Survivorship bias

Drawing conclusions from a sample that excludes the failures.

A study of the returns of companies listed today omits every company that went bankrupt or was delisted, which flatters the average substantially. The same problem affects fund performance data, where closed funds disappear from the record.

For an investor the practical consequence is that historical return estimates drawn from surviving names are too high, and strategy backtests built on a current universe are worse.

Ask what is missing from the sample before believing the average.