Quick answer: the Standardised Approach for Counterparty Credit Risk, or SA-CCR, is a regulatory algorithm for turning a derivatives portfolio into an exposure-at-default number. It does not simply add trade notionals. It first asks what the portfolio would cost to replace today, then adds a rule-based estimate of how much exposure could grow before the position can be closed or margined, recognises legally valid netting and collateral within prescribed limits, and finally applies a regulatory scaling factor. The core relationship is EAD = α × (RC + PFE), where RC is replacement cost, PFE is potential future exposure and α is a supervisory scaling factor.

SA-CCR is a useful piece of applied mathematics because it separates a derivatives exposure into two different questions: what is owed now, and what could become owed before the position can be safely replaced?

Page role: what this article owns

Bukit Timah Tutor already has a broader article on counterparty credit risk. That page owns exposure profiles, netting sets, collateral, CVA and wrong-way risk as a general mathematical problem. The XVA article owns valuation adjustments.

This article owns a narrower question: how does the Basel SA-CCR rule set convert a non-modelled derivatives portfolio into a regulatory exposure amount? It is about algorithmic structure, not personalised finance, trading or regulatory advice.

1. Why not use notional amount?

A derivative can have a very large contractual notional and a much smaller economic exposure. A fixed-for-floating interest-rate swap with S$100 million notional does not mean one counterparty has lent S$100 million to the other. The notional is mainly a scale used to calculate cash flows.

At the same time, using only today’s mark-to-market value is too narrow. A swap that is nearly flat today may move strongly in the bank’s favour tomorrow. If the counterparty then defaults, the bank could face a replacement loss.

SA-CCR therefore rejects both extremes:

Notional alone is usually too crude.
Current mark-to-market alone ignores future exposure growth.

The algorithm combines a current component and a future component.

2. The top-level equation

The Basel Committee describes SA-CCR as retaining two core components: replacement cost and potential future exposure. At a high level:

EAD = α × (RC + PFE)

where:

EAD = exposure at default used downstream in regulatory capital calculations;
RC = replacement cost, representing current exposure after recognised collateral and margin terms;
PFE = potential future exposure, a supervisory estimate of how exposure could grow;
α = the Basel supervisory scaling factor, set at 1.4 in the SA-CCR framework.

The most important conceptual point is that RC and PFE measure different states of the world. RC asks what replacement loss exists now. PFE asks how the state could change before risk is neutralised.

3. Replacement cost: the “what if default happened now?” component

Suppose a bank has a netting set of derivatives with a counterparty. Let V represent the current net market value of the transactions from the bank’s perspective, and let C represent recognised collateral. For a simple unmargined illustration, replacement cost behaves like:

RC ≈ max(V − C, 0)

If the portfolio is worth S$8 million to the bank and there is S$5 million of recognised collateral, the current unsecured exposure is about S$3 million. If collateral exceeds the positive market value, replacement cost is not allowed to become negative.

For margined portfolios, the Basel treatment is more detailed because thresholds, minimum transfer amounts, independent collateral and margin-period assumptions matter. The general lesson is unchanged: collateral matters only to the extent that the legal agreement and regulatory rules allow it to reduce exposure.

4. Potential future exposure: the “what could change before close-out?” component

Potential future exposure is built from supervisory add-ons. Each trade contributes risk according to its asset class, effective notional, direction, maturity and other prescribed features. Trades are then aggregated within hedging sets so that some offsetting relationships are recognised without assuming perfect cancellation.

Schematically:

PFE = multiplier × aggregate supervisory add-on

The aggregate add-on is not a forecast from the bank’s proprietary risk model. It is a standardised regulatory construction using supervisory parameters. This makes the method more comparable across banks, but less tailored than a fully modelled exposure simulation.

5. From notional to effective notional

The algorithm does not treat every contractual notional equally. It converts trade-level notionals into effective risk amounts. For interest-rate and credit derivatives, supervisory duration adjusts the notional for the time interval over which the trade is exposed. For options, a supervisory delta represents direction and non-linearity. For other asset classes, prescribed transformations determine the effective amount exposed to the underlying risk factor.

A useful abstraction is:

effective notional = contractual scale × time adjustment × direction/non-linearity adjustment

This is one of the reasons two S$100 million derivatives can generate very different SA-CCR add-ons.

