What maximum drawdown actually is
Maximum drawdown is the largest peak-to-trough decline in your equity curve, measured as a percentage of the peak. Your account runs up to a high water mark, then falls, and the deepest distance it travels below that mark before reclaiming it is a drawdown. The worst one over the whole record is your maximum drawdown.
Volatility gets most of the attention, but volatility is symmetric: it counts upside and downside the same way. Drawdown only counts the losses, and only the ones that matter for survival. A strategy can have modest volatility and still bury you, because what ends a trading career is rarely a single bad day. It is a slow, grinding descent that erodes both capital and conviction at the same time.
That is why drawdown is the number to respect. It answers the question that the Sharpe ratio quietly ignores: "If I had the worst possible timing, how much pain would this strategy have actually put me through, and would I have stayed in the seat?"
How to calculate maximum drawdown
The calculation is mechanical. Walk the equity curve from left to right. Keep a running peak: the highest equity value seen so far. At every point, compute the drawdown relative to that peak. The maximum drawdown is simply the largest of those values.
Maximum drawdown algorithm
running_peak = max(equity[0..t])
drawdown[t] = (running_peak − equity[t]) ÷ running_peak
max_drawdown = max(drawdown[t]) over all t
Work it through a concrete equity curve. Suppose the account moves through these month-end balances:
Equity / running peak / drawdown
$100,000 / $100,000 / 0%
$118,000 / $118,000 / 0% (new peak)
$110,000 / $118,000 / 6.8%
$95,000 / $118,000 / 19.5%
$82,600 / $118,000 / 30.0% ← trough
$104,000 / $118,000 / 11.9%
$121,000 / $121,000 / 0% (recovered, new peak)
Max drawdown = 30.0%
Note that the peak is the $118,000 high, not the starting balance. Drawdown is always measured from the most recent high water mark, which is why a strategy that is up on the year can still be deep in a drawdown if it has given back gains.
The recovery math that ruins people
The part that surprises new traders and quietly destroys experienced ones is that the gain needed to recover from a drawdown is always larger than the drawdown itself, and the gap widens fast as losses deepen. You lose against the high; you recover against the low.
Drawdown → gain required to break even
10% loss → +11.1% to recover
20% loss → +25.0%
25% loss → +33.3%
33% loss → +49.3%
50% loss → +100.0%
75% loss → +300.0%
90% loss → +900.0%
Recovery = (1 ÷ (1 − drawdown)) − 1
A 50% drawdown is not "halfway to broke." It is a hole that requires a doubling of capital just to get even. If your strategy earns a healthy 20% a year, climbing out of a 50% drawdown takes roughly four years of perfect compounding, and that is assuming the edge survived whatever caused the loss in the first place. Shallow drawdowns are recoverable nuisances. Deep ones are career-enders dressed up as math.
Drawdown duration is its own risk, separate from depth
Two strategies can both have a 25% maximum drawdown and feel completely different. One falls 25% and recovers in three months. The other falls 25% and stays underwater for fourteen months before reclaiming the high. The depth is identical; the experience is not.
Time underwater is where strategies die, not at the trough. The trough is a single frightening moment. Fourteen months below your high water mark is a slow erosion of belief: fourteen monthly statements that confirm you are still losing, fourteen chances to override the system, abandon the rules, or revenge-trade your way to a worse outcome. Most strategies are not killed by their drawdown. They are killed by the trader who could not endure the duration of it.
So measure both. Maximum drawdown depth tells you how much capital is at risk. Maximum drawdown duration, the longest stretch from a peak to the recovery of that peak, tells you how much patience the strategy will demand. Honestly assess whether you have that much patience before you risk a dollar.
Backtest drawdown is the optimistic case
The maximum drawdown in your backtest is the best-case version of reality. Live trading is reliably worse, for reasons the historical simulation cannot fully capture.
- →Slippage and spread: widen exactly when you need to exit, deepening every losing run beyond the modelled fills.
- →Regime change: the future contains market states your sample never saw; the backtest only knows the past it was fit on.
- →Single-path bias: the historical equity curve is one draw from a distribution; the real worst case is almost certainly deeper than the one path you happened to test on.
The professional fix is to stop trusting the single historical path. Use Monte Carlo resampling: shuffle or bootstrap your trade sequence thousands of times to generate a distribution of possible equity curves, then read the drawdown off the 95th percentile rather than the median.
Monte Carlo drawdown (1,000 resamples of the trade sequence)
Historical (single path): 22%
Median resampled drawdown: 26%
95th-percentile drawdown: 38%
Plan for the 38%, not the 22%.
If your backtest drawdown looks suspiciously shallow, treat that as a warning sign rather than a feature. A too-smooth equity curve is often the signature of an overfit strategy that has memorised the noise in your sample and will produce its real, ugly drawdown the moment it meets fresh data.
