Every trader wants to talk about returns. The number that actually determines whether a strategy lives or dies is quieter, and almost nobody leads with it. Maximum drawdown is the largest peak-to-trough decline an account or strategy has suffered over a given period. It is not an average, not a typical bad week, not a guess about the future. It is the worst hole the equity curve has ever dug for itself, measured from the top of a peak to the bottom of the subsequent valley.
We treat it as the first metric, not the last. A backtest showing 40% annual returns is meaningless until you know whether the path to those returns ran through a 15% dip or a 65% collapse. The second path is unsurvivable for most accounts and most temperaments, no matter how attractive the endpoint looks. Drawdown is the toll the market charges to deliver a return, and the size of that toll is what decides whether you keep paying it or quit at the bottom.
Maximum drawdown is a peak-to-trough measurement. You track the running high-water mark of an equity curve, then measure how far below that mark the equity falls before a new high is reached. The deepest of all those declines, expressed as a percentage of the peak, is the maximum drawdown.
The formula is simple:
Maximum Drawdown = (Trough Value − Peak Value) ÷ Peak Value
Work through a small example. Say an account starts at $100,000 and climbs to a peak of $120,000. A losing streak then drags it down to $90,000 before it recovers. The drawdown from that peak is ($90,000 − $120,000) ÷ $120,000, or −25%. Notice the drawdown is measured from the $120,000 peak, not from the $100,000 starting balance. Even though the account is only down 10% from where it began, it has surrendered a quarter of its highest value. That 25% is what the account holder feels, and it is what gets reported as the drawdown for that episode. If no later decline cuts deeper, −25% becomes the maximum drawdown for the period.
Two related numbers matter alongside the depth. The drawdown duration is how long the strategy stays underwater before reclaiming its old high, and it is often more punishing psychologically than the depth itself. A 20% drawdown that recovers in two months is an inconvenience. A 20% drawdown that grinds on for two years is a test of conviction that most people fail.
Returns compound, but so do losses, and the two are not symmetric. This is the part that trips up newcomers who fixate on the win rate or the annualized figure at the top of a tear sheet. A strategy can post a glittering average return and still be uninvestable because the road to that average passes through a drawdown deep enough to end the account, trigger a margin call, or break the trader's nerve before the recovery arrives.
Average return is a statement about the destination. Maximum drawdown is a statement about the worst part of the journey. You only collect the destination if you survive the journey.
You do not get to keep the average return if the worst-case path takes you to zero first. Survival is a precondition for compounding, not a footnote to it.
There is also a behavioral cost that no spreadsheet captures. The deepest drawdowns tend to arrive after a string of strong months, exactly when capital allocation is highest and confidence is highest. That is when a 50% drawdown does maximum damage, both to the balance and to the discipline required to stay in the seat. The metric matters because it predicts the moment you are most likely to abandon an otherwise sound strategy.
Here is the arithmetic that should be tattooed on the inside of every trader's eyelids. The gain required to recover from a loss is always larger than the loss itself, and the gap widens viciously as the loss deepens. A loss of X% does not require a gain of X% to break even, because you are now earning that gain on a smaller base.
Run the numbers and the asymmetry becomes impossible to ignore:
The relationship is governed by a clean formula: Recovery Gain = 1 ÷ (1 − Drawdown) − 1. A 50% loss leaves you with half your capital, and turning half into a whole requires a 100% return on that half. This is why shallow, controlled drawdowns are worth so much more than they look. The difference between a strategy that bottoms at −15% and one that bottoms at −45% is not three times the pain — it is the difference between needing an 18% recovery and needing an 82% recovery. One is a routine quarter. The other can take years, if it comes at all.
These three terms get used loosely, often as if interchangeable. They are not, and conflating them leads to poor decisions. Each answers a different question about risk.
Volatility, usually measured as the standard deviation of returns, describes how much returns scatter around their average. It treats an upside surprise and a downside surprise as equally "risky," which is why it is an incomplete measure of pain. A strategy can have low volatility and still suffer a catastrophic drawdown if its losses cluster, and a high-volatility strategy can have a modest drawdown if its swings are quick and mean-reverting. Volatility is about dispersion. It does not tell you how deep the hole gets.
