What is survivorship bias?
The systematic error of evaluating only surviving entities while ignoring those that failed or were removed. In equity backtests, it inflates apparent performance by silently excluding stocks that went to zero, delisted, or merged.
A backtest run today against the current S&P 500 only sees the 500 companies that survived to be in the index now. The hundreds that delisted, went bankrupt, or were acquired over the test window are invisible. Naively backtested strategies appear better than they would have performed at the time because the dataset has been pre-filtered to remove failures. The correction is to use a point-in-time database that records the index membership and price history of every security at every historical date, including those subsequently delisted.
How invest-like addresses it
The /track-record/ page documents the consensus cohort with locked entry timestamps and includes every stock that entered the cohort, including those subsequently delisted or that went to zero. The working paper at /papers/ documents the survivorship-correction methodology explicitly: the cohort is computed point-in-time, every entry is tracked from its entry date forward including realised failures, and the median return reflects the full failure-inclusive distribution.
The 73.8 percentage-point 5-year alpha on the 7-of-7 cohort is computed including the realised losses on names that subsequently disappointed. The cohort size of 47 reflects the names that passed the screen at entry; the median return reflects all 47 outcomes, not the survivors.
Frequently asked questions
What is survivorship bias?
The systematic error of evaluating only surviving entities while ignoring failures. In equity backtests, it inflates apparent performance by silently excluding delisted, bankrupt, or merged stocks.
Does the invest-like track record have survivorship bias?
No: the cohort is computed point-in-time and includes every stock that entered, including those subsequently delisted. The working paper documents the methodology explicitly.
How can I verify a backtest doesn't have survivorship bias?
Look for explicit documentation of the point-in-time methodology, the inclusion of delisted names in the cohort, and the use of a survivorship-bias-free database (CRSP, FactSet PIT, or equivalent).
Correo semanal gratis
Las acciones más sólidas de la semana, cada lunes.
Cada lunes: las acciones en las que la mayoría de los siete inversores coinciden por primera vez. Un correo, te das de baja cuando quieras.
Investigación, no asesoramiento de inversión.
Educational only. invest-like is not a registered investment adviser; nothing on this page constitutes personalised investment advice.