and prevents secure areas from consuming equal attention forever.

Thirty-four decision laboratories

Priority matrix

When several weaknesses compete, rank each by frequency, marks at risk, transfer value, prerequisite importance, repair time and current confidence. A low-frequency weakness with little transfer value may be postponed even if it feels dramatic; a modest algebra weakness affecting many topics may deserve immediate attention.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Concept versus retrieval

Give the learner generous time and a clear context. If the idea still cannot be explained or executed, treat it as a knowledge problem. If the learner can do it once reminded but cannot retrieve it later, change the spacing and recall system rather than repeating full explanations.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Formula memory

First confirm what must actually be recalled for the relevant syllabus. Then separate remembering the formula from recognising when it applies and mapping variables correctly. A learner can memorise every formula and still choose the wrong one because selection was never trained.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Error-log design

Keep the log short: date, task, first wrong decision, mechanism, repair and verification. Do not paste every wrong question into an archive. The log should help decide what to practise next, not become another revision task that consumes time.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Paper selection

Choose a paper for a reason: breadth, unfamiliarity, timing, difficulty calibration, checking or recovery. Do not select only the hardest papers available. A paper is useful when its demand matches the diagnostic question being asked.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Paper sequencing

Alternate paper work with targeted repair and retrieval. Back-to-back papers can be useful for a narrow high-performing profile, but for many students the best sequence is paper, post-mortem, repair, verification, then the next paper.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Marking a paper

Separate arithmetic and algebra slips from method-selection failures, missing knowledge and time-driven omissions. Mark should-own losses distinctly. This creates a better decision than a total score alone.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Post-paper repair

Choose representative errors rather than redoing the entire paper. Repair the mechanism, then verify it on changed questions. The aim is to alter the next paper, not produce a perfect correction copy of the last one.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Mock realism

A mock should reproduce the relevant timing, materials, interruptions and decision demands closely enough to test the operating system. Do not add artificial stress beyond what is needed. The objective is information and rehearsal, not punishment.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Difficult-question protocol

Before help, require the learner to state the target, known information and one possible representation. If the route stalls, preserve useful work, try another representation, then decide whether to leave. This turns being stuck into a sequence of actions.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Checking hierarchy

Build personal priorities from past errors. A student prone to sign errors may scan high-risk algebra first; another may check units and conditions; another may use substitution. The hierarchy should reflect actual probability, not a generic checklist copied from someone else.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Calculator discipline

Use estimation and structure to predict a plausible answer before or after calculator work. Train accurate entry and sensible precision. If calculator dependence is hiding weak number sense, include short non-calculator reasoning where appropriate.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Speed diagnosis

Measure where minutes are lost rather than treating time as one number. Compare reading, choosing, writing, calculator, checking and stall time. Train only the bottleneck that actually controls completion.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Sleep protection

Set a latest sensible stopping time, especially in the last two weeks. Sleep supports attention, working memory and error monitoring. Extra late-night practice can have negative value when it reduces the quality of the next day.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Weekend structure

Use longer blocks for full papers, deep repair or post-mortems, but pair them with recovery and a stopping point. Weekends should not become unbounded Mathematics marathons that make Monday school performance worse.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Short weekday structure

Use twenty- to forty-minute blocks for retrieval, one repair target, checking practice or a short mixed set. A short high-quality session can preserve memory between larger weekend cycles.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Missed-day recovery

Do not double the next day’s load. Protect the phase priorities, drop lower-value tasks and continue. A resilient plan should survive school events, illness and occasional disruption without turning the following week into punishment.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Prelim integration

Treat prelims as a major diagnostic event inside the runway. Increase familiarity before the prelim where useful, then use the returned script to rebuild the error map. The prelim is evidence for the next cycle, not the final judgment of ability.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

After a poor prelim

Separate widespread structural weakness from paper-specific difficulty. Choose a few highest-return repairs, verify them, then return to paper cycles. Use the dedicated poor-prelim owner for deeper recovery.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

After a strong prelim

Do not stop thinking. Verify repeatability, move secure topics to maintenance and use the remaining runway to reduce variance, strengthen checking and prepare for unfamiliar paper features.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Confidence repair

Replace global labels with specific mechanisms. Show the learner that a sign error, retrieval gap or timing problem has a repair path. Confidence becomes more durable when it is attached to evidence of control.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Overconfidence control

Use delayed retrieval, mixed questions and unfamiliar surfaces. The goal is accurate self-assessment. Strong recent topical results should not be allowed to hide weak transfer or poor paper stability.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Parent role

Keep home support logistical and evidence-based: schedule, sleep, materials, calm review of progress. Avoid making every paper a high-stakes conversation or comparing paper counts with peers.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Tutor role

Shift from explanation toward diagnosis as the examination approaches. Let the student own serious independent attempts. The tutor’s highest-value contribution late in the cycle is often identifying why performance failed, not preventing failure during the attempt.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Three-student group

Use group time for route comparison, error analysis and discussion after independent work. As the examination approaches, include more silent timed sections so each learner’s true paper state remains visible.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

School and tuition coordination

If school already supplies abundant routine practice, tuition should not duplicate it blindly. Use tuition for diagnosis, mixed selection, repair, post-mortem and verification. If school revision is broad but shallow, tuition can target the specific dependencies the learner still needs.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

G1 paper route

Use level-appropriate resources and practical interpretation. Protect clear working, reliable fundamentals and accessible marks. Do not replace good G1 preparation with irrelevant G2/G3 difficulty merely to create a harder session.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

G2 paper route

Use the correct G2 resources, with enough algebra, application and mixed selection practice. If the learner takes G2 A-Math, coordinate retrieval and paper cadence across the two subjects rather than doubling every task.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

