Scenario and stress testing ask what happens to a trading account or portfolio when important assumptions become materially worse. The purpose is not to predict the next crisis. It is to expose where the risk process breaks if volatility, liquidity, correlation, prices, margin or execution conditions move outside the comfortable range.
Historical averages are useful, but they describe what was typical in a sample. Stress testing deliberately focuses on the conditions that can cause disproportionate damage.
Start with the exposure that exists now
A stress test should begin with the actual positions, leverage, currencies and concentrations in the portfolio. A dramatic scenario is not useful if it shocks risks the account does not carry while ignoring the factors that actually drive the positions.
The Portfolio Exposure page provides that inventory. Stress testing then asks what happens when those exposures are hit by adverse changes.
Scenario analysis combines several changes into one story
A scenario describes a coherent set of market conditions. For example, an equity selloff might be combined with higher volatility, wider credit spreads, weaker liquidity and a currency move rather than treating each variable as unrelated.
The scenario does not need to be a forecast. It needs to be plausible enough to reveal how the portfolio behaves when several assumptions fail together.
Sensitivity testing changes one input at a time
A sensitivity test asks how the account changes when one variable moves while the others are held constant. Examples include a 1% interest-rate shift, a 10% asset-price move, a wider spread or a higher volatility estimate.
This is useful for identifying which variables matter most, but it can miss interactions. Scenario testing adds those interactions back in.
Historical scenarios replay known periods with caution
Historical stress tests apply market moves observed during a past disruption to the current portfolio. This can provide concrete magnitudes, but today's positions, market structure and correlations may differ from the historical episode.
The past scenario should therefore be treated as one reference point, not as the worst loss that can happen in the future.
Hypothetical scenarios challenge risks the history may not contain
A strategy may have no historical example of the exact combination that threatens it. Hypothetical scenarios can deliberately combine larger volatility, poorer liquidity, adverse currency movement, gaps or changing correlations to test the portfolio's resilience.
The shock should be severe enough to be informative without pretending that one arbitrary extreme number is a forecast.
Reverse stress testing starts from failure
Reverse stress testing asks a different question: “What combination of events would cause this account to breach its drawdown, margin or capital-survival threshold?”
Starting from the failure condition can reveal vulnerabilities that standard scenarios miss. It also connects naturally to risk of ruin, where the important threshold may be operational failure rather than a zero account balance.
Stress liquidity and execution, not just price
A portfolio can survive a theoretical price shock and still fail operationally if spreads widen, market depth disappears or orders fill much worse than expected. Stop-loss assumptions can become especially fragile under these conditions.
The test should therefore include execution and liquidity where they are material to the strategy.
Correlation is a stress variable
A portfolio that looks diversified under historical correlations can become more concentrated if expected offsets weaken. Stress tests should include scenarios where several positions move adversely together.
The Correlation Risk page explains why historical correlation is evidence rather than a permanent relationship.
Leverage and margin can create nonlinear account consequences
A price move can reduce equity, increase effective leverage and bring the account closer to a margin threshold at the same time. That interaction can force position reductions into unfavorable market conditions.
The Leverage & Margin guide should therefore be included when a stress test involves leveraged exposure.
Stress testing should produce an action, not just a loss number
The test is useful when it changes a decision. If a plausible scenario produces an unacceptable loss, the account can reduce size, lower leverage, diversify a concentration, add liquidity, hedge an exposure or define a contingency rule.
A report that says “the portfolio could lose 25%” without stating what would be changed is descriptive rather than operational.
Do not optimize the portfolio to pass one scenario
Any portfolio can be designed to look good against a small set of known shocks. The danger is overfitting risk controls to the test itself.
Use several scenario families and vary the assumptions. The goal is to understand vulnerability, not to engineer a passing score.
Stress testing is different from Monte Carlo simulation
Stress testing applies selected adverse conditions to examine specific vulnerabilities. Monte Carlo simulation generates many possible paths from a statistical model.
They can complement each other, but they answer different questions. The Research cluster owns Monte Carlo methodology. This page owns deliberate scenario-based risk challenge.
Common stress-testing mistakes
- testing only price changes while ignoring liquidity and execution;
- using one historical crisis as a universal worst case;
- assuming correlations remain unchanged during the shock;
- choosing scenarios that do not match the portfolio's actual exposures;
- producing loss numbers without a predefined response;
- optimizing the portfolio to pass one known scenario.
A practical stress-testing workflow
- Map current exposures, leverage and concentration.
- Identify the variables that can materially damage the account.
- Run single-variable sensitivities.
- Build historical and hypothetical multi-factor scenarios.
- Include correlation, liquidity, gap and margin effects where relevant.
- Compare losses with the risk budget and drawdown limits.
- Define the action required if the result is unacceptable.
- Repeat the test as positions and market conditions change.
Stress testing inside the MFXG framework
The parent Risk Management pillar treats stress testing as a check on model comfort. A process that only works when spreads stay narrow, correlations stay stable and volatility stays average is not robust.
Stress testing does not tell you what will happen next. It tells you whether the account can survive when reasonable assumptions turn out to be wrong.