Find My Mathematics State | What Should I Fix First?

The Mathematics Capability Atlas | How to Locate a Student Properly

A Mathematics student is not a mark. A mark is one measurement produced under one set of conditions. The learner is a moving capability system: some floors are stable, some are fragile, some ideas are deep but slow, some procedures are fast but shallow, some errors recur, and some support should already be fading.

The Mathematics Capability Atlas exists to answer a more useful question than “What grade is the student getting?” It asks: Where is the learner now, what is limiting the next move, and what evidence would show that the state has changed?

The Atlas does not replace the syllabus

The school syllabus tells us what content belongs to the learner’s course and stage. The Atlas tells us how the learner is currently operating inside that content.

Two Secondary 2 students can be studying the same chapter and need completely different interventions. One may lack a prerequisite. Another may understand the concept but retrieve slowly. Another may be fluent in topical work but fail when the method is not announced.

The chapter label is therefore only one coordinate.

Coordinate 1: school stage and curriculum route

Begin with the learner’s actual route: Primary, PSLE, Secondary, Additional Mathematics, JC, IB, IGCSE or another programme.

This tells us the curriculum environment, but not the learner state. The same age can contain different subject levels and different mathematical histories.

Coordinate 2: content object

Name the mathematical object precisely. “Weak in Math” is not a useful diagnosis.

Better descriptions include:

• fraction equivalence;
• ratio and proportional reasoning;
• signed-number operations;
• algebraic equivalence;
• linear equations;
• function notation;
• graph interpretation;
• similarity;
• trigonometric ratio structure;
• calculus rate-of-change meaning;
• probability sample-space construction.

Specific objects make repair possible.

Coordinate 3: lower-floor health

Ask which earlier capability is carrying the present work. A calculus problem may depend on indices and algebra. A trigonometry problem may depend on ratio and geometry. A probability problem may depend on fractions and systematic counting.

The Lower-Floor Law of Mathematics handles this dependency logic.

The Atlas records whether the relevant floor is stable, slow, fragile, missing or simply difficult to retrieve after delay.

Coordinate 4: depth

Depth describes how much of the mathematical structure the learner owns.

A useful depth ladder is:

1. Recognition: the student recognises the method when shown.
2. Reproduction: the student can copy or repeat the method.
3. Explanation: the student can explain the central relationship.
4. Connection: the student can link the idea to another representation or topic.
5. Transfer: the student can use the structure after the surface changes.
6. Generation: the student can construct a new argument, example, model or proof from the idea.

A high mark on routine work may coexist with modest depth if the learner relies on familiar patterns.

Coordinate 5: load

Load asks how much attention the work consumes.

A student may know how to simplify fractions but need so much concentration that the skill interferes with an advanced algebra problem. Another may calculate accurately but become overloaded when language, graph interpretation and multi-step planning appear together.

Fluency reduces lower-level load so working memory can be used for higher-level reasoning.

Coordinate 6: transfer

Transfer asks whether the learner recognises the same structure in a changed form.

Examples:

• a ratio problem moves from recipes to maps;
• a linear equation appears inside geometry;
• percentage becomes compound growth;
• a graph replaces an equation;
• the same function is presented symbolically, numerically and visually.

Transfer is one of the strongest indicators that learning has become portable.

Coordinate 7: retrieval

Can the student bring the Mathematics back after time has passed?

Immediate performance after a lesson can be strong because the explanation remains active in short-term memory. Delayed retrieval shows whether the knowledge has become more durable.

The Atlas distinguishes “understands when reminded” from “can retrieve independently”. The interventions are different.

Coordinate 8: error signature

Errors should be classified by mechanism, not insult.

Common signatures include:

• sign-control error;
• fraction-equivalence error;
• wrong base in percentage change;
• illegal algebraic cancellation;
• representation error;
• method-selection error;
• unit error;
• copying/transcription error;
• domain-condition error;
• checking failure;
• paper-control error.

The Mathematics Fracture and Repair Map develops the repair process once a recurring signature is located.

Coordinate 9: prompt dependence

What must another person supply before the student can continue?

A prompt ladder can be read from heavy to light:

• full worked example;
• method named;
• representation supplied;
• first step given;
• strategic question;
• general encouragement;
• no prompt.

Progress often appears as the first prompt moving later and becoming less specific.

Coordinate 10: method selection

Can the learner decide what to do when the page does not announce the chapter?

Topical practice can hide weak selection because every question sits under the same heading. Mixed work removes the cue.

A student may know ten methods separately and still fail because they cannot decide which structure is present.

Coordinate 11: representation control

Mathematics can be represented through objects, diagrams, tables, graphs, symbols, equations and words.

Strong learners can choose and change representations deliberately. A child may move from bar model to equation. A Secondary student may move from equation to graph. An A-Math student may use a graph to verify function behaviour.

Representation flexibility is a major part of mathematical control.

Coordinate 12: checking and verification

Can the learner gather independent evidence that the answer deserves trust?

Possible checks include:

• estimation;
• reverse operation;
• substitution;
• unit consistency;
• graph behaviour;
• special-case testing;
• alternative representation;
• boundary checks.

The strongest checking is self-initiated rather than prompted.

Coordinate 13: recovery

What happens when the learner gets stuck?