G3 paper route

Use G3 resources to train breadth, method selection and stable execution. Strong topical learners may need unfamiliar paper surfaces and realistic timing more than another cycle of routine chapter practice.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

G2 A-Math route

Separate shared algebraic foundations from A-Math-specific content. Repair shared weaknesses once, verify in both subjects, then allocate A-Math-specific practice to the topics that genuinely need it.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

G3 A-Math route

Protect symbolic accuracy across longer chains. Mix functions, trigonometry, calculus and other relevant areas. Use papers to diagnose where long solutions break under time, then repair the specific mechanism.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Final-week resource choice

Use the smallest set of materials that keeps the system available: personal error notes, selected familiar questions, necessary formula or convention reminders and perhaps one controlled paper early in the week. Avoid a last-minute resource explosion.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Examination morning

Keep recall light and familiar if used at all. Protect travel, calculator, stationery, meals and arrival. The morning is for access to a rehearsed system, not for learning a new mathematical technique.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

After the examination

Record what the preparation cycle taught about the learner: which repairs held, which paper strategies worked, which errors persisted and how independence changed. Good examination preparation should leave a stronger learner for the next Mathematics stage.

Apply the laboratory inside the current phase rather than in isolation. Early in the runway, the decision can justify more teaching; in the middle, it should often change retrieval or mixed practice; late, it should usually refine paper execution or protect the learner from low-value novelty.

Then verify. State what evidence would show that the decision worked—fewer repeated errors, faster retrieval, more independent starts, better completion, higher checking yield or a more stable later paper. Without a verification condition, the plan can continue an intervention long after it has stopped being useful.

Twenty learner cases

Cannot start unfamiliar questions

Train representation and a simple entry routine early, remove topic cues in the middle phase, and track the percentage of unfamiliar questions receiving a justified first move before any hint.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Starts but gets stuck halfway

Diagnose algebraic execution, working-memory overload, missing intermediate representation or inability to change route. Train recovery on shorter problems before full-paper conditions.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Forgets every chapter

Treat this as a retrieval-system problem until evidence suggests a deeper misconception. Build return dates and mixed retrieval instead of restarting each topic from zero.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Accurate but painfully slow

Break time into reading, selection, execution, calculator, checking and stall components. Increase pace only where Mathematics is stable.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Fast but below potential

Inspect rushed reading, compressed working, skipped checks and overconfidence. The goal is not to become generally slower; it is to spend time where it prevents loss.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Panics at a difficult first page

Use simulations with deliberately placed difficult items. Train skip-and-return and emotional reset so paper order stops controlling confidence.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Studies many hours but barely improves

Audit activity. Replace duplicate routine work with mechanism-specific repair, delayed retrieval and mixed selection. Long hours can hide low training quality.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

One catastrophic weak topic

Estimate transfer value, mark exposure and repair time. Use early intensive repair if the topic is high-leverage; otherwise contain it while protecting the rest of the syllabus.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Prelim much worse than school tests

Compare breadth, timing, unfamiliarity and anxiety. Use the prelim as a diagnostic event, not a global verdict.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Prelim surprisingly strong

Verify repeatability, move secure topics to maintenance and use the remaining runway to reduce variance rather than chase ever harder content.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Relies on model answers

Delay model access until after a real attempt. Use the model to compare routes after diagnostic evidence exists.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Overuses the calculator

Require structure or estimation where appropriate, then use the calculator efficiently. Check whether the device is covering weak number sense or simply slowing work.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Repeatedly misreads conditions

Make the reading routine visible, then compress it for time. Paraphrase target and constraints before calculation.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Cannot decide what to check

Build a personal risk list from past papers and link each recurring error to one or two cheap checks.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Does well only on one tutor’s worksheets

Introduce external paper styles and unfamiliar wording gradually. The issue may be cue dependence on a familiar resource ecosystem.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Avoids A-Math because it feels too hard

Separate shared algebra from A-Math-specific concepts and begin with a repair that restores entry. Build shorter successful cycles before raising difficulty.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

A-Math consumes all revision time

Compare marginal returns across subjects. Rebalance if Mathematics is leaking accessible marks. Shared algebra can serve both; paper diagnosis remains subject-specific.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Wants to cram the night before

Use a short retrieval and personal error review, then stop. A rested brain has more value than one extra exhausted paper that cannot be repaired before the exam.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Wants no revision the day before

That can be reasonable if the system is stable. A very light familiar retrieval may help, but rest and logistics are often the higher-value tasks.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

Ready to work independently

Shift tutor time toward diagnostics, post-mortems and occasional high-value explanation. Let the learner own retrieval, papers and correction cycles.

The verification rule is the same: later independent performance must change. If the learner improves only while the tutor is controlling the task, the intervention has improved support conditions but has not yet demonstrated examination transfer.

A weekly parent-and-tutor dashboard

• Current phase: Repair, Integrate or Perform.
• Top two active mechanisms: current highest-value weaknesses.
• One fragile mechanism: improved but not yet trusted.
• One maintenance area: secure enough for light retrieval.
• Prompt burden: how much help familiar questions still require.
• Retrieval: which repairs survived a delay.
• Mixed work: whether method selection is increasingly independent.
• Completion: whether timing leaves significant unfinished work.
• Checking yield: whether verification catches preventable errors efficiently.
• Recovery: whether one difficult question still disrupts later performance.
• Workload: whether the plan remains sustainable with school and sleep.
• Next decision: what evidence will change next week’s plan.

The dashboard is deliberately small. It exists to make the revision system observable, not to create administration. Leading indicators such as fewer repeated errors, stronger delayed retrieval, more independent starts and better paper completion often move before marks fully stabilise.

