A bias in performance analysis that occurs when one only considers assets or strategies that have “survived” until the end of the sample, while ignoring those that dropped out (e.g., failed, delisted, or stopped trading). In backtesting, survivorship bias can lead to overly optimistic results because it excludes the poor performers. For example, if testing a stock strategy on the S&P 500 but only on current constituents, the backtest ignores companies that went bankrupt or were removed from the index in the past – thus overstating returns. In a forex context (where currency pairs generally don’t disappear), this bias can still appear if one only analyzes strategies that are currently successful and ignores those that blew up. Survivorship bias “distorts trading strategies by ignoring failed or delisted assets, leading to inflated returns and underestimated risks.” Mitigating this requires using complete historical data (including failures) or point-in-time datasets.

Back to Robots Glossary Index