全部文章

Kamla 学院

Drawdown: the number that matters more than return

2 分钟阅读

Two systems returned 30% over the past year. The first was never down more than 6%. The second went through a 45% decline before recovering. Same result? Only on paper.

What drawdown is

Drawdown is the distance between a peak in your account and the trough that followed. The maximum drawdown is the worst of those drops across the whole track record, the day your conviction would have been tested the most.

If your account went from 10,000 to 12,000, fell to 9,000 and recovered, the maximum drawdown was 25% (from 12,000 to 9,000), even if the year ended in the green.

Why it matters more than return

  1. The math is asymmetric. Losing 50% requires gaining 100% just to break even. Big drops destroy more than they appear to.
  2. It's the behavior test. Return is what the system did; drawdown is what YOU would have had to sit through without quitting. Most people abandon good strategies at the bottom of a drawdown, turning temporary losses into permanent ones.
  3. It exposes hidden risk. High returns with tiny drawdowns over long periods are extremely rare. When the return is big and the drawdown is nowhere in the sales material, it's because it wouldn't survive publication.

Questions that are worth money

  • What was this system's maximum drawdown, and when?
  • How long did it take to recover (the "underwater period")?
  • What does the system do during a drawdown: carry on unchanged, reduce exposure, halt new buys?
  • Is there an automatic limit (circuit breaker) that cuts risk if the drop exceeds X%?

A mature system has mechanical answers to these, written before the crisis, not improvised during it.

The sign of maturity

Whoever built a serious system talks about drawdown without being asked, because they designed the system around it. Whoever only talks about return is selling you the pretty half of the chart.

In Kamla's public track record, drawdown appears next to return, on the same chart, because only the two together tell the story. See it at kamla.ai/performance.