Survivorship bias
The reason almost every statistic you hear about investing is more flattering than the truth.
When you study a group of successes, you are studying the ones that were still there to be studied. The failures left before you arrived, and their absence changes every number you calculate.
Where it appears in investing
- Fund performance. Funds that do badly are closed or merged away. An average of funds that still exist is an average of survivors.
- Index history. Indices drop failing companies and add successful ones. The long-run record partly measures that maintenance.
- Strategies. Backtests run on companies that exist today cannot include the ones that went bankrupt along the way.
- People. You hear from investors whose concentrated bets worked. The identical strategy that failed produces no interviews.
Survivorship bias never makes results look worse than reality. It systematically inflates, which means the correct adjustment to almost any performance claim you encounter is downward by an unknown amount.
The question that cuts through it
Ask: what happened to everyone who started? Not what happened to the ones being shown to me.
A fund family advertising that most of its funds beat their benchmark over ten years is making a claim about funds that lasted ten years. The ones closed in year four are not in the numerator or the denominator.
Why it is hard to correct
Because the missing data is missing. Good research uses databases that deliberately retain dead funds and delisted companies, and the difference between those and naive figures is usually material.
The gap between survivor-only and full-sample results is large enough, in the long-running academic work on fund performance, to change conclusions rather than merely soften them.