Drawdown measures the decline from a previous equity peak to a later low. It matters because a strategy can be profitable over a long period while still experiencing a decline that is financially or psychologically unacceptable. Looking only at final profit hides that path.
Absolute and percentage drawdown
If an account rises to $12,000 and later falls to $10,800 before making a new high, the peak-to-trough drawdown is $1,200 or 10%. Percentage drawdown is useful for comparing different account sizes, while currency drawdown shows the actual money decline.
Maximum drawdown is the largest observed peak-to-trough decline in the test or live record. It is a historical statistic, not a guaranteed worst case. Future drawdown can be larger.
Recovery is asymmetric
Losses and recoveries are not symmetric because percentages are applied to a changing base. A 10% decline requires an 11.1% gain to recover. A 20% decline requires 25%. A 50% decline requires 100%.
| Drawdown | Gain needed to recover |
|---|---|
| 10% | 11.1% |
| 20% | 25% |
| 30% | 42.9% |
| 40% | 66.7% |
| 50% | 100% |
This is why risk control becomes increasingly important as drawdown deepens.
Win rate does not determine drawdown by itself
A strategy with a high win rate can still suffer a severe drawdown if losses are much larger than wins. A lower-win-rate strategy can be viable if winners are sufficiently larger and losses are controlled. Drawdown depends on the full outcome distribution, position size and the order in which trades occur.
That sequence risk is important. Ten losses spread across a year may feel manageable; ten losses in a row can create a much deeper peak-to-trough decline and can cause traders to abandon a strategy at the worst time.
Estimate losing-streak risk
Historical losing streaks provide context, but they are not hard limits. If a backtest's worst streak is six losses, a live streak of seven or eight is not automatically evidence that the system is broken. The correct question is whether the new behavior remains plausible under the strategy's expected distribution and whether market conditions have changed materially.
Position size is the first drawdown lever
Reducing risk per trade reduces the speed at which a losing streak damages equity. If each loss risks 2% of current equity, a sequence of losses compounds much faster than if each risks 0.5%. There is no universal ideal risk percentage, but larger sizing always increases the dispersion of possible outcomes.
Portfolio concentration can hide risk
Several positions may appear diversified because they use different symbols, yet still respond to the same driver. Long EUR/USD and long GBP/USD can both express USD weakness. Gold and Bitcoin may react together during some risk regimes. A portfolio limit can therefore be more informative than simply limiting each trade individually.
Daily loss caps and pause rules
A daily cap can prevent a malfunctioning or regime-mismatched system from continuing to trade aggressively. A loss-streak pause can serve a similar purpose. However, these rules should be tested because an overly restrictive cap may cut off normal recovery behavior.
GodzillaBTC's EA development includes daily-loss, consecutive-loss, concurrent-position and portfolio-risk controls. They are safeguards, not guarantees against drawdown.
Drawdown in backtests
Backtest drawdown is only as credible as the data and assumptions behind it. If spread and slippage are ignored, the reported drawdown can be too optimistic. If parameters were selected after repeatedly examining the same history, the model may be overfit and the apparent drawdown understated.
Use out-of-sample and walk-forward testing to reduce this risk.
Recovery planning
A drawdown plan should answer in advance: at what level will risk be reduced? When will the strategy be paused for review? What evidence would justify resuming normal size? Without predefined rules, decisions are more likely to be driven by fear after losses or overconfidence after wins.
Common mistakes
- Judging a strategy only by total return.
- Treating historical maximum drawdown as a guaranteed ceiling.
- Increasing size to recover losses faster.
- Ignoring correlated positions.
- Using unrealistically low costs in backtests.
- Changing strategy rules during a normal losing streak without evidence.
For example, a trader might reduce size after a drawdown exceeds the range normally observed in validation, pause new entries if data quality is degraded, and require a documented review before restoring normal size. The important point is that the response is defined before stress occurs.
There is no universal drawdown threshold that applies to every system. A useful policy distinguishes between normal statistical variation and evidence that the model or market regime has changed. Risk reduction can be tied to predefined thresholds, but the thresholds should be chosen from historical and out-of-sample evidence rather than emotion after a losing streak.
When should risk be reduced?
A single historical equity curve is only one ordering of trade outcomes. If the strategy's winners and losers arrived in a different sequence, the maximum drawdown could be larger or smaller even with the same average expectancy. This is why serious strategy evaluation often uses resampling or Monte Carlo-style analysis to explore many possible orderings of observed outcomes. The purpose is not to predict the exact future path; it is to understand how sensitive the strategy is to unlucky sequencing.
Sequence risk and Monte Carlo thinking
Sources and further reading
The CFTC forex advisory explains how leverage can amplify losses. For GodzillaBTC's current evidence standards, see Validation & Performance Evidence.