What Is Maximum Drawdown? The Largest Peak to Trough Fall, Explained

Maximum drawdown is the largest fall, in percentage terms, from a peak in a portfolio’s value to the lowest point that follows before a new peak is set. It answers one blunt question: if you had held this portfolio through its worst stretch, how much of your money would have been gone at the bottom?

That is why many investors treat it as the most honest risk number of all. Volatility describes how much a portfolio wobbles on average. Maximum drawdown describes the single worst thing that actually happened, expressed in the currency people care about, which is how much poorer they felt at the low point.

What it measures

The calculation is easier to describe in words than in symbols.

Start with a series of portfolio values over time, daily or monthly. Walk forward through the series. At each date, keep track of the highest value the portfolio has reached so far. That running high is sometimes called the high water mark or the running peak.

At every date, the drawdown is the percentage by which the current value sits below that running peak. If the portfolio is at a new high, the drawdown is zero. If it has slipped below its best level, the drawdown is negative and grows deeper as the fall continues.

Maximum drawdown is simply the deepest of all those readings over the period you are measuring.

Three dates matter in any drawdown episode:

• The peak. The last day the portfolio was at a high before the fall began.
• The trough. The lowest point reached before a recovery started.
• The recovery date. The day the portfolio finally regained its old peak. This date may never arrive within the period studied, in which case the portfolio is still underwater.

Plotting drawdown at every date produces what is called an underwater curve: a chart that sits at zero when the portfolio is at a high and dips below whenever it is not. It is one of the most revealing pictures in portfolio analysis because it shows not only how deep the falls were but how long they lasted.

The arithmetic that makes losses asymmetric

Drawdown deserves attention because of a piece of arithmetic that catches people out. Losses and the gains needed to undo them are not symmetric.

Take a hypothetical portfolio worth 100. Suppose it falls to 70. That is a drawdown of 30 percent. To get back to 100 from 70, the portfolio does not need a 30 percent gain. It needs a gain of roughly 43 percent, because the recovery is measured off the smaller base.

Push it further. A fall to 50 is a 50 percent drawdown and needs a 100 percent gain to recover. A fall to 25 is a 75 percent drawdown and needs a 300 percent gain. This is pure arithmetic, not a market claim, and it is why deep drawdowns are treated as qualitatively different from shallow ones rather than just larger versions of the same thing.

A 10 percent fall is an inconvenience. A 50 percent fall is a different problem, because the climb back is twice the size of the drop.

How to read it

A maximum drawdown figure means very little in isolation. A few habits make it useful.

Read it with the period attached. A maximum drawdown measured over three calm years and one measured over a decade containing a severe market fall are not comparable. Always ask what window the number covers and what happened in it.

Check the data frequency. Drawdown computed from month-end values will almost always look shallower than drawdown computed from daily values, because the worst intra-month low never appears in the monthly series. Two providers can quote different drawdowns for the same portfolio purely because one sampled monthly and the other daily.

Compare against a benchmark over the identical window. The useful question is rarely “was the drawdown large” but “was it larger or smaller than the market’s over the same dates, and was the return earned worth that difference”. A relevant benchmark makes that comparison fair.

Look at duration, not just depth. Two portfolios can share a 30 percent maximum drawdown. In one, the fall took two months and the recovery took six. In the other, the fall took a year and the portfolio stayed underwater for four more. The second is far harder to hold, and no depth figure captures that. Time to trough and time to recovery are the companion statistics, and they are the subject of drawdown recovery analysis.

Look at the second and third worst drawdowns too. A portfolio whose worst fall is 30 percent and whose next worst is 8 percent has a different risk shape from one whose three worst falls are all around 28 percent. The first may have been unlucky once. The second falls that far routinely.

Relate it to the return. A drawdown figure paired with a growth rate is the basis of the Calmar ratio, which divides annualised return by maximum drawdown to ask how much pain each unit of return cost.

What it does not tell you

Maximum drawdown is a single observation, and that is its central weakness.

It is one number from one history. The maximum drawdown you observe is the worst thing that happened in the sample you looked at, not the worst thing that can happen. A longer history will usually reveal a deeper one. A short track record almost always understates how bad things can get, simply because there has been less time for a bad event to occur.

It carries no confidence around it. Volatility is estimated from many observations, so it has some statistical stability. A maximum drawdown is drawn from exactly one episode. Change the start date by a few weeks and it can move sharply. It is a fact about a path, not an estimate of a parameter.

It does not say why. A 30 percent fall caused by a broad market decline, a fall caused by one concentrated position collapsing, and a fall caused by leverage unwinding all print the same number. The causes carry entirely different lessons. Reading drawdown without reading position and sector concentration leaves the diagnosis half done.

It says nothing about the return earned. A portfolio that avoided every drawdown by holding cash has an excellent drawdown figure and may have compounded nothing. Drawdown is one half of a trade off, never the whole verdict. That is the point of the broader family of risk adjusted return measures.

It ignores cash flows. The tidy calculation above assumes a fixed pot of money. A real portfolio receiving contributions or paying withdrawals has a value path shaped partly by those flows, so a naive drawdown on the account value can flatter or punish the strategy for reasons that have nothing to do with the strategy.

In a backtest, it can be quietly optimistic. A simulated drawdown computed on a universe that excludes companies which were delisted, or on financial data that has since been restated, will usually be shallower than the drawdown a live investor would have suffered. This is why the discipline behind point in time data and survivorship bias matters before any drawdown figure from a backtest is believed.

It is backward looking by construction. Nothing in the number forecasts the next fall. It describes what a path did, in a particular market regime, over a particular window. Regimes change, and so do the drawdowns they produce.

Used properly, maximum drawdown is less a score than a stress question. It asks whether the worst stretch this portfolio has already lived through is one you could have held without abandoning the plan. That question is worth asking before the next drawdown arrives, not during it.

Related reading

• Portfolio metrics explained: the hub that connects every risk and return measure in one place.
• Volatility and standard deviation explained: the other main risk measure, and how it differs from drawdown.
• Drawdown recovery analysis: time underwater, recovery periods, and the arithmetic of getting back to even.
• What is the Calmar ratio: return measured against maximum drawdown.
• Risk adjusted returns explained: why a raw return figure is never the whole answer.

This article is educational. Altys Labs is not a registered research analyst or investment adviser, and nothing here is investment advice or a recommendation to buy, sell, or hold any security.