WhitmanTrading

Drawdown: Depth Is Half the Story

A drawdown is the fall from the highest value an account has reached, measured as a percentage of that peak. Depth is the figure everybody quotes and duration is the one that decides whether a method gets abandoned, because most of the time is spent below a prior high.

How it works

A candlestick chart of the site's shared price history. The headline on the chart reads: The distance from the highest point the account reached.
The distance from the highest point the account reached. Illustrative chart - not real market data.

A drawdown measures from a peak. Not from your deposit, not from the start of the year — from the highest value the account has ever reached.

A gently rising stretch of the long price series with an account equity curve beneath it. The headline on the chart reads: It is measured from a peak, not from what you put in.
It is measured from a peak, not from what you put in. Illustrative chart - not real market data.

That distinction matters more than it sounds. An account up 40% that falls to up 20% is in a 14% drawdown while still being well ahead, and it feels exactly like losing money because in every sense that matters, it is.

A calmly advancing stretch of the long price series with an account curve that breaches its limit. The headline on the chart reads: The deepest on this series was 3.76 per cent.
The deepest on this series was 3.76 per cent. Illustrative chart - not real market data.

On this site’s shared 576-bar history, holding the series itself, the deepest drawdown was 3.76% against a net gain of 3.61% across the whole period.

Duration is the half nobody quotes

A flat, quiet stretch of the long price series with a slowly rising equity curve beneath it. The headline on the chart reads: And 95% of all bars sat below some earlier peak.
And 95% of all bars sat below some earlier peak. Illustrative chart - not real market data.

95% of bars were below a prior peak. Being in drawdown is not the exception; it is almost the entire experience of holding anything.

A strongly rising stretch of the long price series with a gradually rising equity curve beneath it. The headline on the chart reads: The longest stretch underwater ran 73 bars.
The longest stretch underwater ran 73 bars. Illustrative chart - not real market data.

The longest unbroken stretch below a peak was 73 bars. On a daily chart that is most of a quarter of never once making a new high.

A choppy, directionless stretch of the long price series. The headline on the chart reads: Duration is the half nobody quotes.
Duration is the half nobody quotes. Illustrative chart - not real market data.

Every published performance figure quotes maximum drawdown and almost none quote time underwater. The depth is what a risk model needs and the duration is what a person actually experiences — and it is the duration, not the depth, that ends most methods, because a shallow decline that lasts long enough gets abandoned as easily as a deep one.

A declining stretch of the long price series. The headline on the chart reads: A 50% fall needs a 100% rise to get back to even.
A 50% fall needs a 100% rise to get back to even. Illustrative chart - not real market data.

The recovery arithmetic is asymmetric. A 10% fall needs 11.1% to recover, a 33% fall needs 50%, and a 50% fall needs 100%. The relationship steepens sharply, which is the entire mathematical case for limiting size.

In practice

A 72-bar candlestick section of the shared price history with an account curve shown with and without fees. The headline on the chart reads: And costs deepen every one of them.
And costs deepen every one of them. Illustrative chart - not real market data.

Costs deepen every drawdown and lengthen every recovery. They come out during the fall and during the climb back, which is why an active method’s drawdowns are worse than its trade record suggests.

A candlestick chart with a volume histogram beneath it, with the volume histogram emphasised. The headline on the chart reads: Thin conditions produce the slowest recoveries.
Thin conditions produce the slowest recoveries. Illustrative chart - not real market data.

Thin volume lengthens recoveries. Smaller ranges mean less available profit per trade against an unchanged cost, so climbing back takes more trades and more time.

A long-horizon candlestick view of the same price series. The headline on the chart reads: On a daily chart 73 bars is most of a quarter.
On a daily chart 73 bars is most of a quarter. Illustrative chart - not real market data.

Bar counts scale with the timeframe. The same 73 bars is a few days of scalping or a full quarter of daily bars, and only one of those is psychologically survivable without a plan.

A candlestick series containing several opening gaps, with the largest opening gap marked. The headline on the chart reads: A gap can create the whole drawdown at one open.
A gap can create the whole drawdown at one open. Illustrative chart - not real market data.

