Maximum Drawdown
Category
Risiko- & Money-Management
Sub-category
Drawdown-Management
Curated by
Last reviewed
The maximum drawdown is the largest percentage decline of an equity curve from a peak to the subsequent trough before a new high is reached. If an account falls from $112,000 to $103,600, the maximum drawdown is 7.5 %. The metric measures a strategy's historically worst loss scenario – central for prop-firm traders because loss limits enforce a maximum permitted drawdown.
Context & Mechanics
Definition and calculation
The maximum drawdown (MDD) measures the largest decline of an equity curve from a peak to the subsequent trough before a new high is reached. Formula: (peak − trough) ÷ peak. If an account falls from $112,000 to $103,600, the MDD is $8,400 or 7.5 %. Unlike the running drawdown, which describes any current decline, the MDD is the worst-case measure over the entire observation period.
Informative value and limits
The MDD shows which losing phase a strategy has historically gone through – and thus what psychological and financial strain a trader had to endure. Two limits matter: first, the MDD is sample-dependent; the worst drawdown statistically still lies in the future, so backtesting values only mark a lower bound. Second, it says nothing about duration: a shallow, months-long decline can be more taxing than a fast crash. Recovery maths also belongs to the picture: a 10 % loss requires +11.1 % to recover, 20 % requires +25 %, 50 % already +100 %.
Relevance in prop trading
Prop-firm rules are effectively predefined maximum drawdowns: an overall loss limit of 10 % ends the account exactly when the drawdown reaches that threshold. This implies a hard compatibility check: a strategy with a historical MDD of 15 % structurally does not fit a 10 % limit – regardless of how profitable it is long-term. The usual lever is smaller position sizing, which reduces the MDD proportionally, but also the return.
Why it matters for traders
Knowing a strategy's MDD lets a trader align account size, risk per trade and rulebook instead of leaving limits to chance. The GlanWick simulator calculates the maximum drawdown automatically from the equity curve and checks it against any loss limits – GlanWick is a training and simulation tool and not a prop firm itself.
Execution Example
A trader evaluates 100 trades on a $100,000 account ($1,000 risk per trade). The equity curve rose to $112,000 in the interim and then fell to $103,600 before making a new high.
- Identify the peak: highest point of the equity curve = $112,000.
- Identify the trough: lowest point after the peak before a new high = $103,600.
- Calculate the MDD: ($112,000 − $103,600) ÷ $112,000 = $8,400 ÷ $112,000 = 7.5 %.
- Interpretation: with a 7.5 % historical MDD the strategy would stay under a 10 % overall loss limit – but without a safety margin for a worse future path; halving risk per trade ($500) would have cut the MDD to about 3.75 %.
Execution Risk & Errors
Reading the backtest MDD as a guaranteed ceiling for the future
Confusing maximum drawdown with the running drawdown
Ignoring the duration of losing phases and looking only at depth
Underestimating recovery maths (−50 % requires +100 %)
Deploying a strategy with a higher historical MDD than the loss limit in a challenge
Frequently Asked
What is a good maximum drawdown?
It depends on strategy and rulebook. For prop-firm accounts a rule of thumb applies: the historical MDD should sit clearly below the overall loss limit to keep a safety margin for worse paths.
How does the maximum drawdown differ from the drawdown?
The drawdown describes any current decline from the last high; the maximum drawdown is the largest of these declines over the entire observation period – the historical worst-case measure.
Can I reduce the maximum drawdown?
The most effective lever is smaller risk per trade: halving position sizing approximately halves the MDD – but also the absolute return.
Why is the backtest MDD not enough for planning?
Because it is sample-dependent: a strategy's worst drawdown statistically still lies in the future. Backtest values mark a lower bound, not a guarantee.