A fragile student may freeze or wait. A more mature learner has recovery moves:

• return to the givens;
• draw or redraw the representation;
• test a simpler case;
• work backwards;
• estimate the answer range;
• leave the question temporarily;
• locate the last valid line;
• change method.

Recovery is not a minor skill. Advanced Mathematics always contains unfamiliarity.

Coordinate 14: speed and fluency

Speed should be interpreted carefully. Fast and wrong is not fluency. Slow and thoughtful is not automatically weak.

The useful question is where the time is being spent.

If basic algebra consumes attention inside every advanced question, fluency may be limiting. If time is spent choosing among methods, mixed selection may be the issue. If the student writes slowly but reasons accurately, the intervention is different.

Coordinate 15: score stability

One high mark shows a peak. Stability shows whether capability survives different topic mixes and paper conditions.

Watch the lower end of performance as well as the best score. A rising floor—fewer severe collapses—can be a powerful sign of progress.

The Parent Mathematics Dashboard handles this parent-facing evidence layer.

Coordinate 16: learner motion

Static diagnosis is incomplete. The Atlas also asks whether the learner is moving.

Useful motion states include:

• falling: more topics are becoming unstable;
• recovering: one or more bottlenecks are being repaired;
• stabilising: understanding is present but execution or retrieval is inconsistent;
• advancing: current work is secure and future content can be opened;
• deepening: the learner is strong and needs proof, modelling or non-routine transfer;
• commissioning: the system is preparing for a major examination or transition.

The next action depends on motion, not only location.

The eight common Mathematics states

The original Find My Mathematics State route can be made more precise through eight common states.

State A: I do not understand the current idea

Signal: explanations feel disconnected, definitions are unclear, and examples cannot be reconstructed.

Next action: reduce the topic to the mathematical object, choose a clear representation and rebuild meaning.

State B: I understand, but I cannot execute reliably

Signal: the student can explain but makes frequent local errors.

Next action: focused procedure practice, notation discipline and checking.

State C: I can do it now, but I forget later

Signal: strong lesson performance, weak delayed return.

Next action: spaced retrieval and interleaving.

State D: I can do the topic, but not mixed questions

Signal: topical worksheets strong, mixed work weak.

Next action: method-selection practice and transfer.

State E: I am slow

Signal: correct work but excessive time.

Next action: identify the expensive step before adding time pressure.

State F: I repeat the same mistake

Signal: one error signature returns across weeks or topics.

Next action: fracture analysis, minimal repair and delayed retest.

State G: I know the Mathematics but my examination performance is unstable

Signal: homework and topical work strong, papers volatile.

Next action: pacing, mixed retrieval, recovery and checking under time.

State H: current Mathematics is stable and I need the next challenge

Signal: strong independent work, durable retrieval, good transfer.

Next action: teach ahead, deepen, model, prove or extend.

A student can occupy more than one state

A learner can understand algebra deeply and still be slow. They can retrieve well but select methods poorly. They can score highly while depending heavily on tuition prompts.

The Atlas is therefore multidimensional. It does not force the child into one permanent label.

The 90-second routing check

When a student says “I can’t do this”, ask:

1. Do you understand what the question is asking?
2. Can you identify the mathematical object?
3. Do you know one possible representation?
4. Do you know a method that might apply?
5. Can you execute the first step?
6. Can you retrieve the relevant prerequisite?
7. Can you check the work if you finish?

The first “no” is often more useful than the final wrong answer.

Evidence hierarchy: what is stronger?

Different evidence has different strength.

Diagnosis should use enough evidence for the decision, not every available metric.

Locate the first invalid or uncertain move

The final wrong answer can contain several downstream consequences. The most useful diagnostic point is often the first place where the route becomes invalid or uncertain.

If the student expands incorrectly, every later line can be wrong even if the equation method is understood. Repair the first failure rather than treating the entire page as equally broken.

Do not repair everything at once

A student may have many imperfect areas. Tuition time is limited. Prioritise load-bearing constraints.

Ask:

• Which weakness affects the greatest amount of current Mathematics?
• Which one prevents access to the next school topic?
• Which one causes repeated marks loss across papers?
• Which one can be repaired with high leverage?

The Atlas should simplify the next action, not create an endless deficit list.

The minimum-repair rule

Descend only until the first relevant unstable dependency is found. Repair it. Return upward.

This protects the learner from being sent back through years of unrelated worksheets whenever an advanced topic becomes difficult.

The reconnection rule

A repaired prerequisite must return to the original topic quickly. Otherwise the student may become good at the isolated exercise but still fail to recognise the same structure inside current Mathematics.

The delayed-return rule

Immediate success after teaching is weak evidence. Revisit after time has passed. If the learner can retrieve and transfer without the original support, the state has genuinely changed.

The support-fading rule

As control grows, help should become lighter. A tutor who continues to name the method, supply the representation and check every answer can hide the learner’s actual state.

Progress is visible when the student carries more of the route.

The movement test

After any intervention, ask what changed:

• fewer prompts?
• better retrieval?
• same error less frequent?
• changed question handled?
• faster without accuracy loss?
• more independent checking?
• more stable paper performance?

If nothing moves, change the strategy rather than adding more of the same activity.

Primary learner Atlas

For younger students, emphasise number sense, representation, operation meaning, language, willingness to attempt and checking. Do not over-weight speed too early.

A P1 learner can be mathematically healthy even if written speed is modest, provided quantity, decomposition and relationships are developing well.

