Bootstrap Confidence Intervals for Average Trade Results

Summary

This document describes a bootstrap test for judging whether a strategy’s average closed-trade result appears distinguishable from chance. It repeatedly resamples the observed trades with replacement, computes the average for each resample, and uses the resulting distribution to form a 95% interval and estimate the share of averages below zero. Because it does not assume normally shaped returns, the approach can accommodate trade histories in which a few large outcomes have substantial influence.

The example reports a demo account with 44 closed trades whose interval lies above zero; restricting the same history to the last 30 days leaves 21 trades, below the script’s default 30-trade threshold for a verdict. Results depend on the observed sample: with too few trades, resampling cannot represent losses absent from that history and can make uncertainty look too small. The test assesses past trades only, and excludes open positions, deposits, withdrawals, and charges not assigned to trades.

Key ideas

• Bootstrap resampling estimates uncertainty in the average result by repeatedly drawing from observed closed trades with replacement.
• The distribution of resampled averages supplies an interval and an estimate of how often the average falls below zero.
• A minimum trade count can prevent the script from issuing a verdict on a very small sample.
• Resampling cannot reveal losses missing from the observed history, so small samples can understate uncertainty.
• The result describes historical trades and does not establish that an edge will persist.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.