Frequently Asked Questions

Is 90 days enough?

It is enough for meaningful change when priorities are selective. It cannot guarantee complete mastery of every gap for every learner.

Should Day 90 begin with a full paper?

Only if a full paper will produce interpretable evidence. A weaker learner may need a short mixed diagnostic first.

How many full papers?

No universal number. Paper cadence should be limited by the time needed for post-mortem, repair and verification.

When should interleaving begin?

As soon as the learner has enough method stability for selection to be productive. Start narrow and widen.

When should papers dominate?

When most relevant content is known and papers reveal performance variables rather than mainly missing chapters.

Should new topics be learned in the final month?

Yes when they are high-value. The final month is selective, not frozen.

One paper every day?

Not by default. Daily papers can help a narrow strong profile but can also repeat unresolved states and create fatigue.

Scores fall when topics are mixed—is that bad?

Not necessarily. Method selection is harder than cued execution. Diagnose before retreating to topical work.

How should G1/G2/G3 change the plan?

Use the learner’s actual level and correct resources. The training architecture can remain common while mathematical demand differs.

What about Mathematics plus A-Math?

Use separate subject maps plus shared algebraic dependencies. Coordinate total workload.

What should parents watch besides marks?

Error recurrence, prompt burden, delayed retrieval, independent starts, completion, checking, recovery and workload.

What if the learner is already strong?

Train stability, unfamiliar surfaces, efficiency and reduction of avoidable variance rather than only adding harder material.

What if the learner is still weak at Day 30?

Prioritise accessible marks and high-frequency mechanisms. Use targeted sections if full papers remain too noisy.

What if one bad mock causes panic?

Analyse whether it reflects recurring weakness, unusual difficulty or one-off execution. Let the evidence change the plan proportionally.

When should study stop the day before?

When quality falls and further work threatens sleep or confidence. The exact time varies, but tapering should protect cognitive freshness.

Where this owner sits in BTT

• Mathematics Examination Craft — broad examination-performance layer.
• Recover After a Poor Mathematics Preliminary Examination — prelim-specific recovery.
• How to Use Past-Year Mathematics Papers Properly — paper selection and use.
• How to Run Mathematics Mock Examinations Properly — simulation design.
• How to Do a Mathematics Examination Post-Mortem — extracting action from a completed paper.
• How to Recover After Getting Stuck in a Mathematics Examination — question-level recovery.
• How Active Recall Works for Mathematics — retrieval mechanics.
• How Spaced Practice Works for Mathematics — retention across time.
• How Interleaving Works for Mathematics — mixed method selection.
• Secondary Mathematics Tuition — broad Secondary route.

How to use this 90-day owner

This page owns the countdown architecture: how the dominant training job should change as the examination approaches. It does not need to become the deepest page on every examination skill because BTT already has specialist owners for past-year papers, mock examinations, paper post-mortems, getting unstuck and poor-prelim recovery.

The working sequence is repair → retrieve → mix → perform → protect. Early days are more valuable for structural repair because there is time to teach, wait and retest. Middle days are more valuable for retrieval, method selection and mixed work. Late days are more valuable for reliability, pacing, checking, recovery and preserving sleep and cognitive control.

The countdown should change the job, not simply increase the panic.

2026–2027 examination routing

Students graduating in 2026 remain on the relevant existing GCE routes. From the 2027 graduating cohort, the Singapore-Cambridge Secondary Education Certificate replaces the separate GCE N(T), N(A) and O-Level certificates under Full Subject-Based Banding. Students sit subjects at the relevant G1, G2 or G3 level.

For 2027 school candidates, Mathematics is listed as K110 at G1, K210 at G2 and K310 at G3. Additional Mathematics is listed as K232 at G2 and K341 at G3. Confirm the learner’s actual cohort, subject level and whether Additional Mathematics belongs in the schedule before choosing resources or allocating paper volume.

Use the official SEAB SEC overview, G1 syllabus list, G2 syllabus list, G3 syllabus list and MOE’s Full SBB / SEC announcement. BTT’s SEC Mathematics Progression page remains the curriculum owner.

Thirteen countdown gates

Days 90–84: Baseline and ranking

Collect several sources of evidence, classify mechanisms and rank two or three targets by recurrence, transfer value, mark cost and repair time.

Guardrail. Avoid using one remembered score as the entire diagnosis. A short mixed diagnostic plus recent school evidence is usually more useful.

Gate. Exit the block with a written priority list and a clear retest condition.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 83–77: Repair the widest leak

Repair one high-leverage dependency at reduced complexity, then reconnect it to current questions.

Guardrail. Do not remain in foundation drills after the learner can perform them directly. The repair must travel.

Gate. Exit when the mechanism survives at least one changed-surface question.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 76–70: Delay and retrieve

Wait before retesting. Remove notes, examples and prompts, then ask for retrieval.

Guardrail. Use active recall when the idea was understood but forgotten; reteach only if the conceptual model still collapses.

Gate. Exit when the learner can retrieve after a gap without needing the tutor to reopen the route.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 69–63: Reconnect and transfer

Place repaired mathematics inside current syllabus contexts, including graphs, geometry, trigonometry or A-Math where relevant.

Guardrail. Ask the student to identify the hidden dependency before calculation.

Gate. Exit when current questions stop failing for the earlier reason.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 62–56: Begin deliberate mixing

Remove topical labels and mix near-neighbour methods before widening the pool.

Guardrail. Some hesitation is productive because the learner is now selecting rather than following.

Gate. Exit when selection becomes increasingly plausible without tutor cues.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 55–49: Train the first move

Score question entry separately: target, givens, constraints, representation and candidate method.

