Two systems: A returns 100% a year with a 50% max drawdown. B returns 30% a year with an 8% max drawdown. Almost everyone buys A. Almost everyone should buy B. Here is why drawdown, not return, is the number that decides your outcome.
What drawdown actually measures
Drawdown is the decline from an equity peak to the following trough, expressed as a percentage. If your account grows to $120,000 and then declines to $102,000 before recovering, you experienced a 15% drawdown. Max drawdown is the worst such decline over the whole history. Our combined system's max drawdown across 11 backtest years is -13.6% at 1% risk per trade.
The recovery math is brutal and asymmetric
| Drawdown | Gain needed to recover |
|---|---|
| -10% | +11% |
| -20% | +25% |
| -33% | +50% |
| -50% | +100% |
| -75% | +300% |
Losses require disproportionate gains to repair. A system that avoids deep drawdowns has a structural compounding advantage over one that swings for the fences, even when the fence-swinger shows higher average returns.
The psychological breaking point
Backtests do not model the most common failure: the human turning the system off. Research on investor behavior shows most people abandon strategies during drawdowns near their pain threshold, converting a temporary decline into a permanent loss. If a system's historical max drawdown is deeper than what you can watch without intervening, you will not survive the system regardless of its long-term math. This is why we let buyers scale risk: at 0.5% per trade the same system runs a roughly -7% historical max drawdown, a very different psychological experience than -13.6%.
How to read drawdown when evaluating any system
- Demand max drawdown alongside returns. Return without drawdown context is marketing.
- Check drawdown duration too: how long from peak back to new highs? Months of underwater time test discipline more than depth does.
- Assume live max drawdown will exceed backtest max drawdown. The worst case is always ahead of you, not behind.
- Compute return over max drawdown. Our backtest runs at roughly 5:1 (67.8% CAGR vs -13.6% DD). Anything above 2:1 is respectable, below 1:1 is a coin flip.
Reading a drawdown in real time: the three questions
Statistics about drawdown are easy to nod along with in a winning month and impossible to remember in a losing one, so here is the operational version: the three questions to ask while a drawdown is actually happening. One: is the depth inside the historical envelope? Our combined backtest's worst was -13.6% at 1% risk; a live drawdown of -6% is therefore normal weather, not a signal. Two: is the system's structure intact? Check win rate over the last 50+ trades against the long-run figure and confirm every trade followed the rules. Structure intact means the drawdown is variance; structure broken means investigate. Three: has anything changed that the backtest could not know? A broker altering spreads, a pair's session behavior shifting. Absent a yes on question three or a break on question two, the correct action during any in-envelope drawdown is precisely nothing, and having pre-answered these questions is what makes nothing psychologically possible.
Underwater time: the dimension everyone forgets
Depth gets all the attention, but duration does the psychological damage. A drawdown has two phases: the decline from peak to trough, and the recovery from trough back to the old peak. The account is underwater for the entire round trip. In our 11-year backtest, the deepest drawdown took a few months to form and then several more to fully recover: close to a year of no new equity highs, in a system with zero losing years. Both facts are true simultaneously, and that combination is what buyers systematically underestimate. You can own a system with a 67.8% CAGR and still spend ten consecutive months looking at an account below its high-water mark. If your commitment to a strategy has a shorter half-life than its historical underwater periods, the strategy will outlive your participation in it, and the exit will happen at the worst possible point. Check the underwater duration statistics of anything you buy, then honestly compare them against your own patience. That comparison predicts your outcome better than any return figure does.