Monte Carlo Resampling of Backtest Trades to Assess Tail Risk
Summary
The article explains how a single backtest equity curve can hide the range of outcomes produced by different orderings of the same trades. It distinguishes Monte Carlo simulation, which shuffles closed-trade profit and loss values to generate alternative equity paths, from analysis, which summarizes those paths with measures such as bust rate, profitable-run rate, and maximum observed drawdown.
A Python workflow reads closed-trade PnL and initial balance from a MetaTrader 5 Strategy Tester HTML report, runs repeated shuffles, and produces a mean equity curve with a 5–95% band alongside numerical summaries. The outputs are proposed as inputs to risk sizing and strategy review, and the article suggests comparing in-sample and out-of-sample reports. This method explores sequence risk under the observed trade outcomes; it does not create new market behavior or guarantee that live results will fall within the simulated distribution. Its conclusions depend on the report data and the resampling assumptions.
Key ideas
- A single backtest order of trades represents only one possible equity path.
- Shuffling observed closed-trade PnL creates alternative sequences for examining equity-path variation.
- Bust rate, profit rate, drawdown, and percentile bands summarize different aspects of simulated risk.
- The resulting distribution can inform risk sizing and comparison of in-sample and out-of-sample behavior.
- Resampling observed trades is a model-based aid and cannot guarantee live outcomes.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.