Guardrail. Use similar-looking questions that require different methods and different-looking questions that share a structure.

Gate. Exit when more unfamiliar questions are started independently.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 48–42: Use timed sections

Isolate pacing, reading, checking, recovery or execution with shorter examination blocks.

Guardrail. Choose mostly known mathematics so performance variables can be seen clearly.

Gate. Exit when the trained behaviour improves on a comparable section.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 41–35: Start full-paper cycles

Use full papers as integrated diagnostic instruments and classify losses before assigning the next task.

Guardrail. Follow paper → diagnose → repair → verify → next paper.

Gate. Exit when recurring mechanisms begin to shrink rather than merely repeat.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 34–28: Raise the score floor

Protect should-own marks and common structures before chasing rare difficult questions.

Guardrail. Separate avoidable loss from genuinely high-demand loss.

Gate. Exit when ordinary marks become more stable across papers.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 27–21: Build pacing and recovery

Measure where time disappears and train leave-return decisions around the actual bottleneck.

Guardrail. Preserve partial work and practise re-entry rather than restarting from zero.

Gate. Exit when completion improves without a major accuracy collapse.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 20–14: Make checking selective

Build a personal hierarchy of high-yield checks based on error history.

Guardrail. Limit checking time and measure marks recovered per minute.

Gate. Exit when checking becomes shorter, more purposeful and higher-yield.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 13–7: Reduce novelty

Keep fragile high-value mechanisms active and move secure areas to maintenance.

Guardrail. Use familiar resources, selected papers and short retrieval; protect sleep.

Gate. Exit with a stable routine that no longer needs major redesign.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Days 6–1: Taper and protect

Narrow revision to personal risk areas, light retrieval and familiar representative questions.

Guardrail. Stop before fatigue converts practice into noise. Avoid a punishing final mock.

Gate. Exit by arriving rested enough to use the Mathematics already built.

At the end of this block, write one sentence answering: What can the student now do independently that was less reliable seven days ago? If there is no meaningful answer, change the intervention or the evidence rather than simply increasing volume.

Keep the larger phase in view. Early blocks should reduce structural weakness, middle blocks should expand retrieval and method selection, and late blocks should increase reliable examination execution without damaging sleep or confidence.

Forty-five decision clinics for the 90-day plan

Algebra infrastructure

Algebra is a cross-topic operating language. Weak sign control, manipulation, substitution, factorisation or equation solving can damage graphs, geometry, trigonometry and Additional Mathematics.

Days 90–61. Isolate the exact mechanism, repair it at reduced complexity and reconnect immediately to a real current problem. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Hide algebra inside mixed questions so recognition replaces worksheet labels. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Protect accuracy under time and use working that remains auditable. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Recurrence across several topics, delayed retrieval and changed-surface success. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not re-teach every algebra chapter when the remaining weakness is one repeatable mechanism. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Negative numbers

Negative-number mistakes can be conceptual, transcriptional or caused by compressed working. The final wrong sign does not tell you which.

Days 90–61. Use number lines, operation meaning and brackets to identify whether the student understands the sign structure. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Place the same sign logic inside algebra, coordinates and formula substitution. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use a quick personal sign-check routine under time. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Fewer sign errors across different topics rather than one perfect drill. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not call repeated sign mistakes careless without identifying their source. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Fractions

Fraction weakness can surface inside ratio, algebraic fractions, formulas, probability and trigonometry. It deserves high priority when it acts as a shared dependency.

Days 90–61. Repair meaning and operations directly, especially equivalence and common-denominator reasoning. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Mix fraction use across contexts so the learner recognises when exact fractional work is useful. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Preserve exact forms where appropriate and avoid unnecessary decimal conversion. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Independent fraction control inside non-fraction-labelled problems. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not assume calculator use removes the need for fraction sense. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Ratio, rate and percentage

These topics share multiplicative structure but are often memorised as separate procedures. Confusion later can damage many applied questions.

Days 90–61. Use simple representations to rebuild the relationship among quantities. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Mix contexts so the student must decide whether the relationship is ratio, rate, proportion or percentage. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Train unit consistency and quick plausibility checks. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Correct method choice without keyword dependence. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not reward formula matching when the relationship is not understood. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Graphs and functions

Graphs connect numerical, algebraic and visual representations. Plotting skill alone is not enough.

Days 90–61. Connect table, equation, coordinates and graph, asking what each point and feature means. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Mix graph questions with algebra and contextual interpretation. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Train fast extraction of relevant features and use graph sense as a check. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Ability to move among representations and explain the relationship. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not let memorised graph shapes replace relational understanding. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Geometry and diagrams

Geometry failures often begin when the learner assumes a property from appearance or misses a stated condition.

Days 90–61. Mark givens, derived facts and unknowns explicitly. Reconstruct diagrams from conditions when useful. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Integrate geometry with algebra, ratio and trigonometry while requiring justification. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use efficient annotations and check whether the final answer fits the diagram and conditions. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Fewer unsupported assumptions and stronger justification. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not treat every geometry error as a missing formula. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Trigonometry

Trigonometry can fail at diagram reading, rule selection, algebraic setup or calculator execution.

Days 90–61. Diagnose the failed stage before assigning more formula practice. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Mix different trigonometric routes with geometry so method choice becomes necessary. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Standardise calculator mode, angle handling, rounding and reasonableness checks. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Correct first-step selection plus accurate execution. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not compress all trigonometry errors into ‘forgot the formula’. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Statistics

Statistics requires interpretation as well as calculation. A correct mean can answer the wrong question if the data were misunderstood.