A gap can deliver an entire drawdown in a single print, which is the case no intra-session risk control addresses.

A declining stretch of the long price series, with the entry price and the level at which a stop would trigger drawn as horizontal lines. The headline on the chart reads: A stop caps one trade, never the sequence.
A stop caps one trade, never the sequence. Illustrative chart - not real market data.

A stop limits a single loss and does nothing about ten in a row. Drawdown is a property of the sequence, so the controls that work on it are position size and a limit on how much can be lost in a period.

A candlestick chart of the site's shared price history, annotated with the round-trip cost. The headline on the chart reads: Each trade inside it still costs a share of a bar.
Each trade inside it still costs a share of a bar. Illustrative chart - not real market data.

Trading harder to escape one makes it worse. More trades at 2% of a median bar’s range each is a deeper hole, and the urge to accelerate is strongest exactly when the arithmetic is least forgiving.

Planning for it before it arrives

Write down the drawdown you will accept before you need the number. Depth as a percentage, duration in weeks, and what you will do when either is reached — reduce size, stop and review, or continue unchanged because both are within what the method’s history implies.

The decision has to be made in advance because it cannot be made well during. Judgement at the bottom of a long decline is the worst judgement available, and a rule written when nothing was wrong is the only version of yourself worth listening to then. A method with no stated expected drawdown is a method that has not been thought through, whoever is selling it.

One practical measurement is worth adding to any review: the drawdown you are in right now. It takes one line — the highest account value reached, minus the current value, as a percentage of the peak — and it converts a vague sense of how things are going into a number that can be compared with the plan.

Track the running peak rather than recomputing history. A single stored figure updated whenever the account makes a new high is all that is required, and having it visible changes behaviour: a trader who can see 2.1% underwater against a stated 8% tolerance is in a very different position from one who only knows that recent weeks have felt bad.

What a drawdown is not

It is not measured from your deposit. It is measured from the peak.

It is not a loss on a trade. It is a property of a sequence of them.

It is not fixed by a stop. Stops cap trades, not runs of them.

And it is not only about depth. Time underwater is the other half.

When it fails as a measure

A sideways, range-bound candlestick series. The headline on the chart reads: In a range it deepens without any large losing trade.
In a range it deepens without any large losing trade. Illustrative chart - not real market data.

In a range the drawdown deepens with no losing trade large enough to notice. Costs and small losses accumulate, nothing looks wrong in the trade log, and the equity curve is grinding lower.

The second failure is comparing maximums across methods. A 20% drawdown over three weeks and one over three years are not the same risk with the same number attached.

A third is quoting the historical maximum as a limit. It is the worst that has happened so far and not a bound.

A fourth is measuring it on closed trades only. Open positions produce the drawdown you actually live through.

And a fifth is ignoring duration entirely, which is what almost every published figure does.

The original data

On this site’s shared 576-bar history, holding the series itself, the maximum drawdown was 3.76%, the median reading 1.36% and the ninetieth percentile 2.72%. 95% of bars sat below a prior peak, the longest unbroken stretch below one ran 73 bars, and the series finished 3.61% above where it started. The figures are in research/series-measurements.json, produced by site/measure_series.py.

A 72-bar window of the shared price history, cut short at the decision bar. The headline on the chart reads: Down 3% and 40 bars under water. Cut the size?
Down 3% and 40 bars under water. Cut the size? Illustrative chart - not real market data.

A maximum drawdown almost identical to the total gain is the finding worth sitting with. The series made 3.61% and gave back 3.76% at its worst point along the way — which means a holder needed to tolerate a decline larger than everything the position eventually earned. Judge a method by the size of its drawdown relative to its return, not by either number alone, and most performance claims become much easier to read.

Risk management is the framework drawdown control belongs to. Ulcer index combines depth and duration in one figure. And risk per trade is the lever that actually changes it.

What I actually do

The number that changed how I think about this was time, not depth. I can sit through a loss. What nearly made me quit was a stretch of months where nothing was wrong and nothing was working, and I had never once seen anybody quote that figure for a strategy they were selling.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.