What is survivorship bias?

Survivorship bias, explained

In backtests, it happens when the historical data includes only companies listed today. Firms that went bankrupt, were delisted or were acquired at a loss drop out, so a strategy appears to have avoided losers it would in fact have held.

The same bias shows up in marketing. A tool that highlights only its best past picks, or a vendor that quietly retires underperforming models, presents a survivor's view of its record.

To reduce it, look for backtests that use point-in-time data including delisted stocks, and for track records that list every pick made, with its date, rather than a selection of winners.

Example

A backtest of "buy the cheapest stocks in today's S&P 500" starting in 2005 ignores companies that were in the index in 2005 but later collapsed, so its historical returns are likely overstated.

Related terms

• BacktestingBacktesting is testing a trading or investing rule on historical market data to see how it would have performed had it been followed in the past.
• OverfittingOverfitting is when a model or trading rule is tuned so closely to historical data that it captures noise instead of a lasting pattern, so it performs well in backtests but poorly on new data.
• Benchmark (S&P 500)A benchmark is a standard used to judge investment performance; for US stock tools it is usually the S&P 500, an index of about 500 large US companies weighted by market value.
• AlphaAlpha is the part of an investment's return above or below what its benchmark or risk level would predict, often used as shorthand for beating the market.

Where this matters

Survivorship bias comes up most when choosing AI stock pickers. See Best AI stock pickers for the tools we compare, with dated prices.