Is your shallow drawdown real, or overfit? →
A backtest with an unusually clean equity curve and a tiny drawdown is frequently a sign of curve-fitting, not edge. Our Strategy Overfitting Score pressure-tests your results so you can tell a durable strategy from a fragile one before you fund it.
Designing for a survivable drawdown
The most important reframe in risk management is this: size your strategy to the drawdown you can stomach, not the return you want. Return is an output. Drawdown is the input you actually control, through position size and exposure.
Start from your personal limit. If a 20% drawdown is the point where you would genuinely lose faith and intervene, then your strategy's expected drawdown must comfortably fit inside 20% after you account for the live-versus-backtest gap. If the 95th-percentile drawdown is 38%, you are running roughly twice the risk you can tolerate, and you should cut size until it fits.
- →Volatility targeting: scale position size inversely to recent volatility so risk stays roughly constant instead of ballooning in turbulent regimes.
- →Circuit breakers: pre-defined de-risking levels that automatically cut exposure as the drawdown deepens, rather than relying on a decision in the moment.
- →Position-size headroom: run smaller than the math allows so that a worse-than-expected drawdown is uncomfortable, not terminal.
The trader who survives is rarely the one with the best backtest. It is the one who sized small enough that the inevitable drawdown was survivable, and was still in the market when the edge resumed paying.
A practical drawdown policy
Decisions made while staring at a falling equity curve are bad decisions. The fix is to write the rules in advance, when you are calm, and let them execute themselves. A workable policy has three tiers.
Example three-tier drawdown policy (20% tolerance)
Review trigger @ −10%: log it, check for broken assumptions or data issues. No size change yet.
De-risk @ −15%: cut position size by half. Keep trading, smaller.
Hard pause @ −20%: system stops. Full review before any re-deployment.
The review triggers matter as much as the hard limit. A drawdown that reaches your first trigger is your cue to ask whether anything has actually changed: a data feed gone stale, a correlation that broke, a regime the strategy was never built for. Sometimes the answer is "nothing is wrong, this is a normal drawdown, stay the course." Sometimes it is "the edge is gone." The policy does not make that judgment for you, but it forces you to make it deliberately rather than emotionally, and it caps the damage while you decide.
Summary
- Maximum drawdown is the deepest peak-to-trough loss; it matters more than volatility because it decides whether you survive
- Calculate it by tracking the running peak and taking the largest (peak − equity) ÷ peak across the curve
- Recovery is asymmetric: a 50% loss needs a 100% gain, a 75% loss needs 300%. Deep drawdowns are career-enders
- Measure duration as well as depth; time underwater is what breaks the trader, not the trough
- Treat backtest drawdown as the optimistic floor; plan for the Monte Carlo 95th percentile, not the single historical path
- Size to the drawdown you can stomach, not the return you want, and pre-commit a tiered drawdown policy that pauses the system
Frequently asked questions
What is maximum drawdown in trading?
Maximum drawdown is the largest peak-to-trough decline in your equity curve over a period, expressed as a percentage of the peak. If your account climbs to $120,000 and then falls to $84,000 before making a new high, your maximum drawdown is 30%. It measures the worst loss an investor would have experienced if they entered at the worst possible time, which is why it matters more than volatility for survival.
How do you calculate maximum drawdown?
Walk through the equity curve point by point. Track the running peak (the highest equity seen so far). At each point, the current drawdown is (running peak − current equity) ÷ running peak. The maximum drawdown is the largest of those values across the whole series. It is the deepest you ever fell below a prior high.
Why does a 50% drawdown require a 100% gain to recover?
Because the gain is measured against the smaller post-loss balance, not the original peak. Lose 50% of $100,000 and you have $50,000. To get back to $100,000 you must double the $50,000, which is a 100% gain. The math is asymmetric: a 25% loss needs a 33% gain, a 75% loss needs a 300% gain. Deep drawdowns compound the recovery burden non-linearly.
Is backtest drawdown the same as live drawdown?
No. Backtest drawdown is almost always the optimistic case. Live trading adds slippage, wider spreads in stress, partial fills, and regime change the model never saw. A sound process treats the backtested maximum drawdown as a floor, not a ceiling, and uses Monte Carlo resampling to estimate the 95th-percentile drawdown rather than relying on the single historical path.
What is a drawdown policy and why do I need one?
A drawdown policy is a pre-committed set of rules that defines what happens at specific loss levels: a soft review trigger (for example at half your expected drawdown), a de-risking level where you cut size, and a hard limit that pauses the system entirely. Writing it before the drawdown removes the worst decision-maker from the loop: a stressed human staring at a falling equity curve.
A shallow backtest drawdown can be a trap
If your equity curve looks too clean to be true, it probably is. Run your strategy through our free Strategy Overfitting Score to pressure-test the drawdown before live trading discovers the real one for you.