Upper-Primary Atlas

From P3 to P6, watch multiplication/division structure, fractions, proportional reasoning, word-problem translation, model choice, retrieval and mixed-method selection.

As PSLE approaches, add paper pacing, recovery and broad-syllabus integration.

Secondary Atlas

At Secondary level, signed numbers, algebraic equivalence, graph interpretation, proportional reasoning and notation become major load-bearing floors.

Method selection and cross-topic transfer also become more important because the learner encounters more symbolic compression.

Additional Mathematics Atlas

For A-Math, the shared symbolic spine is critical. Indices, algebra, functions and graphs support multiple later topics.

If logarithms, trigonometry and calculus all look weak, do not assume three unrelated failures. Search for a shared floor.

Examination Atlas

When content is broadly secure, locate the examination state separately.

• Can the learner finish?
• Do they spend time in proportion to marks?
• Can they recover after getting stuck?
• Does checking survive the final section?
• Are severe score collapses becoming less frequent?

Examination craft is a layer above content, not proof that content is absent.

Transition Atlas

Major transitions change the load and language of Mathematics.

Examples:

• Primary 6 → Secondary 1;
• Secondary 2 → upper Secondary;
• E-Math → Additional Mathematics;
• Secondary 4 → JC;
• Singapore → IB/IGCSE or another system;
• school Mathematics → university proof and abstraction.

The Atlas asks which old capabilities remain useful and which new operating habits are required.

Teaching-ahead Atlas

A learner is ready to move ahead when current work is stable, retrieval is durable, lower floors are healthy and school participation remains strong.

The Teaching Mathematics Ahead of School guide handles this future corridor.

Strong learner Atlas

High marks do not end diagnosis. The question changes.

Can the learner:

• prove rather than only calculate?
• generalise?
• construct counterexamples?
• model assumptions?
• solve unfamiliar problems?
• compare methods?
• work independently from definitions?

Depth can be more valuable than moving further ahead.

Recovering learner Atlas

For a falling student, locate the highest-leverage repair rather than building a giant backlog list.

One shared algebra floor may improve several topics. One retrieval routine may stop repeated forgetting. One examination-control repair may stabilise marks without re-teaching the syllabus.

Parent use: ask for the coordinate, not the label

Instead of “Is my child weak at Math?”, ask:

• Which object is unstable?
• What prerequisite does it depend on?
• Is the problem understanding, retrieval, execution, selection or paper control?
• How much prompting is currently needed?
• What would improvement look like in actual work?

This turns anxiety into a bounded learning decision.

Student use: convert “I don’t know” into a location

When stuck, identify the first uncertain layer:

• I do not understand the words.
• I do not know what the object is.
• I cannot build a representation.
• I cannot remember the prerequisite.
• I know the method but cannot execute it.
• I can execute but do not know which method belongs.
• I finish but cannot check.

That location is enough to begin repair.

Tutor use: teach the state, not the worksheet

The same worksheet can require different interventions for different students.

One learner needs explanation. Another needs silence and independent attempt. Another needs mixed transfer. Another needs timed execution. The tutor should respond to the learner state rather than deliver the same teaching performance to everyone.

Capability Atlas release standard

The diagnosis is good enough when it produces one clear next action and one clear test of change.

Examples:

• Repair signed-number distribution; retest inside equations one week later.
• Reduce prompts on fraction word problems; check whether the learner starts independently.
• Mix ratio and percentage questions; see whether method selection remains accurate.
• Run timed paper clusters; check whether final-section accuracy stabilises.
• Open the next topic; verify that current retrieval remains durable.

The end point is not perfect diagnosis

The Atlas should not become a permanent monitoring bureaucracy. Once the next action is clear, teach.

After the intervention, sample enough evidence to decide whether the state changed. Then update the route.

Return routes

Use The Parent Mathematics Dashboard when the reader job is tracking progress, The Mathematics Fracture and Repair Map when a recurring error has been found, and Bukit Timah Mathematics Tuition when direct teaching support is being considered.

Closing principle

Locate before prescribing. Repair only what is load-bearing. Verify the change. Then move the learner forward.

The Mathematics Capability Atlas is not a ranking system. It is a navigation system for deciding what should happen next.

Capability Atlas Decision Fieldbook: From Observation to the Next Mathematics Move

The Atlas becomes useful when it changes what happens in the next lesson. A diagnostic description should therefore end in a decision. The fieldbook below translates common observations into more precise hypotheses, tests and interventions.

Observation: “My child is careless.”

Possible underlying states include:

• working is compressed too aggressively;
• sign changes are unstable;
• the student skips unit checks;
• copying from one line to the next is inaccurate;
• the learner rushes because the paper is too long;
• working memory is overloaded by weak lower-floor fluency.

The Atlas should not record “careless” as a personality trait. It should locate the recurring error signature.

Test: collect several errors and classify where the first invalid step occurs.

Next action: repair the dominant mechanism and retest it after delay.

Observation: “My child understands in tuition but not at home.”

This often signals support dependence, weak retrieval or insufficient independent reconstruction.

Test: after a guided example, remove notes and ask for a changed question. Record the first point where help becomes necessary.

Next action: fade prompts, schedule delayed retrieval and require an independent first attempt before support.

Observation: “My child can do worksheets but not exams.”

