Evidence Trading
Backtesting without fooling yourself · Chapter 5 of 6

Monte-Carlo: the same edge, a different path

Advanced6 min read · part of Backtesting without fooling yourself

A backtest gives you one path: your trades in the order they happened. That order is arbitrary in a way people badly underestimate, and the equity curve you are looking at is one sample from a large family of curves the same strategy could equally have produced.

Monte-Carlo makes that concrete by reshuffling the trade sequence many times and looking at the distribution of outcomes. The final result barely changes — the same trades in a different order still sum to the same total — but the worst drawdown changes enormously, and the drawdown is the number that decides whether you can actually trade the thing.

That is the main use. Your historical maximum drawdown is not the strategy’s maximum drawdown; it is the one that happened to occur in the observed ordering. A method showing a fifteen per cent worst drawdown in the backtest can easily show thirty across shuffles, and thirty is the number to size against, because there is nothing special about the order history handed you.

The second use is a sanity check on the equity curve itself. If a strategy’s result rests on a handful of exceptional trades, the shuffles will show it: remove or reorder them and the distribution spreads out dramatically. A curve that stays tidy across thousands of shuffles is telling you the result comes from many trades rather than from a few lucky ones.

The honest limits: reshuffling assumes trades are independent and identically distributed, which is not quite true — losses cluster in regimes, and volatility comes in stretches. So the spread it produces is a lower bound on how bad things can get rather than a worst case. It is still far more informative than the single path, and treating the single path as the truth is the error it exists to correct.

What to take away

  • Reshuffling barely moves the total and moves the drawdown a great deal.
  • Size against the drawdown across shuffles, not the one your history happened to produce.
  • Reshuffling assumes independence, so its spread is a lower bound, not a worst case.

Where it goes wrong

  • Treating the historical maximum drawdown as the strategy’s maximum.
  • Reading the Monte-Carlo spread as a guaranteed worst case.
  • Ignoring that a result resting on three trades will look fine in the total.

This chapter, measured against your own trades

In the app the same chapter ends in your figures rather than an example: how often you did the thing it describes, over your last ninety days. You pick one change to make, and Evidence checks afterwards whether it actually changed — from your journal, arithmetic, no opinion involved. Questions you get wrong come back a week later and again a month after that.

Open the free plan →

Check that it stuck

Answers shown — in the app these are asked before you see them, and the ones you get wrong come back after a week.

What changes most when a strategy’s trade sequence is reshuffled?
The maximum drawdown — The same trades sum to the same total in any order. Where the losses cluster is what moves, and that is the drawdown.
Which drawdown should position size be based on?
The distribution across many shuffles — The observed ordering has no special claim to be the one that repeats.
What is the main limitation of reshuffling?
It assumes trades are independent, while losses actually cluster in regimes — Because clustering is real, the spread it produces understates the tail — a lower bound rather than a worst case.