FRM Exam Part I · Fund Management

Hedge Fund Database Biases and Performance Evidence

Updated 11 October 2026 · Fact-checked

Hedge fund databases rely on voluntary reporting, so reported returns are distorted. Survivorship bias drops dead funds, backfill bias adds prior history when a fund joins, and selection bias reflects who chooses to report. All three usually overstate returns. Evidence on alpha is mixed, and it tends to shrink after bias adjustment.

Understand Hedge Fund Biases and Performance Evidence

Hedge funds are not required to publish returns the way mutual funds are. Managers choose whether to report to a commercial database. That choice creates data problems. Any average return you compute from a database is only as good as the sample behind it.

Survivorship bias arises when funds that stop reporting, usually because of poor performance or liquidation, are removed from the database. The remaining sample holds only the winners that survived. Average returns are pushed up. Risk and failure rates are understated. The bias is larger when attrition is high and when the dead funds were weak performers.

Backfill bias (also called instant history bias) arises when a fund joins a database and its past returns are added. Managers tend to start reporting after a good run, because good results attract investors. So the added history is skewed to strong returns. Backfilled returns therefore overstate performance. A common fix is to drop the backfilled months and use only returns reported after the date of entry.

Selection bias arises because funds choose whether to report. Some top funds are closed to new money and see no reason to report. Some poor funds do not report. The direction of this bias is therefore ambiguous, and it can be positive or negative. Related issues include multi-period sampling bias, where funds with short histories are excluded, and illiquid asset smoothing, where stale prices understate volatility and correlation.

On performance, the evidence is mixed. Raw database returns often look attractive, with positive alpha. After adjusting for biases, for fees, and for exposure to risk factors such as equity, credit and option-like payoffs, alpha is smaller. It also varies across strategies and over time. Some funds persist in performance over short horizons, but persistence is weak and hard to exploit. Treat any headline alpha with caution.

Key formulas to remember

Survivorship bias (approximate): Bias ≈ Return of surviving funds − Return of all funds (including dead); Usually positive. Measured as the annual return gap between a survivor-only sample and the full sample.
Backfill bias (approximate): Bias ≈ Average return including backfilled months − Average return of post-listing months only; Usually positive. Fix: delete returns before the date the fund was added.
Alpha from a factor regression: Rₜ − R_f = α + β₁F₁ₜ + … + βₖFₖₜ + εₜ; Alpha is the intercept: return left after paying for factor exposures. Bias in the data biases alpha.
Direction summary: Survivorship: upward | Backfill: upward | Selection: ambiguous; Memorize this one-line summary for MCQs.

How to solve Hedge Fund Biases and Performance Evidence questions

Use this method for any question on hedge fund database biases or performance evidence.

1Read the question and find the data feature described: dead funds removed, history added on entry, or voluntary reporting choice.
2Match the feature to the bias: removed dead funds is survivorship, added past returns is backfill, who chooses to report is selection.
3Decide the direction. Survivorship and backfill overstate returns. Selection can go either way.
4If numbers are given, compute the gap between the biased sample and the unbiased sample, and check the sign.
5Check for the fix: drop backfilled data, include defunct funds, use multiple databases.
6For performance evidence, ask whether alpha is measured after fees and after controlling for risk factors, and whether biases were corrected.
7Choose the answer that says alpha is reduced after corrections, not one claiming it is always positive or always zero.

Quickest way: Keyword-to-bias shortcut

When to use it: Use on conceptual MCQs where you must name the bias or its direction in under a minute.

Spot the trigger phrase: dropped or defunct means survivorship; added past returns or incubation means backfill; voluntary or choose to report means selection.
Apply direction: first two are upward, selection is ambiguous.
Eliminate options with absolute words such as always or never.
For numeric gaps, subtract the full-sample return from the biased return.

Common mistakes in Hedge Fund Biases and Performance Evidence

Treating survivorship bias and backfill bias as the same thing.
Both overstate returns, so they blur together.
Fix: Survivorship is about funds that disappear. Backfill is about history added when a fund enters. Ask: is the issue missing funds or added months?
Saying selection bias always overstates returns.
Students generalize from the other two biases.
Fix: Selection bias can be upward or downward, since some good funds also decline to report.
Saying survivorship bias understates risk and ignoring it for return.
Focus on only one effect.
Fix: It overstates average return and understates the chance of failure and downside risk.
Claiming evidence shows hedge funds consistently earn large positive alpha.
Raw database averages look impressive.
Fix: State that alpha is smaller after bias, fee and factor adjustments, and varies by strategy and period.
Thinking backfill bias is fixed by adding more funds.
Confusing sample size with data quality.
Fix: Remove the backfilled return months, keeping only returns after the listing date.

Worked examples

Example 1

A database of surviving hedge funds shows an average annual return of 11.2%. Including funds that closed during the period, the average annual return is 9.1%. What is the estimated survivorship bias, and in which direction does it act?

Show the solution

Bias = return of survivors − return of all funds.
Bias = 11.2% − 9.1% = 2.1%.
The sign is positive, so the survivor-only sample overstates returns.

Answer: Survivorship bias is about 2.1% per year, overstating returns.

Example 2

A fund manager launches a fund, posts strong returns for 18 months, and then submits the fund to a database that adds all 18 months of history. Which bias results, and what is the standard fix?

Show the solution

The added history was chosen after good results, which is not representative.
Added past returns on entry is backfill (instant history) bias.
Its effect is to overstate average returns.
The fix is to exclude returns dated before the entry date.

Answer: Backfill bias, which overstates returns; fix it by deleting the backfilled months and using only post-listing returns.

Exam tips

Memorize direction: survivorship up, backfill up, selection ambiguous. This is the most common test point.
Watch for the synonym instant history bias for backfill bias.
For performance evidence, expect answers saying alpha shrinks after adjusting for bias, fees and risk factors.
Look for fixes in options, such as removing backfilled data or including dead funds.

Practice questions from Fund Management

Hedge Fund Biases and Performance Evidence in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

Hedge Fund Biases and Performance Evidence: frequently asked questions

What is backfill bias in hedge funds?

It happens when a fund joins a database and its earlier returns are added. Managers usually start reporting after good performance, so the added history is skewed upward. The average return in the database is then overstated.

What is the difference between survivorship bias and backfill bias?

Survivorship bias comes from removing funds that stopped reporting, leaving only survivors. Backfill bias comes from adding past returns of funds when they enter. Both usually raise reported returns, but they arise from different data problems.

Does selection bias always overstate hedge fund returns?

No. Funds choose whether to report. Poor funds may stay out, which raises the average, while some strong funds that are closed to new investors may also stay out, which lowers it. The net direction is ambiguous.

Do hedge funds generate alpha according to the evidence?

The evidence is mixed. Raw data often shows positive alpha, but it falls after adjusting for biases, fees and factor exposures. Results differ by strategy and period, and persistence is limited.