Monte Carlo simulation for drawdown expectations

Most backtest reports give you a single max drawdown number. That is a dangerous way to look at risk. It assumes your trade history is a fixed sequence, but in reality, the order of your wins and losses could have been vastly different. If your worst losses happened to cluster together in your real-world future, that single number would be meaningless.

To get a better sense of reality, try running a Monte Carlo resampling of your trade list. Instead of testing the trades in their original order, you randomly shuffle the list and calculate the maximum drawdown for each permutation. Repeat this process a few thousand times. This generates a distribution of possible drawdowns rather than a single point estimate.

Here is why this matters:

• The original backtest order is just one possible realization.
• Shuffling reveals the 'luck' component in your sequence of returns.
• You can see the 95th percentile outcome, which is a much more conservative stress test.

If your backtest shows a 10% drawdown but your Monte Carlo simulation shows a 25% drawdown at the 95th percentile, you are likely underestimating your tail risk. Tools like QuantConnect allow you to pull your trade logs and run these simulations easily in Python. Most people find that their 'worst case' in the backtest is actually near the median of the simulation.

Have you ever run a permutation test on your strategy to see how much your drawdown profile shifts when the trade order changes?