Drawdown is the decline in an account from its highest point to its lowest point before a new high is made. It is expressed as a percentage of the peak. An account that reaches £12,000, falls to £10,200 and then recovers has experienced a 15 per cent drawdown. The maximum drawdown is the largest such fall in the account's history, and it is the single most important number in judging whether a strategy can be lived with.
Balance and equity drawdown
The balance is the account after closed trades. The equity is the balance plus the floating profit or loss on open positions. A drawdown measured on balance only counts losses that have been taken; a drawdown measured on equity counts them as they happen. The equity drawdown is the true one, and it is always at least as large as the balance drawdown.
The difference is where grid and martingale systems hide. Their balance curve rises smoothly because losing positions are never closed, while their equity curve shows the growing floating loss. A record that shows balance drawdown alone is showing you the good half.
Depth and duration
Drawdown has two dimensions. The depth is how far the account fell. The duration, sometimes called time under water, is how long it took to make a new high. A strategy with a 15 per cent maximum drawdown that recovered in two months is a different experience from one with the same depth that took fourteen months. Most people can sit through the first. Few sit through the second, and abandoning a strategy in a long drawdown is the most common way a good strategy ends up losing money for its owner.
Where drawdown comes from
Drawdown is produced by the combination of losing streaks, which follow from the win rate, and risk per trade, which multiplies them. Halving the risk per trade roughly halves the drawdown at the cost of roughly halving the return. That trade-off is the whole content of a risk setting. There is no setting that reduces drawdown without reducing return, and any strategy that claims one is hiding the drawdown somewhere.
A strategy risks one per cent per trade at a 42 per cent win rate. Over three years its longest losing run is eleven trades, and during that run some winners were small, so the equity fell 16 per cent from its peak over seven weeks and took a further three months to make a new high. Its maximum drawdown is 16 per cent and its longest time under water is about five months. At two per cent risk the same trades would have produced a drawdown of about 30 per cent. At half a per cent, about 8 per cent, with returns scaled down in proportion.
On an EA's live record, look for the maximum equity drawdown at the risk setting you intend to use, the date it happened, and how long recovery took. Then ask whether the record is long enough to have included a bad period at all. A record with a 4 per cent maximum drawdown over three months has not been tested. A record with a 20 per cent maximum drawdown over three years has, and the honest response is to decide whether you would sit through 20 per cent, or more, since the next drawdown may be deeper than the last.
Some EAs include automatic drawdown controls, reducing position size once the account has fallen a set amount and stopping new trades beyond a limit. These are worth having, and their thresholds should be stated. They do not prevent drawdown. They put a floor under it and slow its approach, which is what makes a strategy survivable when its worst month arrives.