Use historical scenarios to examine documented combinations of factor moves and hypothetical scenarios to explore vulnerabilities outside those episodes. Label adaptations and hybrids explicitly. Apply both to a consistent portfolio snapshot and valuation method so differences in losses reflect scenario design rather than hidden changes in measurement.
Distinguish a replay from an investment history
A historical stress replay applies factor movements from a selected past window to a specified portfolio. It does not imply that the portfolio existed then, held the same quantities, or could have traded at the modeled values. A backtest asks a different question about the behavior of a strategy through time, including its decisions, transactions, and information available at each decision date.
For example, a recently issued bond can be exposed to an older yield-curve change through repricing even though its own price history is short. That is a factor replay, not an observed return for the bond. Keep the distinction in the scenario name and methodology. Calling the resulting loss historical without qualification could make a modeled portfolio outcome look like recorded performance.
Define the observation window before selecting extremes
Choose the start, end, frequency, and market closing convention before extracting historical changes. A common calendar window preserves an observed combination more faithfully than selecting the worst equity day, worst bond month, and worst currency quarter independently. The latter can still be a useful constructed challenge, but it is no longer a single historical episode and should not carry that label.
Specify price versus total-return series, currency treatment, and yield conventions. Check missing holidays and asynchronous closes across regions. An index's published historical composition may differ from its present constituents, while revised economic series can alter the apparent narrative. Preserve the source vintage where possible and explain any substitutions. These decisions can materially change the shocks even before holdings are revalued.
Construct hypothetical conditions for a stated vulnerability
A hypothetical scenario begins with a question that history may not answer cleanly. Perhaps the portfolio combines concentrated technology exposure with foreign-currency borrowing, or holds securities whose contractual features did not exist in an older episode. Specify the joint disturbance needed to examine that exposure. Every factor movement should have a unit, horizon, and documented relationship to the economic story.
The Federal Reserve's 2025 Stress Test Scenarios explicitly distinguishes hypothetical scenarios from forecasts. That distinction applies equally to an educational portfolio exercise. Publishing a constructed path is not a claim that it is likely or imminent. Use official scenarios as examples of clearly defined conditional assumptions, not as investment signals or as automatic calibrations for an unrelated account.
Compare two synthetic scenario vectors
Use a hypothetical USD 500,000 portfolio split between USD 350,000 of equities and USD 150,000 of bonds. In synthetic Scenario A, assign returns of -12% and +3%. Equities lose USD 42,000 while bonds gain USD 4,500, producing a net loss of USD 37,500, or 7.5%. These numbers are invented teaching inputs, not measurements of a named historical crisis.
In synthetic Scenario B, assign -9% to equities and -7% to bonds. The losses become USD 31,500 and USD 10,500, totaling USD 42,000, or 8.4%. Although the equity shock is smaller, the portfolio loss is larger because the bond offset disappears. The comparison shows why a scenario must be evaluated as a joint vector rather than ranked using one dramatic headline variable.
| Synthetic scenario | Equity change USD | Bond change USD | Total change USD | Portfolio return |
|---|---|---|---|---|
| A: equity weakness, bond gain | -42,000 | +4,500 | -37,500 | -7.5% |
| B: both sleeves decline | -31,500 | -10,500 | -42,000 | -8.4% |
Label adaptations instead of hiding them
Historical rates may begin from levels unlike the portfolio's valuation date. Applying the original endpoint can imply a different shock from applying the original change. For equities, an absolute index-point move is not equivalent to a percentage return. Decide which transformation serves the analytical question and retain the untouched source observation alongside the adapted input so the change can be inspected.
A hybrid may retain an observed rate curve move while increasing a sector shock or adding a currency disturbance. Name each replacement and explain its purpose. Avoid presenting an amplified replay as an event that actually happened. The distinction is particularly important when a historical narrative gives a scenario emotional familiarity that its edited numerical components no longer deserve.
Challenge each method's blind spots
CFA Institute's Measuring and Managing Market Risk discusses limitations of both historical and hypothetical scenario measures. Past combinations need not recur, while constructed combinations can misstate magnitudes or co-movements. The practical response is not to declare one method superior in every setting. Examine whether the chosen set covers the portfolio's main vulnerabilities and whether conclusions survive reasonable alternative representations.
A library dominated by one kind of recession can overlook inflationary pressure, concentrated issuer events, or a funding interruption. Conversely, a broad collection of imaginative narratives can share the same underlying shocks and add little analytical coverage. Compare actual factor vectors, not just titles. Include scenarios that challenge familiar offsets and identify conditions under which an apparent hedge would cease to offset losses.
Keep comparisons reproducible
Hold quantities, valuation time, currencies, and pricing models constant when comparing scenario families. If a historical scenario is applied through direct asset returns but a hypothetical one uses factor sensitivities, explain the difference and consider a common-method challenger. Otherwise, an apparent disagreement between scenarios may largely reflect the valuation engine rather than the economic conditions under examination.
Save the original data references, transformations, scenario identifiers, and exclusions with the result. Conclude with conditional statements: which exposure dominates, what changes under the alternative, and which uncertainties remain. Neither historical familiarity nor hypothetical severity supplies an event probability. Together, the methods support a structured examination of portfolio behavior, without offering a trade recommendation or a claim about today's market outlook.
Review checkpoints
- A historical factor replay is not a record of the portfolio's actual performance.
- Adapted historical shocks should be labeled as adapted or hybrid.
- Compare joint factor combinations using consistent holdings and valuation methods.
Source notes
Continue the analysis
How to Define Scenario Severity
Separate shock size, joint plausibility, duration, and portfolio impact when building a transparent severity ladder.
Designing Coherent Portfolio Stress Scenarios
Connect an economic narrative to consistent factor shocks, horizons, and valuation channels without counting the same effect twice.
When diversification changes under stress
Read correlation assumptions, diagnose disappearing offsets, and distinguish a covariance sensitivity from a specified scenario loss.
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.