Reference · Risk & Sizing
Maximum Drawdown
The worst peak-to-trough fall an account has taken — the number that decides whether a strategy is survivable, not just profitable.
Max drawdown = (lowest value after a peak − that peak) ÷ that peak
Maximum drawdown — Measured on the equity curve, from a high-water mark to the lowest point before a new high is made — not per position, and not annualized.
Starter — for Never bought a share
A drawdown is how far your account has fallen from its highest point. The maximum drawdown is the worst such fall it has ever had.
It is measured from your best point
Suppose your account reached $10,000 and later fell to $7,500. That is a drawdown of 25%, because you lost a quarter of what you had. If it then climbs to $12,000 and falls to $9,000, that is a 25% drawdown too — the highs move up with you. The maximum drawdown is simply the deepest one on record.
Why it matters more than the average
Two people can make the same yearly return while one of them loses a third of their money along the way. The first keeps investing. The second often stops, or is forced to stop by a lender, a margin call or their own nerves. The size of the fall decides whether you get to keep playing.
The recovery arithmetic
Losses and gains are not symmetrical. Losing 20% needs a 25% gain to get back to where you were; losing 50% needs a 100% gain. This is why experienced traders care as much about how much they can lose as about how much they can make.
What to take away
- Drawdown is the fall from your highest account value.
- Maximum drawdown is the deepest such fall — the survival test.
- Big losses need disproportionately bigger gains to undo.
Medium — for Invested before, reads the news
Maximum drawdown quantifies path risk: how deep the equity curve went below its high-water mark, how long it stayed there, and what that demanded of the person holding it.
Computing it properly
Walk the equity curve in order, tracking the running maximum. For every point, compute the fall from that running maximum. The largest fall in the series is the maximum drawdown. Two details matter: use the same return frequency throughout (daily and monthly give different answers), and include costs, because friction deepens the trough in real accounts. Drawdown is a peak-to-trough measure on the whole account, not the worst single trade.
Duration is the other half
A drawdown has a depth and a length. Two strategies can both fall 25%, but one recovers in three months and the other stays underwater for three years. The long one is much harder to hold, and it consumes the compounding that would have happened during recovery. Reporting drawdown duration — the number of periods between the peak and the new high-water mark — is what turns a risk statistic into something a person can actually plan around.
Why volatility drags returns
Arithmetic averages overstate compounding when returns swing. A portfolio that gains 50% then loses 50% is not flat; it is down 25%. The bigger the swings relative to the mean, the larger the gap between the average return and the compounded return. Drawdown is the visible form of that gap, which is why risk-adjusted measures such as return divided by drawdown exist.
Calmar and the practical comparison
Divide annual return by maximum drawdown and you get the Calmar ratio — return per unit of worst-case pain. It is a crude measure (sensitive to the sample period, and it ignores duration), but it answers the question a prospective investor actually asks: what do I have to endure to get this return? Use it alongside volatility and the worst single period, never instead of them.
What to take away
- Track the running high-water mark; the deepest fall below it is the maximum drawdown.
- Report duration as well as depth — the same percentage hurts far more over three years.
- Volatility drags compounded returns below the arithmetic average; drawdown is where you see it.
- Calmar (return ÷ max drawdown) is a fast comparability check, not a complete risk measure.
High — for Works with this number already
Maximum drawdown is the risk statistic closest to how capital actually dies: it captures leverage, correlation and liquidity together, which is why survive-the-drawdown is the binding constraint in portfolio construction.
Why realized drawdown exceeds the model
A drawdown computed from historical returns is a sample minimum, and the sample is not the distribution. Three mechanisms make the next one worse than the last: correlations converge to one in a crisis, so diversification that worked in normal times withdraws exactly when it is needed; liquidity evaporates, so exit prices are worse than marks; and leverage is pro-cyclical, with margin requirements tightening as prices fall, forcing sales into weakness. Any sizing rule that assumes the historical maximum drawdown will hold is assuming the crisis will not be a crisis.