6. Supervisory factors convert exposure scale into risk add-ons

After effective notionals are determined, SA-CCR applies prescribed supervisory factors by asset class and risk category. These factors act like standardised volatility-and-risk calibrations. Interest-rate derivatives, foreign-exchange derivatives, credit derivatives, equities and commodities do not receive the same factor because their potential exposure behaviour differs.

Mathematically, a simplified trade contribution looks like:

add-on contribution ≈ effective notional × supervisory factor × maturity factor

The exact Basel aggregation is more structured than this one-line expression, especially for interest rates, credit, commodities and options. The expression is useful because it exposes the three main levers: scale, prescribed risk intensity and time-to-risk-resolution.

7. Hedging sets: why offsets are neither ignored nor assumed perfect

Imagine one interest-rate swap gains value when five-year rates rise and another loses value under a similar movement. Treating them independently would overstate exposure. Assuming they cancel perfectly could understate exposure because maturities, indices, basis risk and timing can differ.

SA-CCR therefore groups trades into prescribed hedging sets and applies aggregation rules that allow some netting while retaining residual risk. This is an important mathematical compromise:

complete additivity assumes correlation of +1;
complete cancellation assumes correlation of −1 for the relevant risk;
regulatory aggregation tries to sit between those extremes.

Special treatments may apply for basis transactions, volatility transactions and other cases where the primary risk drivers differ from ordinary directional exposure.

8. Maturity factor: time changes potential exposure

A six-day residual exposure window is not the same problem as an uncollateralised multi-year contract. SA-CCR uses a maturity factor to scale potential future exposure according to how long the position can move before exposure is reset, margined or terminated.

For unmargined trades, remaining maturity matters subject to regulatory floors and caps. For margined transactions, the margin period of risk becomes central. That is the period during which the portfolio may continue to change after the last successful margin exchange and before positions can be closed or re-hedged.

The weak link is obvious: the model is only as meaningful as the assumed time needed to resolve the exposure. Operational delays, disputes, illiquid positions and very large netting sets can all make the relevant window longer.

9. The multiplier: collateral can reduce PFE, but not erase it completely

If a netting set is heavily overcollateralised, it would be unreasonable to charge exactly the same PFE as an otherwise identical uncollateralised set. SA-CCR therefore uses a multiplier that can reduce the aggregate add-on when the current portfolio value net of collateral is sufficiently negative.

However, the multiplier has a floor. This prevents the algorithm from concluding that future exposure is literally zero merely because the portfolio is overcollateralised today. Prices can move, collateral can lose value or become unavailable, and close-out is not instantaneous.

This is an example of evidence polarity: collateral is evidence for lower exposure, but it is not evidence that all future exposure has disappeared.

10. A worked conceptual example

Suppose a simplified netting set has:

positive net market value after recognised collateral of S$4 million;
aggregate supervisory add-on of S$6 million;
multiplier equal to 1 because there is no overcollateralisation benefit in this example;
α = 1.4.

Then:

RC = S$4m

PFE = 1 × S$6m = S$6m

EAD = 1.4 × (4 + 6) = S$14m

The S$14 million is not a forecast that the bank will lose S$14 million. It is a regulatory exposure measure used as an input into capital calculations. The distinction between exposure, expected loss and capital is essential.

11. Why legal netting is a mathematical input

A spreadsheet can net two trades in one second. A regulator cannot assume the same economic offset unless the relevant close-out netting agreement is legally enforceable in the applicable jurisdictions and circumstances.

This means the algorithm depends on a non-mathematical fact: which trades are legally allowed to be treated as one netting set? If legal enforceability changes, the mathematics changes because the aggregation boundary changes.

That is a useful systems lesson. Some numbers are outputs of equations; other numbers are outputs of classifications that decide which equations are allowed to see each other.

12. Inputs and outputs

Typical inputs include: trade type, asset class, notional, start and end dates, option terms, current market value, collateral, margin agreement terms, legal netting-set membership, settlement/margin frequency and supervisory parameters.

Typical intermediate outputs include: effective notional, supervisory delta, maturity factor, asset-class add-ons, aggregate add-on, multiplier, RC and PFE.

Final output: EAD for the netting set, which then feeds downstream credit-risk capital calculations and may also matter in other regulatory measures.