Days 90–61. Repair concepts such as average, spread, sample, frequency and representation at the required level. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Mix tables, graphs and contextual questions so the learner must decide what information matters. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Train careful reading and efficient calculation under time. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Ability to explain what the computed value means in context. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not let calculator fluency hide interpretation errors. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Probability

Probability becomes difficult when intuition is not supported by a clear event model or sample space.

Days 90–61. Use small sample spaces, complements and explicit event descriptions. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Mix representations such as tables, trees and diagrams so the student chooses appropriately. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Focus on disciplined notation, condition reading and plausibility. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Clear event definition before calculation and probabilities that make sense. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not substitute formula repetition for representation. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Calculator control

Calculator errors can come from brackets, mode, stored values, rounding or transcription rather than Mathematics itself.

Days 90–61. Observe actual keying and build a consistent entry routine. Require estimates before acceptance of output. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Combine symbolic and numerical work so the student knows what should remain exact. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Standardise mode checks, rounding and re-entry under time. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. A falling rate of calculator-caused errors across papers. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not tell the student merely to be careful when a reproducible routine can fix the issue. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Formula knowledge

Recall and selection are separate. A student may remember many formulas but still not know which relationship applies.

Days 90–61. Confirm what genuinely needs recall for the actual syllabus, then connect each formula to meaning and conditions. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Remove formula labels and require justification before substitution. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Keep high-frequency relationships accessible while prioritising selection and accurate use. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Fewer formula-selection errors, not just quicker recitation. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not spend the final month memorising what is already known if selection is the real weakness. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Question reading

Reading a Mathematics question is part of the mathematics. The target, relevant data, conditions and relationships must be extracted before calculation.

Days 90–61. Require paraphrasing and representation before computation on selected questions. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use similar wording with different mathematical structures so keywords cannot carry the decision. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Train a quick routine: target, data, constraints, representation, route. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Fewer errors before the first calculation begins. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not label misreading as a calculation mistake. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Representation

A learner may know the necessary mathematics but fail to convert the problem into an equation, diagram, table or graph.

Days 90–61. Teach multiple representations and ask which one exposes the relationship most clearly. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use changed contexts so the student must select a representation without copying an example. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Train rapid externalisation of information in long problems. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Better problem entry and fewer stalls caused by keeping everything mentally. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not force one favourite representation on every question. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Method selection

Method selection is the bridge between knowing procedures and solving mixed examination questions.

Days 90–61. Contrast nearby methods and ask why one applies rather than another. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Remove topic labels and score the first minute of unfamiliar questions. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Train efficient choices and willingness to change route when evidence rejects the first method. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Higher rate of plausible independent starts. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not count only continuation after the tutor names the method. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Working and notation

Working reduces memory load, communicates reasoning and makes checking possible. Compression can be efficient until it becomes error-prone.

Days 90–61. Establish clear transformations and consistent notation. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Teach what can be shortened safely and what should remain explicit. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use efficient but auditable working that supports quick error location. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Fewer transcription and sign errors linked to working structure. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not confuse decorative neatness with functional clarity. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Active recall

Rereading notes creates familiarity but may not produce examination access.

Days 90–61. Use recall after explanations and corrections are no longer visible. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Broaden retrieval across older topics without chapter cues. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use short recall to maintain secure areas while paper work increases. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Successful retrieval after increasing delays. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not treat same-day fluency as durable memory. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Spacing

Spacing schedules retrieval after some forgetting, making memory more durable and the test more honest.

Days 90–61. Return to repairs after short gaps and gradually lengthen them. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Keep a rotating set of older mathematics alive alongside new mixed work. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use brief spaced maintenance for secure topics in the final month. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Knowledge remains accessible without full re-teaching. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not revisit every topic at identical intervals. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Interleaving

Interleaving teaches the student to distinguish methods rather than execute a named procedure.

Days 90–61. Start with a small contrast once methods are stable enough. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Widen the mix as selection improves and include older topics. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use realistic paper order as the final natural form of interleaving. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Selection accuracy improves even when question surfaces change. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not mix so widely that the student cannot form any stable schema. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Error log

An error log should guide decisions, not collect every wrong question forever.

Days 90–61. Record mechanism, context, repair and retest date for high-value recurring errors. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use the log to decide which repairs still need mixed verification. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Reduce the log to a personal risk register before the examination. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Mechanisms move from active repair to fragile, repaired and maintenance. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not keep historical errors in intensive revision after they are secure. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Prompt burden

A smooth lesson can hide tutor dependence when the tutor supplies method, first step or error signal.

Days 90–61. Record the kind of help required for familiar questions and remove one layer at a time. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Protect silent first attempts and use graduated hints only after evidence of the student’s own route. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Complete serious papers without intervention and review afterward. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Lower minimum prompt level and a smaller supported-versus-independent gap. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not mistake supported performance for examination readiness. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Paper pacing

Pacing is a system of allocation, not a command to move faster.

Days 90–61. Measure reading, selection, calculation, checking and stall time separately. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use timed sections to train the actual bottleneck. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Rehearse whole-paper checkpoints and leave-return decisions. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. More complete papers without a disproportionate accuracy drop. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not use whole-paper repetition when one specific time leak dominates. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Checking

Checking should target likely errors with methods that can actually disconfirm an answer.

Days 90–61. Teach checks tied to the student’s history: substitution, estimate, units, sign, graph or domain. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Run checking-only drills with a fixed time budget. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use a personal hierarchy when paper time is limited. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Higher checking yield and fewer preventable losses. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not spend equal checking time on every question. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Recovery after getting stuck

Strong students still get stuck. The skill is preventing one stall from damaging the rest of the paper.

