Every trading record has a number that gets quoted and a number that gets lived. The quoted one is the return. The lived one is the drawdown: how far the account fell from its best point before it recovered, and how long that took. You can read about returns. You feel drawdowns.
The definition
A drawdown is the decline from a peak in account value to the lowest point that follows it, before a new peak is made. It is usually stated as a percentage of the peak. If an account grows to 10,000, falls to 9,370, and later climbs past 10,000 again, that episode was a 6.3% drawdown. The maximum drawdown of a record is the worst such episode over the whole window.
Three things about the definition matter in practice:
- It is measured from the peak, not from where you started. A strategy can be up for the year and still be in a painful drawdown.
- It has a length as well as a depth. Six months under water at the same depth is a different experience from six days.
- It is a property of a window. Shorten the window and the drawdown shrinks; that is why a record's full window matters more than its best stretch.
Why it matters more than the return
Returns compound only if you are still following the strategy when they arrive. Most people who stop following a system do not stop because it lost money over a long period; they stop in the middle of a drawdown, because the drawdown exceeded what they had imagined, and they imagined it from the return.
So the practical question a drawdown answers is: would I have kept going? A strategy with a modest return and a shallow drawdown often does more for a real person than a strategy with a higher return and a drawdown they would have abandoned at the bottom. The second strategy's return was never available to them; only the drawdown was.
What a drawdown looks like in a labeled record
CopyOrders publishes its record with every figure labeled backtest, paper or live, and the labels are never blended. On the six-sleeve backtest, hypothetical and over the full window including the down year, the numbers read: 9.5% compound annual growth, a maximum drawdown of −6.3%, and −2.2% for 2022, the worst calendar year in the window.
Read those three together, not separately. The return is only meaningful next to the drawdown it cost, and both are only meaningful with the label attached: a backtest shows how the rules would have behaved, not what anyone earned. The paper and live records are shorter and separate, and their drawdowns will be their own.
How to read someone else's drawdown
When a trading page shows a drawdown figure, ask:
- Over what window? A drawdown quoted over the last good year is not a drawdown; it is a selection. Ask for the full window, including the worst year.
- Measured how often? Daily closes miss what happened inside the day. A monthly figure hides most of it.
- With what label? Backtested, paper-traded, or live? A backtest's drawdown is a lower bound on what reality would have done, because the backtest knew the rules in advance and never slipped on a fill.
- Is it shown at all? A record that quotes a return without a drawdown has told you half of a sentence.
Drawdown and position size
Drawdown is also where risk per trade becomes concrete. If a system risks 1% of the account per trade and has a historical run of eight losses in a row, that streak is roughly an eight-point drawdown before anything else goes wrong. Risk three times as much per trade and the same streak becomes a drawdown near a quarter of the account. The rules did not change; the size did. That is why the desk's sizing is a rule, 1% per trade from the distance to the invalidation, rather than a feeling.
The honest summary
Drawdown is not a flaw in a strategy; every strategy has one. It is the price of the return, paid in advance and in discomfort. A record that shows it, labeled, over the full window, has told you what the strategy would have asked of you. A record that hides it has told you what it wants you to feel.
Nothing here is personalised investment advice. Backtested figures are hypothetical, do not represent actual trading, and have inherent limitations; past performance of any kind does not predict future results.