13. Assumptions and weak links

Legal enforceability: netting benefits require valid legal support.
Trade classification: wrong product or asset-class mapping changes supervisory factors and aggregation.
Market value quality: bad valuations distort replacement cost.
Collateral records: stale, duplicated or ineligible collateral can understate exposure.
Maturity data: bad dates alter duration and maturity factors.
Margin terms: thresholds, transfer amounts and dispute status affect exposure treatment.
Hedging-set mapping: false offsets can appear if risk factors are grouped incorrectly.
Supervisory calibration: a standardised parameter is necessarily an approximation rather than a bespoke forecast for every portfolio.

14. Failure modes and counterexamples

Same notional, different risk: two S$100 million derivatives can have different maturity, asset class and direction, producing very different add-ons.
Same mark-to-market, different future exposure: two portfolios both worth S$2 million today can have very different PFE.
Apparent collateral, weak legal claim: operational records show collateral but legal or eligibility conditions prevent recognition.
False hedge: two trades have opposite signs but different bases or maturities, so the offset breaks under stress.
Margin illusion: daily margin is assumed to make exposure tiny, but disputes or illiquid close-out materially extend the risk window.
Model equivalence error: a bank’s internal PFE simulation and SA-CCR produce different answers; this does not automatically mean one is wrong because they answer the problem using different calibrations and constraints.

15. Diagnostics and falsifiers

Recalculate the portfolio with collateral set to zero. Does RC and the PFE multiplier respond in the expected direction?
Split one netting set into legally separate sets. Does exposure increase as netting benefits disappear?
Extend maturity while holding other variables constant. Does potential exposure rise where the framework says it should?
Flip a trade’s direction. Does supervisory delta and hedging-set aggregation react correctly?
Move a trade into the wrong asset class deliberately in a test environment. Is the control able to detect the resulting supervisory-factor jump?
Compare trade population counts between the source system and SA-CCR engine. Are any trades missing or duplicated?
Reconcile current market value and collateral totals to independent ledgers.

A useful falsifier for “our SA-CCR exposure is correctly netted” is finding a trade pair that the engine offsets even though the legal agreement does not permit close-out netting between them.

16. Alternatives and limits

SA-CCR is deliberately standardised. It is not intended to reproduce every feature of a bank’s internal exposure simulation. Approved internal-model approaches can model exposure distributions more directly, while stress tests can explore scenarios that prescribed supervisory factors do not fully represent.

The standardised approach has important advantages: comparability, lower modelling discretion and clearer supervisory rules. Its limitation is the mirror image of that advantage: fixed factors and prescribed aggregation cannot perfectly fit every portfolio or market regime.

SA-CCR also does not replace CVA, wrong-way-risk analysis, liquidity stress, collateral optimisation or legal review. Those are connected but distinct jobs.

17. Verification and update triggers

new derivative products or payoff structures;
changes to collateral or margin agreements;
legal opinions affecting close-out netting;
migration between margined and unmargined treatment;
changes in settlement or margin frequency;
large growth in trade count within a netting set;
regulatory changes to supervisory parameters or local implementation;
valuation-system changes affecting current exposure;
repeated reconciliation differences between trading, collateral and regulatory systems.

18. Connections across the finance-algorithms lane

Counterparty credit risk — the broader exposure problem SA-CCR standardises.
Derivatives XVA — valuation adjustments that depend on exposure, funding, credit and margin.
Collateral optimisation — how eligible assets are allocated once collateral requirements exist.
CCP margin — a different algorithmic treatment of current and future derivatives risk.

Research anchors

The deeper mathematical lesson

SA-CCR is a compact example of model design under constraints. It decomposes one difficult quantity into current state and possible future state, uses legal boundaries to decide where netting is allowed, uses prescribed parameters to make different portfolios comparable, and deliberately refuses to let present collateral eliminate all future uncertainty. The chain is:

trade terms → legal netting set → current value/collateral → effective notional → hedging-set aggregation → maturity adjustment → PFE → EAD.

If any link is wrong, the final number can be wrong even when every subsequent equation is calculated perfectly.

Educational boundary: This article explains public Basel mathematics and computational logic. It is not regulatory advice, investment advice, trading advice or a capital calculation for any specific institution.

Small Group Tutorials

How SA-CCR Algorithms Turn Derivatives into Counterparty Exposure: Replacement Cost, PFE, Netting, Collateral and Maturity Factors