Possible issues include:

• method selection;
• mixed-topic switching;
• time allocation;
• retrieval under pressure;
• recovery after a hard question;
• checking disappears late in the paper.

Test: compare topical performance with a mixed timed set.

Next action: train the layer that collapses under integration rather than re-teaching every topic.

Observation: “My child needs more practice.”

This statement is incomplete until the practice target is named.

Practice can train:

• retrieval;
• fluency;
• method selection;
• transfer;
• paper pacing;
• checking;
• proof reconstruction.

More questions of the same type may be low value if the real problem is mixed selection or conceptual misunderstanding.

Observation: “My child forgot everything after the holidays.”

Do not assume the concept is gone. Retrieval can weaken while underlying structure remains.

Test: give one clean problem without support. If the learner is stuck, provide one small cue and see whether the whole route returns.

Interpretation: rapid recovery after a small cue suggests retrieval weakness rather than complete conceptual loss.

Next action: spaced retrieval rather than full re-teaching.

Observation: “My child is very slow.”

Locate where time accumulates.

• reading the question;
• building a representation;
• choosing a method;
• basic algebra or arithmetic;
• writing every step;
• checking repeatedly;
• recovering after mistakes.

Speed is an output of several mechanisms. Timed drills are appropriate only when the bottleneck is fluent execution.

Observation: “My child is fast but makes many mistakes.”

The learner may be compressing too many steps, skipping representation or not checking.

Test: require the student to slow only at high-risk decision points while preserving natural speed elsewhere.

Next action: build selective checking and notation discipline rather than slowing the entire paper indiscriminately.

Observation: “My child gets the answer but cannot explain.”

This may indicate procedural strength with shallow conceptual depth.

Test: ask for another representation, a reason the method works or a counterexample showing when it would fail.

Next action: deepen the object and its conditions without discarding the useful procedural fluency.

Observation: “My child can explain but cannot get the answer.”

This may indicate execution instability, notation weakness or weak retrieval of routine facts.

Test: isolate the local transformations needed after the concept is understood.

Next action: focused fluency practice and error checking rather than another long conceptual lecture.

Observation: “My child only struggles with word problems.”

Possible bottlenecks include:

• language parsing;
• relationship identification;
• representation choice;
• operation selection;
• multi-step planning.

Test: present the same relationship first as a diagram or equation. If execution becomes easy, the main issue lies before calculation.

Observation: “My child relies on keywords.”

Keywords can help early access but become dangerous when the same word appears in different structures.

Test: create pairs of questions using similar vocabulary but different operations.

Next action: train relationship recognition and representation before operation selection.

Observation: “My child cannot remember formulas.”

Separate formula recall from formula use.

A student can remember a formula and misuse it because conditions or variable roles are unclear. Another can understand the relationship and reconstruct it but recall slowly.

Test: provide the formula and see whether the student can map variables, state conditions and verify the output.

Next action: choose recall training only if recall is genuinely the limiting layer.

Observation: “My child uses the calculator for everything.”

Calculator dependence may reflect weak arithmetic fluency, low confidence or a habit formed by easy tool access.

Test: ask for estimation before calculation and a simple non-calculator equivalent.

Next action: preserve useful technology while rebuilding number sense where it is still load-bearing.

Observation: “My child does not check.”

The student may not know what checking means beyond repeating the calculation.

Test: ask for a different verification route.

• estimate;
• substitute;
• reverse;
• inspect units;
• compare with a graph;
• test a special case.

Next action: attach a specific check to each high-value topic until self-initiation develops.

Observation: “My child loses confidence after one difficult question.”

This may be a recovery problem rather than a content problem.

Test: give a mixed set containing one deliberately hard item early. Observe whether the learner can leave it, preserve the rest of the work and return later.

Next action: build paper recovery as an explicit examination skill.

Observation: “My child always wants the teacher to confirm the answer.”

This is external verification dependence.

Test: before giving feedback, ask the student what evidence they have that the answer is plausible.

Next action: require an independent check before tutor confirmation.

Observation: “My child is strong but bored.”

The learner may need depth rather than more routine volume.

Possible extensions:

• prove a familiar result;
• generalise a pattern;
• compare two methods;
• construct a counterexample;
• model a real system;
• solve an unfamiliar olympiad-style problem;
• teach the concept to another learner.

A strong student should not be forced to repeat easy work merely to fill lesson time.

Observation: “My child is ahead of school.”

Being ahead is not itself a state of mastery.

Check:

• are current topics durable?
• are older topics still retrievable?
• can the learner handle unseen variations?
• does school remain useful?
• is support dependence falling?

If yes, the future corridor may remain open. If not, reduce syllabus distance and consolidate.

Observation: “My child wants A-Math but E-Math is inconsistent.”

Do not answer this only from one mark. Inspect the load-bearing algebraic floor.

Additional Mathematics increases symbolic density. Algebraic equivalence, indices, functions and graph understanding become more expensive if current foundations are unstable.

Test: sample algebraic manipulation, function language, graph interpretation and delayed retrieval.

Next action: repair shared symbolic floors before or while opening the A-Math corridor.

Observation: “My child scores well but makes strange basic mistakes.”

Strong learners can compensate for weak floors using speed, memory or pattern familiarity.

Test: isolate the basic operation in a clean form, then test it inside an advanced problem.

Next action: refine the floor without treating the learner as globally weak.

