Stress testing with explicit assumptions, not forecasts

A stress test asks how an account might behave under specified adverse conditions and assumptions. It does not predict an event’s timing or certain probability, nor necessarily identify the largest possible loss. Its value lies in explicit assumptions and limitations rather than a frightening or deceptively precise final number.

Record the starting account snapshot

Positions, quantities, currencies, collateral and pending orders should refer to a stated time. Disclose stale or incomplete data. A scenario mixing today’s balance with yesterday’s positions may be uninterpretable. Also specify whether the model concerns a master account or an individual investor’s actual outcome.

Define explainable shocks

Price movement, wider spreads, reduced liquidity, higher margin requirements and lost access are different shocks. Examine individual effects before relevant combinations. Severity may come from historical experience or an independent assumption, but its source and rationale must be clear. Do not present a hypothetical number as an observed event.

A hypothetical linear sensitivity

Suppose a fully linear position loses ten reporting-currency units for each adverse unit of price change. A three-unit shock produces a hypothetical loss of thirty if quantity remains fixed and costs are excluded. The relationship does not apply to every derivative or liquidation path. It states neither shock probability nor an account loss ceiling.

The path can change the outcome

A final price does not tell the whole story. Stops may trigger, margin may change or part of a position may close along the way. A static model excluding those responses must disclose the limitation. Assuming immediate, complete execution at a chosen price requires justification in a liquidity-stress scenario.

A scenario sheet

• Snapshot time, account scope and data quality.
• Assumed shock severity, sequence and duration.
• Valuation, cost and currency-conversion rules.
• Assumed order, margin and manager responses.
• Outcomes, model exclusions and review responsibility.

Do not hide fixed correlation or perfect responses

Asset co-movement under pressure can differ from calm periods. Assuming a manager always reacts at the right moment can make the outcome optimistic. Where relevant, examine delayed responses or inability to act. This is not a trading recommendation or an instruction to change positions.

Use results to ask better questions

If a scenario reveals a weakness, record questions about data quality, authority boundaries or host rules. Changing assumptions only to improve the number defeats the exercise. Passing several scenarios does not establish safety under every future condition. Retain discarded scenarios and reasons for selecting the final set.

When reviewing trading accounts, read stress results with historical evidence and model descriptions. Calculation tools can examine some assumptions, but should not be called a complete stress engine without that actual capability. This is an educational framework, not a market forecast, risk guarantee or personal allocation recommendation.