Drawdown is path-dependent and asymmetric. It only cares about declines, and only about the worst one. It captures the lived experience of holding a strategy — the actual peak-to-trough loss you would have endured — in a way that a volatility figure never can.
Risk of ruin is a probability: the odds that an account is wiped out, or drawn down past a point of no return, given a strategy's edge, bet size, and the randomness of outcomes. Drawdown tells you how bad it has been; risk of ruin tells you the likelihood it gets bad enough to end the game entirely. The two are linked — a strategy with large drawdowns and aggressive sizing carries a high risk of ruin — but they are distinct lenses, and a disciplined process tracks all three. For a fuller treatment of how these fit together, see our guide to risk management in algorithmic trading.
Once you accept drawdown as a first-class metric, it changes how you read a backtest. Two strategies with identical annual returns are not equal if one bottomed at −12% and the other at −38%. The first is doing the same job with less than a third of the worst-case pain, and on a risk-adjusted basis it is the vastly superior system.
A few practical tests we apply when comparing systems:
The honest version of this question — whether any of it is worth the effort versus simply indexing — is one we take seriously in is algorithmic trading worth it. Drawdown discipline is a large part of what makes the answer "sometimes, if done right."
There is no universal "correct" maximum drawdown. The acceptable figure is the one you can hold through without abandoning the strategy at the worst possible moment, and it depends on your capital, your time horizon, and your tolerance. For most retail accounts, a maximum drawdown beyond roughly 30% to 35% starts to threaten both the math and the psychology — the recovery gains required grow steep, and the emotional pressure to quit grows with them. Many disciplined systems are engineered to keep the worst-case decline well below that.
The crucial insight is that drawdown is not fixed by the strategy alone — it is a function of position sizing, which you control. A strategy with a raw, fully-invested drawdown of 40% can be sized down so the realized drawdown on your capital lands closer to 20%. The trade-off is direct and linear in expectation: halve the position size, and you roughly halve both the drawdown and the return. The art is finding the size at which the drawdown is survivable and the return is still worth pursuing.
Practical levers for controlling drawdown include fixed-fractional position sizing (risking a small, constant percentage of equity per trade), volatility targeting (scaling exposure down when markets get choppy), and hard portfolio-level stops that cut risk when a predefined drawdown threshold is breached. None of these improve the strategy's edge. What they do is convert a raw edge into one you can actually hold onto through the inevitable bad stretch — which, given the recovery math, is the whole game.
Define the worst case before you risk a dollar, size so you can survive it, and the returns have a chance to take care of themselves. Get that order backwards and the best edge in the world will not save you. That risk-first sequencing is the core of how we build everything at Algo Alpha.
There is no single right answer, because it depends on your capital, horizon, and tolerance. As a rough guide, many disciplined systems aim to keep maximum drawdown below 30% to 35%, since the recovery math and the psychological pressure both grow harsh beyond that point. What matters most is whether you can hold the strategy through its worst stretch without abandoning it.
You track the running peak (high-water mark) of an equity curve, then measure the largest percentage decline from any peak to the subsequent trough before a new high is made. The formula is (Trough − Peak) ÷ Peak. The deepest such decline over the period is the maximum drawdown.
Because the gain is earned on a smaller base. A 50% loss leaves you with half your capital, and turning that half back into the whole requires doubling it — a 100% return. The general formula is Recovery Gain = 1 ÷ (1 − Drawdown) − 1, and the required gain accelerates sharply as losses deepen.
No. Volatility measures how much returns scatter around their average and treats upside and downside equally. Maximum drawdown measures only the worst peak-to-trough decline and is path-dependent. A low-volatility strategy can still suffer a large drawdown if its losses cluster together.
Yes — through position sizing. Drawdown scales roughly with exposure, so reducing position size lowers both the drawdown and the expected return in tandem. Volatility targeting and portfolio-level stops are additional levers. They do not improve the edge, but they convert a raw edge into one you can realistically hold through a bad stretch.