Observation: “My child has many tuition classes but is still struggling.”

The Atlas should ask whether support systems are aligned or merely additive.

Possible issues include:

• too little independent study time;
• conflicting methods;
• insufficient retrieval between lessons;
• fatigue;
• help arriving before a genuine attempt.

Sometimes the next action is to simplify the support environment.

The capability profile is not permanent

A student can move quickly once the correct bottleneck is repaired. The Atlas should therefore carry dates or recent evidence mentally rather than turning an old diagnosis into a permanent identity.

“Weak algebra” may have been accurate three months ago and wrong today.

Atlas state: Repair

Use when a load-bearing prerequisite or recurring error is limiting current work.

Lesson emphasis:

• isolate the mechanism;
• make the relationship visible;
• practise narrowly;
• reconnect upward;
• retest later.

Atlas state: Stabilise

Use when the learner understands but performance is inconsistent.

Lesson emphasis:

• retrieval;
• fluency;
• notation;
• independent checking;
• varied practice.

Atlas state: Transfer

Use when topical work is strong but unfamiliar surfaces cause collapse.

Lesson emphasis:

• change representations;
• remove chapter labels;
• mix methods;
• use new contexts;
• ask the learner to explain the structural clue.

Atlas state: Perform

Use when content is broadly secure and the next constraint is examination expression.

Lesson emphasis:

• time per mark;
• mixed papers;
• recovery;
• checking;
• answer presentation;
• score stability.

Atlas state: Advance

Use when current work is stable and future content can be introduced calmly.

Lesson emphasis:

• essential vocabulary;
• new notation;
• central relationship;
• prerequisite bridge;
• one or two clean examples.

Atlas state: Deepen

Use when the learner needs richer mathematical demand rather than curriculum speed.

Lesson emphasis:

• proof;
• generalisation;
• modelling;
• unfamiliar transfer;
• method comparison;
• counterexample construction.

Atlas state: Transition

Use when the learner is moving into a new mathematical environment.

Examples include:

• Primary to Secondary;
• E-Math to A-Math;
• Secondary to JC;
• Singapore to IB/IGCSE;
• school to university.

Lesson emphasis should focus on what the new environment assumes, what notation changes and what old capability must be carried forward.

Atlas state: Release

Release is the state in which external support can be reduced.

Evidence:

• independent start;
• durable retrieval;
• reliable transfer;
• self-checking;
• stable school performance;
• help-seeking only after a genuine attempt.

A tuition system should recognise release as success.

A monthly Atlas review

Once a month, ask five questions:

1. What is the learner’s strongest stable capability?
2. What is the main current bottleneck?
3. What evidence shows it?
4. What intervention is being used?
5. What would count as release from that intervention?

This is enough to preserve direction without turning learning into constant measurement.

A term Atlas review

At the end of a school term, zoom out.

• Did lower floors strengthen?
• Did prompt dependence fall?
• Did retrieval last longer?
• Did transfer improve?
• Did score volatility narrow?
• Did the learner move from repair toward stability or extension?

The term review should shape the next phase.

A year Atlas review

The largest question is developmental:

Can the learner now carry more Mathematics with less external control than one year ago?

A rising syllabus level without rising independence is incomplete progress. The system should mature as well as expand.

Capability Atlas and the small-group classroom

Three-student tuition can support high-resolution observation because the tutor can watch working closely enough to distinguish concept, retrieval, method selection and checking.

But small size alone does not create diagnosis. The tutor still needs to use the visibility intelligently: inspect first attempts, compare representations, vary problems and fade help.

Capability Atlas and parents

Parents do not need access to a complicated scoring dashboard. A useful update can remain simple:

Stable: what the child can now carry alone.

Constraint: the main thing currently limiting progress.

Evidence: one or two observations.

Next action: what the next cycle will target.

Capability Atlas and the student

The student-facing version should remain even simpler:

“Here is what you can already do. Here is the one thing that is slowing you down. Here is how we will train it. Here is how we will know it is fixed.”

That language protects agency and keeps the learning job bounded.

Capability Atlas and examination seasons

As major examinations approach, the Atlas should not keep inventing new diagnoses. The emphasis shifts toward commissioning known capability under realistic constraints.

Priorities become:

• retrieval breadth;
• paper pacing;
• error containment;
• checking;
• recovery;
• sleep and sustainable workload.

Capability Atlas and post-examination transitions

After PSLE or another major examination, the Atlas can shift from performance back to development. Repair hidden floors, open future notation gently and prepare the next mathematical language without carrying examination urgency forward unnecessarily.

The Atlas should never become a rank

Do not score children on a single “capability number”. Mathematical profiles are multidimensional and state-dependent.

A learner can be deep but slow, fast but fragile, conceptually strong but examination-unstable, or highly independent in one domain and dependent in another.

The Atlas exists to route action, not label worth.

The Atlas should never become surveillance

Do not record every mistake as a permanent data point. Sample enough evidence to understand the pattern, intervene, then allow the student to learn without feeling constantly inspected.

The Atlas should never become an excuse to delay progress forever

Foundations do not need to be perfect before the learner can move forward. They need to be stable enough to carry the next layer.

Progress and repair can happen together.

Final routing map

If the problem is:

• unclear state → start here at Find My Mathematics State;
• unstable prerequisite → Lower-Floor Law;
• repeated error → Fracture and Repair Map;
• progress evidence → Parent Mathematics Dashboard;
• future preview → Teaching Mathematics Ahead;
• service fit → Bukit Timah Mathematics Tuition.

