Survivorship bias is the error of evaluating a strategy, an asset class, or an investment approach based on the outcomes of survivors—the funds still in operation, the strategies that produced positive returns, the investors who are publicly visible—while ignoring the much larger population of failures that has been silently removed from the data. This bias systematically overstates the performance of investment approaches, understates the failure rates of investment vehicles, and creates the impression that strategies which actually produce poor outcomes for most practitioners are reliable generators of wealth.
The most straightforward example is the mutual fund industry. At any given time, the universe of available funds represents the survivors—the funds that have performed well enough, or at least not badly enough, to remain in operation. Funds that performed poorly were liquidated or merged into better-performing funds, removing their track records from the available data. The average performance of the surviving fund universe therefore looks substantially better than the average performance of all funds that were actually available to investors during the relevant period, because the worst performers have been systematically removed. Studies that adjust for survivorship bias consistently show that the typical actively managed fund underperforms its benchmark by more than naive analysis of the current fund universe would suggest.
The same bias operates in the evaluation of investment strategies. The strategies that attract attention and study are the ones that produced impressive results for their practitioners. The far larger number of practitioners who applied similar approaches and produced poor results did not write books, give interviews, or attract academic study. The investment approach that appears, from the visible evidence, to have consistently produced strong returns has actually produced strong returns for the visible minority of practitioners while producing poor or negative returns for the invisible majority. The survivorship selection process creates an evidence base that is exactly as misleading as the evidence would be if one evaluated the safety of Russian roulette by interviewing survivors.
The individual investor is particularly vulnerable to survivorship bias in the context of social proof. The investment ideas that circulate in social networks are overwhelmingly the successes—the positions that appreciated, the calls that proved correct, the strategies that happened to work in the relevant period. The failures are not discussed because they are embarrassing and because the investors who experienced them are not promoting their approaches. The resulting social information environment systematically overstates the probability of success for the approaches being discussed, because the selection process for what gets discussed is survival.
Correcting for survivorship bias requires the deliberate search for evidence about failures, not just successes. Before adopting an investment approach that has produced impressive results for visible practitioners, the investor should ask: what is the full population of investors who have tried this approach? What fraction succeeded? What happened to those who failed? These questions are harder to answer than questions about the successes, because the failures have been removed from the visible data, but they are the questions that provide accurate information about the actual distribution of outcomes the approach is likely to produce.