From drawdown tolerance to position size
The practical chain runs backwards from the answer. Decide the maximum fall you can withstand without changing behaviour — not the maximum you would tolerate in theory, but the one that does not make you sell. Divide that by the drawdown you believe a single position can produce, and you have the position limit. This is why a stop placed by structure rather than by a fixed percentage matters: a wide, structurally-justified stop shrinks the size the account permits, which is exactly the trade-off being priced.
Sequence risk and the withdrawal case
For anyone withdrawing — retirement, or a fund with redemptions — drawdown is not symmetric in time. The same average return produces a permanently lower terminal value if the losses arrive first while money is being taken out. This is sequence-of-returns risk, and it is why a strategy that looks acceptable on an average-return basis can be ruinous for a decumulating portfolio. The mitigations are lower initial withdrawal rates, a cash buffer sized to the expected drawdown duration, and de-risking as the horizon shortens.
Measuring it honestly
Use net-of-cost, daily returns where available; report depth, duration and time-to-recovery; state the sample and the worst rolling window rather than only the single historical maximum; and stress the number with an explicit crisis scenario (a correlation of one, a gap through the stop, a funding shock) rather than trusting the observed trough. Where leverage is involved, the leverage itself is a drawdown multiplier, so model the forced-deleveraging path, not just the price path.
The point of the number
Maximum drawdown is not a score to minimise at all costs — a portfolio that never draws down is usually one that never earns anything. It is the constraint you design inside: choose the exposure, the stops and the cash buffer so that the worst plausible path does not end your ability to keep compounding or meeting the obligation you have. Everything else in risk management is a detail of that sentence.
What to take away
- Stress the historical trough rather than trusting it: correlations, liquidity and margin all worsen realized drawdowns.
- Derive position size backwards from the drawdown you can hold, not forwards from conviction.
- For withdrawing portfolios the sequence matters — a cash buffer sized to drawdown duration is the mitigation.
- Report depth, duration and time-to-recovery net of costs, on a stated frequency, with an explicit crisis scenario.
Two accounts with the same return and very different risk
Both accounts start at $100,000 and both end the year at $120,000. Account A rises steadily. Account B rises to $150,000 by March and falls to $96,000 by June before recovering.
- Account A: peak $118,000, trough $106,000 → max drawdown = (106,000 − 118,000) ÷ 118,000 = −10.2%
- Account B: peak $150,000, trough $96,000 → max drawdown = (96,000 − 150,000) ÷ 150,000 = −36.0%
- Identical 20% annual returns; B required surviving three and a half times the loss
- Recovery required: A needs +11.4% to regain its peak; B needs +56.3%
The usual mistakes
- Reporting average return without drawdown. A leveraged strategy can show a great average and still be unsurvivable, because the sequence of losses forces the exit before the average is realized.
- Measuring drawdown from the start of the period rather than from the high-water mark, which understates every loss that followed a gain.
- Comparing a drawdown measured monthly with one measured daily. Daily data always finds a deeper trough, so the comparison is meaningless unless the frequency matches.
- Assuming a past maximum drawdown is the worst that can happen. It is the worst that happened, over a specific sample; the distribution has a tail.
Terms this entry defines
Sources
- SEC investor bulletin — margin and forced liquidationU.S. Securities and Exchange Commission, Investor Bulletins on margin accounts
- CFA Institute — risk and return measuresCFA Institute curriculum, portfolio risk measures (drawdown, volatility, risk-adjusted return)
- Magdon-Ismail and Atiya — maximum drawdownMagdon-Ismail, M. and Atiya, A., "Maximum Drawdown", Risk Magazine (2004)
Learn content is for education only — not individualized financial advice, a recommendation, or a solicitation to buy or sell any security. Options involve substantial risk. Examples are simplified and historical patterns never guarantee future results.