Metrics / Max Drawdown
Max Drawdown
4 min read · Metric
The largest peak-to-trough loss in your equity curve. The single best predictor of whether you'll actually keep trading the strategy.
What it is
Max drawdown (MaxDD) is the largest percentage loss measured from any prior peak equity value to the subsequent trough, before the strategy makes a new high. It's the worst-case experience of holding the strategy as a function of when you started — and it's usually the metric that determines whether you turn the strategy off in real life.
It's a path-dependent statistic: two equity curves with identical start and end points can have drastically different MaxDDs depending on the order returns arrived in. A long, slow grind down looks the same on CAGR as a sharp dip and recovery, but the MaxDD reveals the difference.
Formula
# Track running peak, compute drawdown from peak at each point
running_peak[t] = max(equity[0:t+1])
drawdown[t] = (equity[t] - running_peak[t]) / running_peak[t]
max_drawdown = min(drawdown) # most negative value
# Reported as a positive percentage:
max_dd_pct = abs(max_drawdown) * 100Typical ranges
- SPY 1928–present: 86% (1929–1932). Post-1980: 55% (2008–2009).
- Diversified systematic strategies: 10–30%.
- Defensive strategies (low-vol, quality): 5–15%.
- Aggressive single-name momentum: 30–50%.
- Levered strategies: scale linearly with leverage (3x leverage roughly triples the MaxDD).
Common mistakes
- Not noting underwater duration. A 20% MaxDD that recovered in 3 months is not the same trade as a 20% MaxDD that took 4 years to recover. The duration matters because human conviction evaporates around 18 months underwater.
- Confusing MaxDD with worst calendar year. They're different. A strategy can finish a year flat but have had a 30% intra-year drawdown.
- Backtest doesn't span a stress event. A 3-year backtest in a calm regime will show a misleading small MaxDD. Always include 2008 and 2020 if your data goes that far back.
- Treating MaxDD as a hard upper bound. The future is allowed to be worse than the worst historical experience. MaxDD is a sample minimum; the population minimum is somewhere worse.
What the platform flags
Quantis shows MaxDD with the date the trough occurred and the time-to-recovery (or “still underwater” if the strategy hasn't made a new high). The result card pairs MaxDD with Calmar (CAGR/MaxDD) so you see both the worst-case loss and how the return compensated for it.