Resolve each instrument, preserve its contract details, and map it to the relevant factors using defined sensitivity units or a pricing model. Distinguish direct measurements from proxies, retain signed derivative exposure, and reconcile both position values and factor contributions. Unsupported holdings must remain visible rather than becoming zero-risk rows.
Establish instrument identity before assigning exposure
A ticker alone may not identify the correct share class, exchange, currency, maturity, or derivative contract. Begin with stable identifiers and the contractual fields needed for valuation. Record quantity, price, currency, multiplier, maturity, and ownership sign as applicable. Keep the original source row available so a later reviewer can trace a transformed exposure back to what was actually recorded in the account.
Separate missing identifiers from missing sensitivities. The former can prevent reliable instrument recognition; the latter may allow a documented proxy once identity is known. Automatically assigning an unknown security to a broad equity bucket hides the distinction. A mapping process should produce explicit states such as directly modeled, proxy modeled, unresolved identity, and unsupported valuation, with reasons for each exception.
Choose factors that correspond to the valuation model
Equities may use market and residual factors, bonds may use benchmark curves and spreads, and foreign assets require translation inputs. Derivatives may depend on several underlyings, volatility, and time. CFA Institute's multifactor reading distinguishes different types of factor models and their applications. In a stress engine, the important requirement is that factor definitions and instrument sensitivities use the same mathematical convention.
A sensitivity to a percentage return cannot be multiplied by an index-point shock without conversion. A money sensitivity per basis point cannot be treated as modified duration. State the unit of both exposure and shock in the mapping record. Where a full pricing function is used instead of a sensitivity, preserve its market inputs and model version so the result remains reproducible.
Distinguish market value from economic sensitivity
Market value contributes to the starting portfolio balance sheet. Notional describes a contractual reference amount. Sensitivity measures the response to a specified input change. These quantities can be very different. A newly entered derivative can have a value near zero while carrying substantial market exposure, and cash posted as margin can be an asset without representing the derivative's full economic size.
Signed exposures are essential for shorts and hedges. Calculate the derivative contribution using direction, quantity, multiplier, and the relevant valuation relationship. Do not assign a positive exposure merely because the contract count was stored as an absolute value. Also retain gross positions where netting would conceal basis differences or funding obligations. A small net factor exposure does not prove that all risks have canceled.
Reconcile a hypothetical futures mapping
Consider a synthetic portfolio with USD 250,000 of equities, beta 1.1 to a chosen index, and USD 50,000 of cash, including posted margin. It also holds two short index futures with an assumed index level of 2,000 and multiplier of USD 50 per point. Assume the futures initially have zero marked value. Their signed index notional is -USD 200,000, not zero.
For an illustrative 10% index decline, the equity beta model gives a USD 27,500 loss. The futures gain USD 20,000 because a 200-point decline times USD 50 times two short contracts reverses the loss sign. Cash is unchanged before settlement transfers. The net change is -USD 7,500 on USD 300,000 starting net value, or -2.5%, excluding basis, costs, and financing.
| Hypothetical position | Starting marked value USD | Signed index exposure USD | Change at -10% USD |
|---|---|---|---|
| Equities, beta 1.1 | 250,000 | 275,000 | -27,500 |
| Two short futures | 0 | -200,000 | +20,000 |
| Cash including margin | 50,000 | 0 | 0 |
| Portfolio | 300,000 | 75,000 | -7,500 |
Preserve proxy quality and avoid wrapper duplication
A fund can be mapped through disclosed holdings, an index proxy, or estimated factor sensitivities. These methods answer related but different questions. Record which was used, the disclosure or calibration date, and the portion represented. If holdings are expanded into underlying positions, do not also count the entire wrapper value as an additional exposure in the same calculation. Maintain a parent-child reconciliation instead.
A proxy should have a documented reason beyond a similar name. Compare currency, sector mix, duration, leverage, and contractual features where relevant. An unleveraged benchmark may be inappropriate for a leveraged product, while a broad corporate index may poorly represent a concentrated loan fund. Keep the original proxy result and a challenger where uncertainty is material rather than silently changing mappings after viewing losses.
Test conservation, signs, and factor isolation
First reconcile marked holdings, cash, and liabilities to the starting portfolio value. Then apply a zero-shock test under a frozen clock; unexpected value changes require explanation. Isolate one factor at a time and verify direction for representative long and short positions. Finally, compare position-level and factor-level totals. These checks catch accounting and unit mistakes, though they do not independently validate a strategy or scenario.
BCBS 239 addresses accuracy, completeness, and risk-data aggregation in banking. The educational application is to make transformations traceable and exceptions visible. Retain raw quantities, normalized units, mapping decisions, and any manual override. The futures example also shows why settlement must be handled carefully: transferring a recognized futures gain into cash does not create a second economic gain to add to the portfolio.
Treat the mapping as a versioned analytical artifact
Save identifiers, factor definitions, sensitivity dates, pricing-model versions, and unresolved issues with each run. When a mapping changes, distinguish an exposure correction from a genuine change in holdings. A reduced stress loss caused by a repaired multiplier should not be described as improved portfolio resilience. Reproducibility requires preserving the earlier assumptions as well as the corrected representation.
The finished map should let a reviewer answer which factors affect each position, how the response is calculated, and where the representation is weak. Report coverage using both value and economically relevant exposure measures. This supports conditional portfolio analysis without implying that every risk is captured, that a hedge will execute as modeled, or that the worksheet satisfies a regulatory data-aggregation standard.
Review checkpoints
- Instrument identity and contract terms come before factor assignment.
- Market value, notional, and sensitivity are different quantities.
- Reconcile mappings and preserve proxy choices, exceptions, and versions.
Source notes
Continue the analysis
Coverage and Missing Positions in a Stress Test
Measure what a stress calculation includes, expose unsupported holdings, and interpret partial losses without silently treating missing risk as zero.
Finding Concentration Risk Behind Portfolio Labels
Look through funds, aggregate connected exposures, and distinguish capital weights from conditional loss concentrations.
Reading exposures that change as markets move
Recognize where delta-based estimates fail, distinguish payoff diagrams from valuation, and review nonlinear positions under joint shocks.
Challenge the model, not just its arithmetic
Separate implementation tests from economic validation, construct informative challengers, and document unresolved limitations and permitted uses.
First edition, 16 October 2026. Hypothetical values are teaching inputs, not observed portfolio or provider results.