Survivorship bias
Survivorship bias is a selection distortion that arises when only the subjects that survived to be observed — funds still trading, firms still listed — are analysed, so the results overstate typical success.
See it move
A fund database starts with 200 funds. Five years later, 40 have closed; the 160 survivors report an average return of 9%. But those 40 closed funds averaged −15% before shutting down, so the honest, count-weighted average across all 200 original funds is (160×9% + 40×−15%) ÷ 200 = 4.2%. Reporting 9% instead of 4.2% overstates performance by 4.8 points.
Check yourself
A researcher studies 150 startups founded five years ago. Only the 90 still operating are surveyed, and they report average revenue growth of 20% over the period. The 60 that shut down are excluded from the survey, but company records show they averaged −10% growth before closing. What is the true average revenue growth across all 150 original startups?