Days 90–61. Teach how to identify what remains valid after a failed route. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Practise changing representation or trying a simpler case before seeking help. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Rehearse leaving cleanly, moving on and re-entering later. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Shorter unproductive stalls and better downstream completion. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not equate strategic leaving with giving up. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Past-year papers

Past papers are resources whose value depends on selection and analysis, not trophies to count.

Days 90–61. Extract questions that test current repair targets. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use mixed sections for retrieval and selection. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use complete papers selectively and reserve time for post-mortem and repair. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Later papers improve because earlier papers changed training. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not consume papers merely to generate scores. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Mock examinations

A mock tests the entire operating system: timing, reading, emotional control, checking and recovery.

Days 90–61. Avoid heavy simulation when missing content dominates the result. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use mocks once enough syllabus control exists for performance variables to matter. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Run selectively and early enough that repairs remain possible. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Simulation produces clear paper-management changes. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not make mock frequency the measure of seriousness. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Post-mortems

A post-mortem converts a paper into the next training decision.

Days 90–61. Focus early paper reviews on structural weakness and missing knowledge. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Separate retrieval, selection and transfer in the middle phase. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Emphasise pacing, checking, recovery and should-own losses late. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Recurring categories shrink across papers. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not re-solve every wrong question equally. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Should-own marks

The final score often improves more by protecting common marks than by chasing rare difficult questions.

Days 90–61. Identify common structures that are still unreliable. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Expose them inside mixed work so reliability survives without cues. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Maintain them cheaply and count every preventable loss. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. The learner’s score floor rises across different papers. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not neglect ordinary marks because the hardest questions feel more prestigious. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Confidence

Confidence is most useful when grounded in independent evidence.

Days 90–61. Use successful repair and delayed retrieval to show that weaknesses can change. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Build evidence through unseen starts and mixed work without prompts. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use stable timed performance and successful recovery as the strongest proof. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Greater willingness to attempt and faster recovery after errors. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not manufacture confidence only with easy work. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Sleep and workload

Revision quality depends on the whole student. Sleep and competing subjects affect mathematical retrieval and control.

Days 90–61. Allow more learning volume early when there is room to recover. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Protect consistency and avoid duplicating school workload unnecessarily. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Reduce low-yield late-night work as the examination approaches. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Extra hours produce measurable benefit rather than fatigue. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not optimise Mathematics by damaging every other subject. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Parent communication

Parents help most when they understand what the plan is trying to change.

Days 90–61. Explain high-leverage repair targets and why they were chosen. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Report retrieval, mixed work and prompt reduction rather than chapter count. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Report paper completion, checking, recovery and stability. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Weekly decisions become clearer and less driven by panic. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not use paper count or hours studied as the sole evidence of progress. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Tutor role

The tutor’s role should shift from repair architect to integration coach to performance diagnostician.

Days 90–61. Teach explicitly where genuine misconceptions need rebuilding. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Observe first attempts and reduce prompts as knowledge becomes available. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Let serious paper work happen independently and add value through diagnosis afterward. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Tutor intervention falls on familiar work while student control rises. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not let examination pressure cause the tutor to take back decisions the student needs to own. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Far-behind learner

A far-behind student needs prioritised re-entry, not a promise to relearn everything equally.

Days 90–61. Repair the smallest high-leverage set of prerequisites that unlocks current work. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Begin controlled mixing before complete confidence so repaired skills become usable. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Protect accessible marks and use sections when full papers remain too noisy. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. More questions become accessible and startable independently. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not spend all ninety days inside old chapters. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Strong but inconsistent learner

The problem is reliability rather than broad missing knowledge.

Days 90–61. Compare several scripts to identify variance drivers such as stalls, overconfidence or checking. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use unfamiliar surfaces and mixed sections to stress the decision system. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Reduce avoidable variance through stable paper routines. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Paper-to-paper spread and should-own losses decrease. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not respond with more routine content when knowledge is already strong. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Slow but accurate learner

Slow work requires a time diagnosis before speed training.

Days 90–61. Measure where time is actually spent and protect correct structure. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use timed micro-sections to build fluency in the true bottleneck. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Train whole-paper pacing and selective checking without sacrificing accuracy. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Completion improves while accuracy remains broadly stable. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not accelerate everything equally. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Many avoidable mistakes

A high mistake rate needs an error taxonomy, not the label careless.

Days 90–61. Name the top recurring mechanisms and design prevention routines. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Carry those routines into mixed and timed work. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use a personal checking hierarchy in final papers. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Recurrence falls by mechanism. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not treat random slips and structural error as the same problem. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Plateaued learner

A plateau often means current practice no longer targets the limiting mechanism.

Days 90–61. Change the diagnostic task using retrieval, mixed work, first-attempt analysis and timing. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Design practice around the exposed bottleneck rather than increasing volume. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use papers to verify that the new intervention changes real performance. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. The limiting mechanism moves even if total marks move later. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not intensify a strategy simply because it is familiar. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Mathematics plus A-Math

Two subjects need separate tracking with a shared algebraic foundation layer.

Days 90–61. Repair shared dependencies once and verify in both subjects. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Coordinate retrieval and workload so one subject does not disappear. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Schedule heavy papers to preserve recovery and protect the other subject. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Shared and subject-specific errors become easier to distinguish. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not mechanically double every homework activity. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

G1 Mathematics route

G1 planning should use the actual G1 syllabus and examination requirements while preserving the same repair-retrieve-perform architecture.

Days 90–61. Prioritise meaningful application, core numerical and algebraic control and the student’s highest-frequency barriers. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use mixed application and retrieval appropriate to the level without importing irrelevant higher-level demands. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use correct G1 papers and practise reliable execution of the Mathematics actually required. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Improvement is judged against the learner’s actual subject level and paper. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not treat G1 as a label for the whole learner or use G3 resources indiscriminately. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

G2 Mathematics route

G2 planning should match the G2 syllabus and the learner’s current school route, including Additional Mathematics where actually taken.

