Survivorship bias in trading backtests
Survivorship bias appears when a backtest universe only contains assets that still exist today. Strategies look stronger because the losers have already been removed from the sample.
What survivorship bias is
Survivorship bias happens when the test universe includes only assets or strategies that survived until today. Delisted stocks, failed funds, and abandoned strategies are missing, so the average performance is flattered.
Why it distorts results
A strategy backtested on survivors looks more profitable than it really is, because every asset that went to zero has been silently removed. The real-world universe was harder than the test one.
How to reduce it
Use a point-in-time universe that includes delisted assets, or test across a broad basket and discount results that rely on a few surviving names.