Hypothetical worked example and primary educational references; not observed market data.
The visible universe may be incomplete
A current list of funds contains investments that still exist. It may exclude funds that closed, merged, or otherwise left the dataset. If an analyst evaluates the past using only that current list, selection can depend on outcomes that occurred after the starting date. Investor.gov cautions against treating past fund performance as a prediction; survivorship analysis adds an original data-quality question before even reaching that caution. Ask what population existed at the beginning and what happened to every member. A historical comparison cannot describe the full original opportunity set if it silently removes inconvenient departures.
Supporting references: 1. Investor.gov: Mutual Funds, Past Performance
Construct a complete hypothetical starting set
Imagine four invented funds, each available at the start of one period. Their returns are plus 10%, plus 6%, minus 20%, and minus 12%. Suppose the two negative-return funds subsequently close and disappear from a current directory. The full starting-set arithmetic average is minus 4%, while the average among remaining funds is plus 8%. These figures are constructed to expose the selection mechanism; they are not statistics about fund closures. The apparent improvement arises entirely from excluding observations. No investment changed its realized return when the analyst revised the sample.
Hypothetical worked example
Full starting set versus surviving subset
Model setup
Read these assumptions before comparing the scenarios.
- All monetary amounts in this example are hypothetical U.S. dollars (USD), not current prices, quoted account terms or observed market data. Percentages, counts, time periods and index levels retain their stated units.
- Four invented fund returns are 10%, 6%, -20%, and -12%.
- Only the first two remain in the hypothetical current directory.
Arithmetic mean return (%)
- 1. Original starting set
- -4.00
- 2. Survivors only
- 8.00
Shared zero-based value scale. Range: -4.00 to 8.00, in the units above. Case numbers match the labels; no probability or time trend is implied.
Hypothetical model, not observed market data or a forecast. Assumptions above; original values and calculations below.
| Sample | Included funds | Arithmetic mean return (%) |
|---|---|---|
| Original starting set | 4 | -4.00 |
| Survivors only | 2 | 8.00 |
Calculation details
- Full mean
- -4.00 %
((((10 + 6) + -20) + -12) / 4) - Survivor mean
- 8.00 %
((10 + 6) / 2)
Closure is not always equivalent to failure
Funds can close or merge for many reasons, so a departing fund should not automatically be assigned a poor return. The problem is excluding departures without accounting for their actual histories, not assuming all departures performed badly. A defensible dataset records outcomes and distinguishes missing information from zero returns. For mergers, the investor's continuing exposure may also need to be traced. If a record cannot be reconstructed, report the gap instead of filling it with a convenient value. The direction and size of bias depend on how selection relates to outcomes in the specific data being evaluated.
Investor experience follows capital, not directory entries
An investor owning a fund that later disappears still experienced its gains, losses, distributions, and any transfer or redemption. A survivor-only table can erase that capital path even though the financial consequences were real. When testing a strategy, retain securities that were actually eligible at each point in time and use information then available. Avoid choosing only today's successful companies for a historical simulation. That would use future knowledge to rewrite the starting universe. The important unit is the investment opportunity and its subsequent cash-flow path, not whether its name remains easy to find now.
Arithmetic averages are only one possible summary
The hypothetical average treats each fund return equally for one period. A money-weighted investor result, an asset-weighted category result, or a compounded multi-period strategy would require another calculation. Survivorship problems can affect each of these measures, but their denominators and weighting rules differ. Specify the statistic before discussing bias. Investor.gov's shareholder-report guidance offers information about an individual fund's performance, yet a report for a surviving fund cannot by itself reconstruct a missing population. Claims about an entire universe need evidence covering that universe rather than a collection of individually valid survivor reports.
Supporting references: 2. Investor.gov: How to Read a Mutual Fund or ETF Shareholder Report
Read rankings as selected evidence
A ranking should disclose inclusion dates, treatment of closures and mergers, weighting, costs, and whether failed records were retained. If those details are unavailable, the reader should narrow the conclusion rather than assume completeness. A survivor-only result may describe the survivors accurately while failing to describe the original choice set. That distinction is the central lesson. Historical evidence remains conditional even after selection issues are addressed; it still does not promise future outcomes. Careful universe reconstruction improves the integrity of the comparison without turning a backward-looking analysis into a dependable forecast or a recommendation of current winners.
Test another situationCarry closed-fund proceeds through the comparison endpoint
A separate hypothetical case. Its inputs differ from the main worked example above.
Carry closed-fund proceeds through the comparison endpoint
Suppose a hypothetical strategy starts with USD 4,000 in each of three funds. At a common year-end, the first is worth USD 4,400 and the second USD 4,200. The third closed earlier and returned USD 2,000, retained as non-interest-bearing cash through year-end. Combined resources are exactly USD 10,600. Omitting the closed fund would erase part of the initial capital and misdescribe the investor's experience. Reinvesting its proceeds would create another cash-flow path that needs stated dates and prices. This illustration concerns accounting for departures; it supplies neither observed closure returns nor a prediction that closed funds systematically perform worse.
Distinguish a missing identifier from vanished wealth
Investor.gov cautions against projecting prior fund performance into future results. Before making even a historical statement, reconcile each initial investment through liquidation, merger, renaming or continued ownership. A discontinued ticker is not enough to determine the outcome. Follow cash receipts and successor shares where evidence permits, and mark unresolved histories as missing rather than zero. This may leave a narrower defensible sample, which should be disclosed with its limitations. A perfectly accurate table of surviving names can still answer the wrong population question. Preserve the original eligibility list so later revisions can be checked against the opportunities actually available then.
Supporting references: 1. Investor.gov: Mutual Funds, Past Performance
Check this case’s calculations
- Resources including closed-fund cash
- 10,600.00 USD
((4400 + 4200) + 2000)
Limits of this additional case
- Closure proceeds are invented; actual histories may require merger conversions, distributions and dated reinvestment records.
Limits of this analysis
- Departures can occur for reasons unrelated to performance.
- The example uses equal-weight arithmetic averages rather than an investor wealth path.
- The official references support performance interpretation; the selection example is independent analysis.
Sources and scope
- Investor.gov: Mutual Funds, Past Performance
Past fund performance does not necessarily predict future results.
Scope: U.S. investor education; supports the stated mechanism, not the hypothetical scenario or a forecast.. Reference accessed 2026-10-06.
- Investor.gov: How to Read a Mutual Fund or ETF Shareholder Report
Reports provide performance, benchmark and portfolio context; past returns are not predictions.
Scope: U.S. investor education; supports the stated mechanism, not the hypothetical scenario or a forecast.. Reference accessed 2026-10-06.
Illustrative calculations are not forecasts or personal recommendations. Editorial standards