Final Atlas compact

Locate → name the limiting mechanism → choose one intervention → verify movement → reduce support → move again.

The student is not a fixed point on the map. The whole purpose of the map is to help them move.

Capability Atlas Diagnostic Task Bank: Small Questions That Reveal Large States

A good diagnostic task is not merely “easy” or “hard”. It is chosen because a particular response reveals something about the learner’s mathematical state. The tasks below are deliberately small. Their value comes from what the tutor watches while the student works.

Diagnostic 1: number sense before written method

Ask a Primary learner:

Which is closer to 500: 487 or 462? How do you know without subtracting exactly?

This reveals magnitude sense and estimation. A student who immediately reaches for a written algorithm may still be correct, but the response shows whether approximate reasoning is available.

Diagnostic 2: decomposition flexibility

Ask:

Show three different ways to make 20 using two addends.

A learner who produces 10+10, 12+8 and 17+3 is demonstrating flexible number decomposition rather than one memorised bond.

This matters because later algebra and mental calculation depend on seeing quantities in multiple equivalent forms.

Diagnostic 3: multiplication meaning

Ask:

What is the difference between 4 groups of 6 and 6 groups of 4? Are the answers the same? Is the story the same?

The numerical product is the same, but the representation is different. This distinguishes arithmetic result from relational meaning.

Diagnostic 4: fraction magnitude

Ask:

Which is larger: 3/5 or 4/7? Explain without converting both to decimals first.

Possible routes include cross-products, benchmark fractions or common denominators. The tutor is looking for fraction structure, not one preferred technique.

Diagnostic 5: same whole

Ask:

Is 1/2 always bigger than 1/3?

The correct answer depends on the same whole. Half of a small object can be smaller than one-third of a large object.

This question reveals whether the student treats fraction symbols independently of the quantities they describe.

Diagnostic 6: ratio as multiplicative comparison

Ask:

A:B = 2:3. If A increases by 4, can you say the new ratio is 6:3?

The tutor is checking whether the learner understands that changing one quantity changes the relationship and that ratio parts are not the quantities themselves unless a scale is established.

Diagnostic 7: percentage base

Ask:

A price rises from $80 to $100. What is the percentage increase? If it then falls from $100 to $80, is the percentage decrease the same?

Increase: 20/80 = 25%.

Decrease: 20/100 = 20%.

This reveals whether the learner can identify the reference quantity.

Diagnostic 8: signed-number floor

Ask:

-3 - (-5)

Do not only record the answer. Ask the learner to explain what subtracting a negative means or represent it on a number line.

A correct answer obtained by a memorised sign rule may still be fragile if the relationship is not understood.

Diagnostic 9: equality meaning

Ask:

What number makes this true? 7 + 5 = __ + 4.

A learner who writes 12 because they interpret the equals sign as “write the answer next” is showing an equality misconception.

The correct missing number is 8.

Diagnostic 10: algebraic equivalence

Ask whether:

3(x+2) and 3x+6

are the same, and what “the same” means.

The tutor is looking for equivalence across representations, not only expansion skill.

Diagnostic 11: illegal cancellation

Ask the learner to simplify:

(x+2)/x.

If they “cancel the x” and write 2, factor structure is unstable. The numerator is a sum, not a product with x as a common factor.

This tiny task can explain errors across algebraic fractions, equations and A-Math.

Diagnostic 12: equation balance

Ask:

Why does adding 4 to both sides of an equation preserve the solution?

The student may know the rule operationally. The explanation reveals whether equality is understood as a preserved relationship.

Diagnostic 13: function notation

Let f(x)=2x+3. Ask:

• What is f(4)?
• What does f(4) mean?
• Is f the same thing as f(x)?

This separates substitution fluency from understanding the function as an object.

Diagnostic 14: graph representation

Give a simple linear graph and ask:

What does the gradient tell us about the relationship between x and y?

The tutor is checking whether gradient is a number to calculate or a rate relationship to interpret.

Diagnostic 15: graph and equation connection

Show y=2x+1 and ask how the numbers 2 and 1 should appear in its graph.

This reveals whether symbolic and visual representations are connected.

Diagnostic 16: method selection

Give two equations:

x²-5x+6=0

and

x²-5x+1=0.

Ask which method the learner would try first for each and why.

The first factorises neatly; the second may invite the quadratic formula or completing the square. The state being tested is not execution but route selection.

Diagnostic 17: trigonometric ratio meaning

Ask:

Why does sin θ have the same value for every right triangle with the same acute angle θ?

A deeper answer invokes similarity and proportional side lengths. A shallow answer may only repeat “opposite over hypotenuse”.

Diagnostic 18: calculus lower-floor split

Ask the student to differentiate x^-2.

If they understand the power rule but cannot work confidently with negative indices, the calculus concept may be fine while the index floor is unstable.

This prevents unnecessary re-teaching of differentiation.

Diagnostic 19: probability sample-space construction

Roll two fair dice and ask:

What is one elementary outcome? Are the sums 2 through 12 equally likely?

The student who treats sums as equally likely has compressed the sample space too far.

Diagnostic 20: statistics and the average

Give:

Set A: 5,5,5,5,15.

