Start freeSign in
Research/Glossary/Maximum drawdown

Maximum drawdown

Reference

Maximum drawdown is the largest peak-to-trough decline an equity curve experiences, measured as a fraction of the peak value.

Maximum drawdown is the largest peak-to-trough decline an equity curve experiences, measured as a fraction of the peak value.

An equity curve tracks the cumulative value of a strategy or portfolio over time. As the curve rises, each new high sets a reference peak. If the curve then falls, the drawdown is the size of that decline relative to the most recent peak. Maximum drawdown is the single worst such decline observed across the full history being measured.

The mechanism is straightforward. First identify each running peak in the equity curve. Then measure how far the curve falls from that peak before a new peak is reached. For any point in the decline, drawdown can be expressed as current equity divided by peak equity, minus one. Maximum drawdown is the minimum value reached by that series, or equivalently the largest loss from a peak to a subsequent trough as a share of the peak.

For example, if an equity curve reaches 100, falls to 80, and later recovers, the drawdown from that peak is 20 percent. If no larger peak-to-trough decline occurs anywhere else in the sample, the maximum drawdown is 20 percent.

In Sonar’s backtest overfitting audit tool, drawdown is calculated from the equity curve by comparing equity at each time to the running maximum of equity. This makes maximum drawdown a path-dependent risk metric. Two strategies can end at the same final value while having different maximum drawdowns if one experienced a deeper interim decline.

This metric is useful because it summarizes the worst historical capital contraction visible in the tested path. It complements return-focused and distribution-focused statistics by showing how severe the deepest realized decline was within the sample. As with any backtest statistic, it depends on the observed data path and does not by itself describe outcomes outside that tested history.

Covered in depth in the Strategy research fundamentals pillar hub.

Apply this and the related checks to your own results with the Backtest Overfitting Audit.Open the audit
ShareXLinkedInFacebookEmail