Survivorship bias is the error of studying only the winners while ignoring the losers. In finance, it occurs when an analysis includes only funds, stocks, or strategies that still exist and excludes those that failed, merged, or disappeared.

This bias distorts results in a consistently optimistic direction. If a database only contains funds that survived, the average performance of those funds looks better than the true average of all funds that ever existed. The same problem arises in stock indices, strategy backtests, and academic research. Recognizing and correcting for survivorship bias is essential to drawing accurate conclusions from historical data.

Definition

Survivorship bias is a form of selection bias (a distortion caused by looking at a non-representative subset of data) that occurs when failed or discontinued observations are excluded from a dataset. The remaining sample contains only "survivors," which systematically overestimates performance, success rates, or the reliability of a strategy.

The term applies broadly across many fields, from business (studying only successful companies) to medicine (analyzing only patients who completed a trial). In quantitative finance, it is one of the most common and most consequential sources of misleading results.

How Survivorship Bias Occurs

Survivorship bias can enter financial analysis through several channels. Each one removes poor performers from the dataset, leaving behind an artificially rosy picture.

Fund Databases

Mutual fund and hedge fund databases typically remove funds that close or merge. A fund that loses 40% of its assets and shuts down vanishes from the database. Future researchers who query "all funds in this category" will never see it. The average return of the remaining funds is higher than the true average because the worst performers have been erased.

This is especially severe in hedge fund databases, where reporting is voluntary. Funds that perform poorly often stop reporting before they close. A database that appears to contain "all hedge funds" may actually contain only those that chose to keep reporting, which skews toward better performers.

Stock Indices

Stock market indices like the S&P 500 regularly add and remove companies. When a company's market value shrinks or it goes bankrupt, it is removed from the index. Researchers who backtest a strategy using "today's S&P 500 constituents" are implicitly selecting for survivors. Companies that failed, were acquired at distressed prices, or shrank out of the index are missing from the analysis.

Strategy Research

Quantitative researchers often test many strategies before publishing the one that worked best. The strategies that failed are never published. A reader who sees a published strategy with strong historical performance has no way of knowing how many failed strategies were tested alongside it. This is closely related to overfitting and the pitfalls of backtesting.

Practical Example: Mutual Fund Performance

Consider a simplified example with 100 mutual funds launched in 2010.

The 35 funds that closed had poor performance. Removing them raised the average return from 6.2% to 8.5%. A researcher using a survivor-only database would conclude that mutual funds in this category earned 8.5% per year, overstating true performance by 2.3 percentage points annually. Over a 15-year period, that difference compounds dramatically.

How Much Does Survivorship Bias Distort Results?

The size of survivorship bias varies by asset class and time period, but multiple academic studies have quantified it.

Even a 1% annual bias is meaningful. Over 20 years, the compounding effect of a 1 percentage point overstatement grows substantially, potentially changing investment conclusions entirely. What appears to be a modest annual difference becomes a significant gap in cumulative wealth when compounded over a full investment horizon.

How to Mitigate Survivorship Bias

Completely eliminating survivorship bias is difficult, but several practices reduce its impact.

• Use survivorship-bias-free databases. Some data providers maintain records of defunct funds and delisted stocks. The CRSP (Center for Research in Security Prices) database at the University of Chicago, for example, includes delisted stocks and is considered the gold standard for U.S. equity research.
• Include dead funds in any comparison. When evaluating a fund category, actively seek out funds that closed or merged during the study period. If the database does not track them, note this as a limitation.
• Use point-in-time constituent lists. When backtesting strategies on an index, use the actual index constituents at each historical date, not today's constituents applied backward. This is sometimes called "as-reported" or "point-in-time" data.
• Apply out-of-sample testing. Test a strategy on data it was never trained on. If performance degrades sharply out of sample, survivorship bias (or overfitting) may explain the in-sample results. See backtesting pitfalls for more detail.
• Report the attrition rate. State how many funds, stocks, or strategies dropped out of the sample during the study period. A high attrition rate is a warning sign that survivorship bias may be significant.

Limitations of Survivorship Bias Corrections

Further Reading

• Elton, E.J., Gruber, M.J., and Blake, C.R. (1996). "Survivorship Bias and Mutual Fund Performance." The Review of Financial Studies, 9(4), 1097–1120.
• Brown, S.J., Goetzmann, W., Ibbotson, R.G., and Ross, S.A. (1992). "Survivorship Bias in Performance Studies." The Review of Financial Studies, 5(4), 553–580.
• Fung, W. and Hsieh, D.A. (2000). "Performance Characteristics of Hedge Funds and Commodity Funds: Natural vs. Spurious Biases." The Journal of Financial and Quantitative Analysis, 35(3), 291–307.
• Harvey, C.R., Liu, Y., and Zhu, H. (2016). "...and the Cross-Section of Expected Returns." The Review of Financial Studies, 29(1), 5–68.

Related Terms