Definition
Survivorship bias arises when closed funds, bankrupt companies or other objects that did not survive disappear from a historical comparison. The result becomes a history of today’s participants, not the outcome of the earlier selection rule. It is a specific selection problem relevant when checking claims of performance. The SEC recommends attention to the choice of periods and successful investments in performance presentations.[27](#source-market-performance)
Example and limitation
If four hypothetical investments returned plus twenty, plus ten, minus ten and minus one hundred percent, the equal-weighted result is minus twenty percent. Excluding the last one changes the average to approximately plus 6.67%. The sample improved; the strategy did not.
Not every disappearance means a total loss: an acquisition and payment may occur. Determine the actual event result or state the uncertainty explicitly. Preserve the historical constituents and position-closing rules when testing. Removing this bias is necessary, but by itself it does not prove future profitability or manager skill.