Read identity and coverage first, reconcile the signed valuation total, separate cash needs, then examine assumptions and limitations. A result is conditional on its scope and scenario, not a forecast or a universal pass certificate.
Read the report identity before the loss
Find the holdings timestamp, valuation timestamp, reporting currency, scenario version, and calculation version. These identify the object being reviewed. A result calculated today from last month's positions is not a current-portfolio stress result. Likewise, two reports bearing the same scenario name may use different shocks or mappings. Treat missing identity fields as substantive limitations rather than administrative details to fill in later.
State the measurement horizon and whether holdings remain fixed. An instantaneous revaluation, a multiday liquidation exercise, and a projected balance-sheet path answer different questions. The Federal Reserve's 2026 scenario publication explicitly distinguishes hypothetical stress conditions from forecasts. That distinction is useful here, while its banking scope should not be confused with the capabilities or regulatory status of an illustrative portfolio tool.
Determine what the total actually covers
Reconcile the input holdings with the modeled holdings. Look for unsupported instruments, missing prices, duplicate positions, stale identifiers, and holdings assigned to broad proxies. Coverage by market value is useful but insufficient for derivatives with small market values and large sensitivities. Inspect notional exposure, risk concentration, and the materiality of excluded positions as additional views rather than assuming one percentage describes completeness.
A missing holding is not a zero-loss holding. It might increase the stress loss or provide an offset, so a partial total is not automatically a lower bound. Keep exclusions visible beside the result and identify why each matters. The Basel Committee's risk-data principles offer an institutional reference for dependable aggregation; the practical reviewer still needs a concrete reconciliation for this particular portfolio.
Hypothetical calculation: expose the denominator
Assume an invented portfolio has USD 10,000,000 of starting market value, of which USD 9,000,000 is modeled. Modeled losing positions contribute minus USD 780,000 and modeled gaining positions contribute plus USD 150,000. Net modeled change is minus USD 630,000. This is a 7% decline relative to the modeled starting value, or 6.3% relative to the entire portfolio's starting value.
The USD 1,000,000 excluded sleeve remains unassessed. Dividing by the full portfolio does not make the calculation complete, and dividing by modeled value does not estimate the omitted sleeve. Label both denominators if both are useful. These hypothetical numbers describe the same partial valuation result from two perspectives, not two competing estimates of what the complete portfolio would lose.
| Hypothetical report item | Value | Interpretation |
|---|---|---|
| Modeled losing positions | -USD 780,000 | Signed valuation contribution |
| Modeled gaining positions | +USD 150,000 | Signed valuation offset |
| Net modeled change | -USD 630,000 | Sum of modeled contributions |
| Modeled-value denominator | USD 9,000,000 | -7.0% modeled change |
| Full-value denominator | USD 10,000,000 | -6.3% partial contribution; 90% value coverage |
Follow the bridge into the headline
Adopt one sign convention and trace every subtotal. Signed profit and loss usually shows losses as negative values, while a loss-magnitude column may show them as positive. Both can work, but mixing them can reverse an offset. Verify that position contributions reconcile to sleeve totals and that sleeves reconcile to the portfolio. Investigate unexplained residuals instead of placing them under miscellaneous without comment.
Then determine whether the headline includes transaction costs, financing costs, accrued income, or assumed management actions. A gross market revaluation should not be compared with an after-liquidation result as though only exposure changed. Preserve a bridge between stages. Separate modeled effects from manual adjustments and identify who supplied each adjustment, why it was needed, and whether it was included consistently across scenarios.
Read cash demands on their own terms
A margin call or redemption outflow belongs to a cash analysis, not automatically to the market-loss total. Review the earliest deadline, the required currency, and the availability of eligible resources. A positive final cash balance can conceal an earlier shortfall. Conversely, a large collateral transfer can restrict liquidity without reducing economic value by the full amount transferred under the modeled accounting treatment.
Ask whether offsetting gains are realizable before obligations fall due. A hedge that improves the end-of-scenario valuation may still require interim financing. Where the report contains no cash-flow model, say that liquidity is unassessed. Do not infer funding resilience from a favorable market-loss figure or assume that an asset classified as liquid can always produce cash by the relevant operational cutoff.
Interrogate apparent precision and comparisons
A narrow range across scenarios can indicate genuine robustness, but it can also result from shared omissions or nearly identical assumptions. Inspect what actually varies. If every scenario preserves the same favorable correlation or excludes the same complex holding, the range says little about those vulnerabilities. Round conclusions to a precision justified by the inputs while retaining calculation precision for reconciliation and testing.
When comparing reports across dates, distinguish holdings changes, market starting-point changes, scenario revisions, model changes, and data corrections. A smaller headline loss does not identify which of those caused the improvement. Request a controlled comparison that holds other dimensions fixed. The report should support an explanation of change rather than inviting a favorable narrative based solely on the direction of one number.
End with a bounded interpretation
A useful summary states the conditional modeled loss, the principal drivers, the coverage boundary, and the most consequential assumption. It also names any cash shortfall separately. Avoid unqualified labels such as safe or passed unless the report defines an approved criterion and shows how it was evaluated. Even a satisfied internal criterion does not establish that untested scenarios or omitted exposures are harmless.
Assign follow-up questions to the evidence that can resolve them: position reconciliation, contract terms, pricing validation, or timing confirmation. Preserve the report and its inputs so a later reviewer can reproduce the interpretation. This approach makes an illustrative stress result useful without presenting it as current investment advice, an observed historical return, or certification under a regulated bank stress-testing program.
Review checkpoints
- Read identity, horizon, and coverage before the headline loss.
- A full-portfolio denominator does not repair partial modeling coverage.
- Reconcile valuation changes and cash obligations separately.
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.
Explain a scenario loss without inventing a unique cause
Reconcile position and factor contributions, expose interactions and ordering choices, and distinguish attribution from causal explanation.
Keep stress-report revisions explainable and reproducible
Assign ownership, preserve report versions, bridge changed results, and prevent corrections or assumption changes from rewriting the record.
First edition, 16 October 2026. Hypothetical values are teaching inputs, not observed portfolio or provider results.