Maximum drawdown: the depth of the worst hole

Maximum drawdown is the largest fall from a peak to a later low. It is the number most people look at first, and for good reason: it is roughly the worst loss you would have had to sit through. But it is one number from one episode. It says nothing about how long the strategy took to recover, or whether it spent most of its life a few percent below its high.

Same depth, different experience

Over the last five years on SPY, Z-Score Mean Reversion and Volatility Breakout both had a maximum drawdown of about 10% (−10.4% and −10.1%). On a leaderboard sorted by drawdown they look alike. They were not:

  • Z-Score Mean Reversion was below its previous high on 32% of days; its longest drawdown lasted 224 days. Ulcer index: 1.2%.
  • Volatility Breakout was below its previous high on 98% of days (including long stretches out of the market while still below an earlier peak); its longest drawdown lasted 1,006 days — almost three years without a new high. Ulcer index: 5.3%.

The first strategy had one sharp dip and recovered. The second spent nearly the whole period slightly underwater. Annual returns followed the same pattern: 8.0% against 2.5%.

The Ulcer index: depth and duration in one number

The Ulcer index is the root mean square of the daily drawdown: every day below the previous peak counts, and deeper days count more. A strategy that falls 10% once and recovers quickly gets a low value; one that hovers 3–5% below its high for years gets a high value, even with a smaller maximum drawdown. That matches how drawdowns feel to the person running the strategy — long stretches without progress are what make people give up. On Quant Arena you find it in the risk tab of every strategy page, next to the longest drawdown in days and the share of time underwater.

Rolling Sharpe: one ratio, many periods

A Sharpe ratio averages return per unit of risk over a whole period, and the period matters a lot. On SPY:

  • Volatility Breakout: Sharpe 2.49 over the last year, 1.07 over three years, 0.47 over five, 0.14 over its full history.
  • Ichimoku Cloud: −0.24 over the last year, but 1.00 over three years and 0.70 over its full history.

Judged on the last year alone, you would pick the first and drop the second — the opposite of what the longer record suggests. The rolling Sharpe chart on each strategy page shows the ratio over a moving one-year window, so you can see whether a good number is typical or a recent streak.

A simple checklist

  • Read maximum drawdown together with the longest drawdown in days and the Ulcer index.
  • Prefer strategies whose Sharpe ratio is reasonable across 1, 3 and 5 years over ones with one spectacular window.
  • Check the 90% bootstrap range for the drawdown: the worst case you live through can be deeper than the one the backtest happened to produce.
  • Ask yourself how long you could watch a strategy go nowhere before switching it off — then compare that with its longest drawdown.