Days 90–61. Repair the prerequisites that carry the relevant G2 mathematics and identify readiness issues accurately. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Increase mixed problem solving and method selection at the appropriate demand. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use correct G2 papers, pacing and checking routines. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Evidence is gathered from the actual G2 task, not from mismatched papers. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not assume G2 is simply a diluted version of another level. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

G3 Mathematics route

G3 planning should match the G3 syllabus and, where relevant, coordinate G3 Additional Mathematics separately.

Days 90–61. Repair high-leverage algebraic and representational dependencies early. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use mixed unfamiliar questions and integrated upper-Secondary work to train selection. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Use correct G3 papers and emphasise stability, timing and recovery. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. The learner shows strong retrieval and method selection across realistic G3 demand. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not let advanced question volume replace precise diagnosis. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Subject-level movement

Movement between G1, G2 and G3 should be based on evidence, not aspiration alone.

Days 90–61. Map prerequisite differences, retrieval, independence and workload capacity. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use bridging tasks to test whether the learner can sustain the greater or different demand. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Protect examination preparation for the actual registered route while transition evidence is gathered. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Readiness is shown by stable independent performance on bridging work. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not use one recent score as the sole movement criterion. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Poor preliminary examination

A poor prelim is a high-information event, not a command to panic.

Days 90–61. Classify the script and choose one or two high-return repairs. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Retest those repairs before another full paper and rebuild mixed retrieval. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Return to full-paper work with clearer priorities and protect confidence through evidence. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Later papers show fewer recurring mechanisms rather than only more attempts. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not assign another paper immediately merely to erase the emotional effect of the previous one. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

High performer

A high performer still needs an operating plan; the bottleneck may be stability, method efficiency or unusual surfaces.

Days 90–61. Use deeper reasoning, route comparison and independent diagnosis rather than basic volume. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Increase unfamiliarity and reduce cues while maintaining older retrieval. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. Optimise checking, pacing, elegance and resilience on difficult papers. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Peak scores become more repeatable and should-own losses approach zero. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not equate high marks with the absence of trainable mechanisms. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Final ten days

The final ten days should reduce novelty and protect the system already built.

Days 90–61. Use one final substantial diagnostic only if it has a clear purpose, then narrow repair. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Shift toward familiar mixed retrieval and personal risk review. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. In the last few days, shorten sessions, protect sleep and avoid punishing simulations. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. The student arrives with accessible knowledge and a familiar routine. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not try to manufacture completeness through exhaustion. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Examination day

Examination day is for execution, not new learning.

Days 90–61. Review only what is already familiar if any review is needed at all. This is the period with enough time to change the underlying mechanism and verify that the repair survives beyond the original lesson.

Days 60–31. Use the practised entry, pacing, leave-return and checking routines. The emphasis now shifts from isolated competence toward retrieval, recognition and selection when several mathematical possibilities compete.

Days 30–1. After a difficult question, recover and protect the rest of the paper. With less time remaining, the intervention should become selective, examination-facing and proportionate to the marks or disruption the mechanism can still cause.

Evidence. Performance follows the trained system rather than improvisation under stress. Record whether the learner can reproduce the improvement independently after a delay and, where appropriate, under mixed or timed conditions.

Failure to avoid. Do not let one early mistake define the remaining paper. If this failure appears, change the training decision rather than simply increasing the quantity of the same activity.

The clinic should remain subordinate to the countdown. Once the mechanism becomes reliable, reduce its training dose and return attention to the next highest-value risk. The plan is strongest when priorities are allowed to change.

Ten paper-cycle recipes

Repair-cycle paper

Use one paper or section primarily to reveal structural weaknesses. Select no more than two high-leverage repairs afterward, then retest before the next comparable paper.

When to use it. Best early when the paper is informative but still exposes significant knowledge gaps.

End the cycle with one or two next actions. If the paper produces a long unfocused list, prioritise by recurrence, transfer value, mark cost and repair time. The paper is useful only when it makes the next decision better.

Retrieval-cycle paper

Use a paper after several days without topical rehearsal to see what remains genuinely available.

When to use it. Best in the middle phase when durable access matters more than same-day fluency.

End the cycle with one or two next actions. If the paper produces a long unfocused list, prioritise by recurrence, transfer value, mark cost and repair time. The paper is useful only when it makes the next decision better.

Method-selection paper

Review first-method choices and question entry, even on questions eventually solved correctly.

When to use it. Best when topical work is strong but mixed work is weak.

End the cycle with one or two next actions. If the paper produces a long unfocused list, prioritise by recurrence, transfer value, mark cost and repair time. The paper is useful only when it makes the next decision better.

Accuracy-cycle paper

Use mostly familiar mathematics and classify preventable losses, then apply personal prevention routines.

When to use it. Best from the middle phase onward.

End the cycle with one or two next actions. If the paper produces a long unfocused list, prioritise by recurrence, transfer value, mark cost and repair time. The paper is useful only when it makes the next decision better.

Pacing-cycle paper

Record time at checkpoints and analyse leaving decisions as carefully as marks.

When to use it. Best when completion is the dominant issue.

End the cycle with one or two next actions. If the paper produces a long unfocused list, prioritise by recurrence, transfer value, mark cost and repair time. The paper is useful only when it makes the next decision better.

Recovery-cycle paper

Use a paper likely to contain several difficult questions and review how the student behaves after getting stuck.

