What is Survivorship bias?

Survivorship bias is the mistake of drawing conclusions only from what is still around, while ignoring whatever dropped out along the way. The survivors are usually the successes, so the picture looks better than reality. In investing, it appears whenever a dataset contains only the funds, companies or strategies that still exist today. Funds that do badly are often closed or merged into other funds, and failing companies are removed from stock indices, so their poor results vanish from many lists and comparisons.

Say 100 funds launch at the same time, each with $10,000 of investors' money. Over ten years, 40 of them perform poorly and are closed; while they existed, they averaged 2% a year. The 60 that remain averaged 8% a year. A database listing only today's funds shows an average of 8%. Counting every fund gives a different answer: 60 times 8 is 480, 40 times 2 is 80, and 560 divided by 100 is 5.6%. The survivors overstate the typical result by 2.4 percentage points a year. Over ten years, $10,000 growing at 8% reaches about $21,600, while at 5.6% it reaches about $17,200, a gap of roughly $4,300.

The same trap appears in backtests, which are hypothetical tests of how a set of rules would have performed on past data. A test that picks stocks from today's members of an index such as the S&P 500 includes only companies that grew large enough to be there now. It leaves out those that shrank, were taken over or went bankrupt, so the hypothetical results look stronger than anything achievable at the time. A common misunderstanding is that index funds suffer from the same problem. Their reported returns do not: a fund tracking the index held those companies before they were removed, so their losses are already included. The bias lies in the analysis, not in the index itself.

For a private investor building a long-term portfolio, survivorship bias matters because so much published performance is assembled after the fact. Lists of top funds, stories of people who made fortunes and impressive hypothetical strategy results all tend to come from the winners' side. Careful investors ask how many funds, stocks or strategies started out, not only how the remaining ones did. Accounting for the failures usually leads to more modest and more realistic expectations.

By Dr. Harry Hamann

What is an example of survivorship bias in investing?
Fund comparisons that include only funds still open today. Poorly performing funds are often closed or merged, so their weak results disappear. The average return of the remaining funds then looks higher than what investors in the whole group actually earned.

How does survivorship bias affect backtests?
A backtest that uses only companies or funds that exist today ignores those that failed or were removed. The hypothetical results then look better than anything achievable in real time. More reliable tests use point-in-time data that includes delisted and closed securities.

Do index funds have survivorship bias?
Not in their reported returns. An index fund held companies while they were in the index, so losses before their removal are already included. The bias arises when researchers study only today's index members as if those companies had been chosen in advance.