Maximum drawdown and investment risk

Drawdown measures loss from a previously reached peak

Finishing with a profit does not mean the journey avoided large losses. At each observation, drawdown compares current value with the highest value already reached. Maximum drawdown, MDD, is the worst decline from a prior peak to a subsequent trough, conventionally shown as a positive loss magnitude.

The peak must precede the trough. Pairing the lowest value with a high reached only later is invalid. Drawdown also differs from profit relative to the initial investment: an account may remain above its starting value while falling sharply from an interim peak.

Follow the peak and trough through six observations

Consider equally spaced values 100, 120, 90, 80, 110 and 125. The second observation sets a peak of 120. Value 90 is a 25% drawdown; value 80 is 1 − 80/120 = 33.33% below that peak. At 110 the old peak is still unrecovered; 125 finally exceeds it.

Maximum drawdown is therefore 33.33%, not the 20% fall from the initial 100 to 80. The final 25% gain over the starting value does not erase that interim loss experience. Assess the ending wealth and the path separately.

Recovery magnitude and waiting time are different

Returning from 80 to 120 requires a 50% gain: 40 is half of 80. A 33.33% gain after a 33.33% loss does not restore the peak because the denominators differ. Deeper losses require disproportionately larger recovery gains.

Here the decline from 120 to 80 spans two observation intervals. From the 120 peak to the first recovery at 125 spans four intervals. No daily or monthly frequency is assigned, so four intervals must not be called four days or months. If recovery is absent before the sample ends, report it as unrecovered rather than inventing a date.

Change the trough to see why the path matters

The control changes the fourth value while keeping the others fixed. Raising 80 to 110 makes the third value, 90, the worst trough and reduces MDD to 25%. Lowering the fourth value to 50 produces 1 − 50/120 = 58.33%. The ending value remains 125 despite very different interim losses.

Watch the peak, worst interval and recovery marker together rather than reading one percentage alone. This helps discuss loss tolerance and the risk of a forced sale during a funding need. It does not forecast the next trough.

Historical MDD is not a complete risk measure

The model needs positive historical peaks and a comparable wealth series. Sampling frequency can miss intraday troughs, and changing the sample period can change MDD. Deposits and withdrawals are not returns; raw account values may misclassify a withdrawal as a loss. Align cash-flow treatment and total-return conventions before comparing paths.

MDD describes the worst observed decline, not a limit on future losses. It does not directly measure volatility, liquidity, margin calls or tail probabilities. Strategies with equal MDD may have different recovery times and funding needs, so other risk information remains necessary.

Scope and limitations

Use 2–1,000 total-return values (first positive, subsequent nonnegative), adjusted for external deposits/withdrawals. No frequency is assumed and no annualization is applied; periods are observation intervals, not days. Historical MDD does not bound future loss.