Set B: 7,7,7,7,7.

Both means are 7.

Ask what the mean fails to tell us. This reveals whether the learner understands spread and distribution rather than treating the mean as a complete summary.

Diagnostic 21: retrieval after delay

Do not announce the topic. One week after teaching simultaneous equations, place one simultaneous-equation problem inside a mixed set.

The task tests two things at once:

• does the method return?
• does the learner recognise when it belongs?

Diagnostic 22: transfer across context

Teach ratio using recipes. Later test scale maps or mixture concentration without announcing “ratio”.

If the learner can identify the multiplicative structure, the knowledge is becoming portable.

Diagnostic 23: checking behaviour

Give a problem where one plausible answer violates an obvious bound, such as a probability greater than 1 or a length that is negative.

Observe whether the student notices without prompting.

The state being tested is not only calculation but structural checking.

Diagnostic 24: recovery after a planted obstacle

Give a mixed set containing one question deliberately beyond the student’s current comfort.

Observe:

• do they freeze?
• do they skip and return?
• do they simplify?
• do they identify what is unknown?
• do they preserve the rest of the paper?

This is a realistic examination-control diagnostic.

Diagnostic 25: prompt dependence

On a familiar question, wait longer than usual before helping.

The tutor should observe whether the student can generate a first move independently when the automatic rescue is removed.

Silence can be diagnostic.

Diagnostic 26: proof readiness

Ask:

The first five examples support a pattern. Is that enough to prove it for all integers?

The learner should distinguish evidence from universal proof.

At stronger levels, ask for a counterexample to an over-broad claim.

Diagnostic 27: notation load

If a student struggles with formal notation, translate the same idea into ordinary language.

If understanding immediately improves, the bottleneck may be notation access rather than concept.

The intervention should then build translation fluency rather than re-teach the entire mathematical object.

Diagnostic 28: working-memory load

Break one multi-step problem into its component operations. If each component is individually secure but the full problem collapses, coordination load may be the issue.

Possible interventions include external representations, clearer working and improved fluency on high-use components.

Diagnostic 29: examination pacing

Take a completed school paper and mark where time was spent.

Look for:

• too much time on low-mark questions;
• late-paper rush;
• repeated full restarts;
• excessive checking of already secure work;
• failure to leave a blocked item.

Pacing problems should be treated as a control layer, not automatically as weak content.

Diagnostic 30: score volatility

Compare several papers rather than one result.

Ask which topics or conditions cause large drops. A high average with occasional severe collapse suggests a different state from consistently moderate performance.

The floor mark can be as informative as the peak.

Diagnostic 31: readiness to teach ahead

Before opening a future topic, test:

• current topic retrieval;
• prerequisite health;
• mixed transfer;
• school independence.

If those are stable, a future corridor can be useful. If not, acceleration may hide unfinished learning.

Diagnostic 32: readiness to reduce tuition

Ask whether the learner can:

• start school homework independently;
• retrieve older work;
• seek help after a genuine attempt;
• check answers;
• maintain stable performance.

If yes, reduce support gradually. Release is part of good tuition design.

Diagnostic 33: strong learner depth test

Take a familiar theorem or formula and ask:

• Why is it true?
• When does it fail?
• Can you generalise it?
• Can you prove a special case?
• Can you construct a counterexample to a related false claim?

This distinguishes syllabus familiarity from mathematical depth.

Diagnostic 34: model judgement

Give a simple linear model and ask:

What assumption would make this model stop being trustworthy?

The learner who can calculate but cannot state assumptions has only part of the modelling capability.

Diagnostic 35: AI/calculator audit readiness

Show a machine-generated solution with one planted domain error.

Ask the student to find the first invalid or incomplete step.

This tests whether technology output is being treated as authority or candidate evidence.

Diagnostic 36: accessibility versus mathematics

Present the same mathematical task in a clearer layout. If performance improves sharply without any change to the mathematical information, the original interface may have been adding irrelevant load.

This does not automatically diagnose an individual condition. It simply reveals that presentation can affect access.

Diagnostic 37: learner explanation versus memorised wording

Ask the student to explain the idea using a different example from the one taught.

If the explanation survives the changed example, understanding is stronger than if the learner reproduces a memorised sentence tied to one case.

Diagnostic 38: concept versus vocabulary

If a learner cannot answer “What is the gradient?” try “How much does y change when x increases by one?”

If the relationship is understood but the term is not, vocabulary is the bottleneck.

Diagnostic 39: method versus representation

If a word problem is hard, supply only the representation—not the method. If the learner can then solve it, representation construction is the likely bottleneck.

This is a more precise diagnosis than “word problems weak”.

Diagnostic 40: final Atlas release question

After any repair, ask one changed problem without announcing the repaired skill.

If the student recognises the structure, executes reliably and checks independently, the repair is becoming operational.

How the Atlas Prevents Over-Teaching

One danger in tuition is solving the wrong educational problem very thoroughly. A student who needs retrieval gets another explanation. A student who needs transfer gets another topical worksheet. A student who needs examination pacing gets another chapter review.

The Atlas prevents this by forcing the tutor to name the mechanism before prescribing the intervention.

The correct teaching response can be “do less”

If the student already understands the concept, more explanation can create passivity. The better move may be to remove hints.

If the student already retrieves well, more drill can waste time. The better move may be transfer.

If the student is independent and stable, the better move may be to reduce tuition.

