Max drawdown is the largest peak-to-trough fall in an equity curve. Learn to compute it, why losses need bigger gains to recover, and how long it takes.
Risk Metrics · Updated 2026-10-07
Max drawdown (MDD) is the largest percentage fall from a running peak to a later low in an equity curve over a chosen period. It answers a simple question: if you had started at the worst possible moment, how much of your account would you have watched disappear before things turned around? Return figures describe where a strategy ended up; drawdown describes how rough the road was along the way.
Drawdown is always measured against the highest value reached so far, not against the starting capital. A strategy that doubles and then falls back to +40% has not lost money overall, yet it has suffered a 30% drawdown, because 1.4 is 30% below 2.0. That is why MDD is reported as a negative number, and why it often feels worse than the total return suggests.
You need only the sequence of account values, ideally daily. Then work through three steps. First, keep a running peak: at each date, the peak is the highest value seen up to and including that date. Second, compute the drawdown at each date as value ÷ running peak − 1. Third, the max drawdown is the most negative of those drawdowns, and the dates of that peak and of the low that followed it define the episode.
A small example makes it concrete. Suppose an account is valued at 100, 120, 90, 110, 130, 104 and 125 on seven consecutive dates. The running peak is 100, 120, 120, 120, 130, 130, 130. Dividing each value by its running peak and subtracting one gives drawdowns of 0%, 0%, −25%, −8.3%, 0%, −20% and −3.8%. The max drawdown is −25%, from the peak of 120 to the low of 90. Notice that the later −20% dip, from 130 to 104, is a separate and smaller episode, and that the curve ends at 125, still 3.8% under its peak.
Measurement frequency matters. A curve sampled monthly can hide a low that happened inside the month, so a daily MDD is usually deeper than a monthly one for the same strategy. When you compare two numbers, check that they come from the same frequency and the same period.
The gain needed to get back to the old peak is 1 ÷ (1 − drawdown) − 1. Because the gain is earned on a smaller base, it is always larger than the loss. A −10% drawdown needs +11.1% to recover. A −20% drawdown needs exactly +25%. A −35% drawdown needs +53.8%. A −50% drawdown needs +100%, a doubling just to get back to where you were. In the small example above, climbing from 90 back to 120 required +33.3%, not +25%.
The gap widens quickly past −30%, which is why many systematic investors treat depth as a hard constraint rather than a cosmetic statistic. The drawdown recovery calculator applies the same formula and also estimates how many years recovery takes at an assumed annual return; for instance, a 30% drawdown needs +42.86%, which takes about 2.55 years at 15% a year.
Depth is how far the curve fell. Duration is how long it stayed under water, measured from the peak to the date the old high was regained. The two are different risks. A sharp fall that recovers in weeks is psychologically and practically different from a shallow slide that keeps capital below its high for two years. It helps to split duration into the decline phase, from peak to trough, and the recovery phase, from trough back to a new high.
The strategy backtests on this site re-run eight US ETF and stock strategies on roughly ten years of daily data (24 October 2016 to 5 October 2026 for the two below, analysis run on 7 October 2026). Two of them show how differently drawdowns can behave.
Kaufman Efficiency Index has the deepest max drawdown in the group, −48.6% on the daily curve. It peaked on 26 January 2018, fell for 565 days to a trough on 14 August 2019, and regained the old high on 19 February 2020, 754 days (about 25 months) after the peak. Over the same peak-to-trough stretch SPY returned +1.9%, so this was not a market-wide sell-off: the loss belonged to the strategy. By the recovery formula, a −48.6% fall needs +94.6% to get back to the peak. Its next two drawdowns were −29.0% (July 2024 to April 2025, recovered in August 2025) and −28.3% (February to June 2020, recovered in August 2020).
Growth-Value Style Selection shows a different shape. Its deepest drawdown on the daily curve was −35.0%, from a peak on 26 January 2018 to a trough on 24 April 2018, only 88 days of decline. Recovery took far longer than the fall: the old high came back on 8 January 2020, 712 days after the peak. SPY returned −7.9% over the decline. A −35.0% drop needs +53.8% to recover. Growth-Value Style Selection is an aggressive-tier strategy: it can hold 3x leveraged ETFs (SPXL, TQQQ, UPRO), it was selected from many backtested candidates so luck cannot be ruled out, and it has no live or forward record yet.
The comparison also shows why duration deserves its own look. Kaufman's deepest drawdown was the deeper of the two, but its climb from trough back to the high took 189 days, while Growth-Value Style Selection needed 624 days to recover a smaller loss. Its second-deepest drawdown, −29.9%, ran from 29 December 2021 to a recovery on 13 June 2023, 531 days in all. Neither depth nor duration alone tells the whole story.
Treating MDD as the worst case. It is the worst case that happened in this sample. A longer history or a different market can produce a deeper one, and a ten-year backtest has lived through only a handful of real stress periods.
Comparing across different windows or frequencies. A bot with three weeks of history has had little chance to experience a large drawdown; a strategy with ten years has had many. Short records tend to flatter MDD.
Ignoring what the market did. A drawdown during a broad sell-off and a drawdown while the index was flat say different things about a strategy. Kaufman's −48.6% came while SPY was up 1.9%, which points to a strategy-specific weakness rather than a bad market.
Looking only at MDD. Pair it with return to judge whether the pain was paid for; the Calmar ratio divides CAGR by the size of the max drawdown for that reason. Kaufman's is 25.09 ÷ 48.62 ≈ 0.52 and Growth-Value Style Selection's is 29.54 ÷ 34.97 ≈ 0.84. Also look at the second and third drawdowns, which show whether the worst one was an outlier.
Max drawdown is backward-looking. Backtests are hypothetical, rely on modelled costs and fills, and do not guarantee that future drawdowns will be as shallow or as short. Leverage, concentrated positions and changing market conditions can all produce losses larger than any historical figure. Use MDD to size positions and set expectations, not as a promise. Nothing here is investment advice.
Educational content, not investment advice. Guide content and league rankings are independent of any exchange or sponsor relationship.