Stress testing takes your current portfolio and subjects it to the market conditions that prevailed during historical crises. The goal is not to predict which crisis will happen next but to understand how your current positions would behave under severe but plausible stress scenarios. If your portfolio cannot survive the replay of a past crisis, it probably cannot survive the next one either.
The standard scenarios worth testing against include the 2008 Global Financial Crisis (broad risk asset sell-off, credit freeze, correlation spike), the March 2020 COVID crash (sudden liquidity crisis with rapid recovery), the 2022 rate hiking cycle (sustained pressure on duration-sensitive assets), and for crypto specifically, the 2022 Luna/FTX period (crypto-specific cascading failure).
For each scenario, you apply the actual asset class returns that occurred during that period to your current portfolio weights. A simple implementation takes the daily returns of each asset class during the stress period and compounds them through your current positions. A more thorough approach also adjusts for the correlation changes that occurred during the crisis, since the correlation structure during stress differs significantly from normal periods.
The output should answer specific questions. What would my total portfolio loss be? How long would the drawdown last based on the historical recovery time? Would any margin calls or liquidation thresholds be triggered? Would the loss exceed my maximum acceptable drawdown? If the answer to the last question is yes, the portfolio needs adjustment now, not after the next crisis arrives.
Reverse stress testing is equally valuable. Instead of asking what would happen under scenario X, you ask what scenario would be required to cause a specific loss level (for example, a 50% portfolio drawdown). This identifies the vulnerable points in your portfolio and reveals which assumptions your current positioning depends on. Sometimes the answer is surprising: a seemingly diversified portfolio can be destroyed by a single factor (like a correlation spike or a liquidity withdrawal) that affects all positions simultaneously.
Scenario design should include not just replays of past events but hypothetical scenarios that have not occurred but are plausible. What if US and China had a severe trade conflict that froze cross-border capital flows? What if a major stablecoin depegged permanently? What if the Fed raised rates by 200 basis points in a single meeting? These hypothetical scenarios are harder to model precisely, but even rough estimates of portfolio impact are more valuable than ignoring the possibility entirely.
The frequency of stress testing matters. Running it quarterly captures portfolio drift from market movements and allocation changes. Running it after significant position changes ensures new trades do not introduce vulnerability that was not present before. Running it when macro conditions shift (yield curve inverts, credit spreads widen, volatility spikes) tests whether the portfolio is positioned for the emerging environment.
One common mistake is treating stress test results as theoretical rather than actionable. If a stress test reveals that a specific scenario would cause a loss beyond your tolerance, the appropriate response is to modify the portfolio to reduce that vulnerability, not to file the report and hope the scenario does not occur. The entire value of stress testing is in the adjustments it motivates.