The correct teaching response can be “go down one floor”

If advanced work keeps failing because of one shared prerequisite, a short targeted repair can unlock more progress than pushing the current chapter harder.

The correct teaching response can be “stay at the same level but deepen”

A strong learner does not always need next year’s syllabus. Proof, modelling, method comparison and unfamiliar transfer can create deeper mathematical growth without racing ahead.

The correct teaching response can be “switch to examination craft”

Once content is broadly secure, the bottleneck may be pacing, answer presentation, recovery or checking. Re-teaching content then has diminishing value.

The Atlas Compact for Parents

Ask four questions:

1. What is stable?
2. What is the main current constraint?
3. What evidence shows that?
4. What will we do next, and how will we know it worked?

If those four answers are clear, the learning plan has enough resolution.

The Atlas Compact for Students

Ask:

1. What part do I understand?
2. Where exactly do I first get stuck?
3. What earlier skill is this using?
4. Can I represent it another way?
5. How will I check the result?

The student does not need to diagnose everything alone. They need a language for locating uncertainty.

The Atlas Compact for Tutors

Observe → locate → isolate → teach → reconnect → vary → delay → verify → fade.

That sequence protects against generic remediation and permanent dependence.

Final Thought

The Mathematics Capability Atlas is valuable because it gives a learner somewhere more precise to stand than “good at Math” or “bad at Math”.

A student can be located. A limiting mechanism can be named. A repair can be chosen. Progress can be tested. Support can be reduced.

Once that happens, the map has done its work. The learner can move.

Capability Atlas Release Standard: When the Diagnosis Has Done Enough

The Atlas should stop diagnosing when the next action is already clear. Its purpose is not to create a permanent learner file containing every weakness. It is to give enough resolution to choose the next useful move, then get out of the way while the learner changes.

Release test 1: the bottleneck is specific

“Weak at Mathematics” is not specific enough. “Signed-number distribution is causing recurring algebra errors” is specific. “Mixed-topic method selection is unstable” is specific. “Retrieval falls sharply after two weeks” is specific.

A useful diagnosis names a mechanism the tutor can actually train.

Release test 2: the intervention matches the mechanism

Concept problems need explanation or representation. Retrieval problems need spaced return. Execution problems need focused practice. Transfer problems need varied surfaces. Examination-control problems need timed integration and recovery training.

If the treatment does not match the mechanism, more effort may produce little change.

Release test 3: one piece of evidence can show movement

The learning plan should name what will look different if it is working.

• fewer sign errors in mixed algebra;
• less prompting before the first step;
• older topics retrieved after delay;
• changed wording handled successfully;
• paper completion rate improves without accuracy collapse;
• checking becomes self-initiated.

This makes progress visible before the next large examination.

Release test 4: the learner can carry the repaired capability into real work

A foundation is not repaired because a remedial worksheet was completed. It is repaired when the skill reappears inside current Mathematics and functions there without special announcement.

That is why reconnection and transfer are part of every serious repair.

Release test 5: support can fade

If the learner improves only while receiving the same level of prompting, the state has changed less than it appears.

A strong intervention should make some external support unnecessary over time.

Release test 6: the next state is clearer

After repair, the learner may move into stabilisation. After stabilisation, into transfer. After transfer, into examination performance or extension. The Atlas is dynamic.

The student should not remain trapped under the label of the problem that brought them to tuition.

Final parent sentence

A useful parent update can fit into one sentence:

“This is what is stable, this is the main constraint, this is the evidence, and this is what we are doing next.”

When that sentence is clear, the Mathematics Capability Atlas has enough resolution to guide the next cycle.

Final student sentence

“Here is what I can already do, here is where I first get stuck, and here is the one thing I am training next.”

That is a healthier and more useful internal model than “I am bad at Math”.

Final principle

Locate accurately. Intervene narrowly. Verify movement. Reduce support. Move forward.

The Atlas is complete enough when the learner has a route.

Closing Navigation: A Map Is Useful Only When It Changes Direction

The Mathematics Capability Atlas should leave every reader with a simpler next step than they had before. A parent who arrives with “My child is weak at Math” should leave knowing whether the main issue is a lower-floor prerequisite, retrieval, execution, transfer, examination control or something else. A student who arrives with “I don’t know” should leave knowing where the uncertainty first begins. A tutor should leave knowing what to train and what to stop over-teaching.

If the learner is confused

Return to the mathematical object and representation. Reduce symbolic compression until the relationship becomes visible.

If the learner understands but cannot perform

Stabilise execution through focused practice, clean notation and independent checking.

If the learner performs but forgets

Use spaced retrieval and delayed return instead of repeating the original explanation in full.

If the learner succeeds only on familiar worksheets

Change the surface. Mix topics. Remove labels. Ask the learner to identify the structural clue that makes the method appropriate.

If the learner knows the Mathematics but examination results remain unstable

Move to pacing, recovery, mixed-paper control and checking under time.

If the learner is strong and independent

Open depth, proof, modelling, unfamiliar transfer or the next future corridor rather than adding routine volume.

If the learner no longer needs the same support

Reduce it. The Atlas is not designed to keep students inside tuition forever. It is designed to help support become more precise and then less necessary as control moves inward.

The final rule remains compact: locate the state, name the mechanism, choose one next action, verify movement, and let the map update as the learner changes.