When to use it. Best once basic pacing is stable enough to isolate recovery.

End the cycle with one or two next actions. If the paper produces a long unfocused list, prioritise by recurrence, transfer value, mark cost and repair time. The paper is useful only when it makes the next decision better.

Checking-cycle paper

Reserve a defined checking budget and record which errors are found.

When to use it. Best in the final month when checking must become selective.

End the cycle with one or two next actions. If the paper produces a long unfocused list, prioritise by recurrence, transfer value, mark cost and repair time. The paper is useful only when it makes the next decision better.

Mock-cycle paper

Reproduce realistic materials, timing and conditions, then review the entire operating routine.

When to use it. Best when enough syllabus control exists for simulation to mean something.

End the cycle with one or two next actions. If the paper produces a long unfocused list, prioritise by recurrence, transfer value, mark cost and repair time. The paper is useful only when it makes the next decision better.

Confidence-cycle paper

Use a representative paper after a repair block to gather evidence that the new system can work independently.

When to use it. Best after a damaging result or confidence loss.

End the cycle with one or two next actions. If the paper produces a long unfocused list, prioritise by recurrence, transfer value, mark cost and repair time. The paper is useful only when it makes the next decision better.

Taper-cycle paper

If a late full paper is still warranted, schedule it early enough that any useful repair can occur before the examination.

When to use it. Best used sparingly because the closer the examination, the higher the bar for another full paper.

End the cycle with one or two next actions. If the paper produces a long unfocused list, prioritise by recurrence, transfer value, mark cost and repair time. The paper is useful only when it makes the next decision better.

A weekly parent-and-tutor dashboard

• Current phase: Repair, Integrate or Perform.
• Top two active mechanisms.
• One fragile mechanism.
• One maintenance area.
• Prompt burden.
• Delayed retrieval status.
• Mixed-question method selection.
• Paper completion.
• Checking yield.
• Recovery after difficulty.
• Workload and sleep.
• Next decision and evidence required.

This dashboard exposes useful progress before marks fully stabilise. Fewer recurring sign errors, stronger delayed retrieval, more independent starts, better completion and higher checking yield are all leading indicators that the system is becoming more reliable.

Frequently Asked Questions

Is 90 days enough to improve Mathematics?

Meaningful improvement is possible when high-leverage mechanisms are targeted. Ninety days cannot guarantee complete mastery of every weak area, but it is enough time to repair important dependencies, strengthen retrieval and improve paper execution.

Should Day 90 begin with a full paper?

Not automatically. A strong learner may use one as baseline evidence. A weaker learner may gain more from a short diagnostic because a full paper could mostly repeat known gaps.

How many papers should a student do?

There is no useful universal number. Paper count is constrained by how much analysis, repair and verification happens between papers.

When should mixed practice begin?

As soon as the learner has enough stability to discriminate methods productively. Small mixed sets can begin during repair and should expand in the middle phase.

When should revision become examination-focused?

When enough relevant content is known and timed conditions reveal performance issues rather than only missing chapters.

Should new Mathematics still be learned in the final month?

Yes when the gap is high-value. The final month is about selectivity, not a ban on learning.

Should there be a full paper every day in the final month?

Not by default. Papers need marking, post-mortem, repair and retrieval. More papers are useful only when later papers improve because of earlier ones.

What if mixed practice lowers scores?

That can reveal method-selection weakness hidden by topical cues. Narrow the mix and train classification before abandoning it.

What if the student takes Mathematics and A-Math?

Coordinate both plans. Keep separate subject maps plus shared algebraic dependencies, and balance heavy paper days.

How do G1, G2 and G3 affect the plan?

Confirm the actual subject level and use the correct content and papers. The countdown architecture remains useful across levels, but the mathematical demand differs.

What should parents track besides marks?

Repeated-error frequency, prompt dependence, delayed retrieval, mixed-question starts, paper completion, checking yield, recovery and workload sustainability.

What if the student already scores very highly?

Focus on stability, unfamiliar surfaces, efficient selection, checking and reduction of avoidable variance rather than simple volume.

What if the learner is still weak at Day 30?

Prioritise accessible marks, high-frequency mechanisms and targeted sections. Do not abandon learning merely because the calendar has entered the final month.

What if one bad paper causes panic?

Analyse it. Separate recurring weakness from paper-specific difficulty and convert the result into a small number of actions.

When should revision stop on the final day?

At a sensible point that protects sleep and cognitive freshness. Stop before further work becomes low-quality.

Where this page sits in the BTT examination system

• Mathematics Examination Craft — broad paper-performance layer.
• Recover After a Poor Mathematics Preliminary Examination — recovery after a poor prelim.
• How to Use Past-Year Mathematics Papers Properly — paper selection and use.
• How to Run Mathematics Mock Examinations Properly — simulation design.
• How to Do a Mathematics Examination Post-Mortem — extracting action from a completed paper.
• How to Recover After Getting Stuck in a Mathematics Examination — question-level recovery.
• How Active Recall Works for Mathematics — retrieval mechanics.
• How Spaced Practice Works for Mathematics — retention across time.
• How Interleaving Works for Mathematics — method selection from mixed work.
• Secondary Mathematics Tuition — broader Secondary route.

Final Thought: the countdown should change the job, not increase the panic

Ninety days out, repair what is structurally expensive.

Sixty days out, make Mathematics retrievable and selectable across mixed work.

Thirty days out, convert capability into stable paper performance.

Ten days out, reduce novelty and protect the system.

One day out, arrive rested enough to use what has already been built.

Repair → retrieve → mix → perform → protect.

The examination date is fixed. The productive response is not to make every remaining day equally intense. It is to make every phase do the work that phase is best suited to do.