In brief

Evaluating a mutual fund requires more than comparing returns. Start with the fund’s objective and holdings, select a benchmark that matches what it actually owns, compare total returns over identical periods, and then examine risk, fees, taxes, manager tenure, and consistency of style.

Alpha can help separate market exposure from unexplained return, but alpha depends on the model and sample. A positive historical alpha is not proof of skill, and a top-decile year is not evidence that performance will persist.

Step 1: identify what the fund is supposed to do

Read the prospectus and latest shareholder report. Record:

  • investment objective;
  • principal strategy and permitted securities;
  • benchmark;
  • portfolio turnover;
  • expense ratio and other fees;
  • manager names and tenure;
  • major holdings and sector exposure; and
  • distribution and tax history when held in a taxable account.

The SEC requires a standardized fee table in a mutual fund prospectus. Its Mutual Fund and ETF Fees and Expenses bulletin explains where to find shareholder fees and annual operating expenses.

Step 2: choose the right benchmark

A benchmark should resemble the opportunity set and risk exposures the manager actually uses. Comparing a small-value fund with the S&P 500 can make style cycles look like skill or failure. Comparing a global balanced fund with an all-equity U.S. index is worse.

Use three layers when available:

  1. Prospectus benchmark: the index selected for required presentation.
  2. Category or style benchmark: a second index reflecting the observed portfolio.
  3. Simple investable alternative: a low-cost fund an investor could realistically own instead.

If the conclusion changes completely with a reasonable benchmark change, it is not robust.

Step 3: compare total return on the same clock

Total return includes price changes and reinvested distributions. Compare the same start and end dates, and distinguish cumulative from annualized return.

For a holding period of n years:

Annualized return = (ending value ÷ beginning value)^(1/n) − 1

Do not compare a fund’s calendar-year return with an index’s trailing 12-month return. Also check whether the published result is before or after fund expenses and whether sales loads are included.

Step 4: ask how much risk produced the return

Useful measures answer different questions:

Measure Question Limitation
Volatility How widely did periodic returns vary? Treats upside and downside variation alike
Maximum drawdown What was the deepest peak-to-trough loss? One historical path, not a future worst case
Beta How sensitive was the fund to its market benchmark? Depends on benchmark and sample
Sharpe ratio How much excess return occurred per unit of total volatility? Sensitive to period and risk-free input
Alpha What return was unexplained by the selected model? Changes with the model and may be statistically noisy

The existing CAPM guide explains beta and one-factor alpha. The Fama–French three-factor and Carhart four-factor articles show how adding size, value, and momentum can change the attribution.

What alpha means—and does not mean

In a simple CAPM regression:

Fund excess return = alpha + beta × market excess return + error

Alpha is the intercept. A monthly alpha of 0.10% is sometimes annualized by multiplying by 12, producing approximately 1.2% under a simple convention. But the estimate needs a standard error and statistical test. A positive point estimate can arise from luck.

Alpha is model-dependent. A small-value manager may appear to have CAPM alpha because CAPM omits size and value exposures. A multifactor model can attribute part of that return to systematic factors instead of manager selection.

A transparent hypothetical comparison

Consider a fictional fund and benchmark over five years:

Item Hypothetical fund Benchmark
Annualized total return 9.0% 8.5%
Annualized volatility 17.0% 14.0%
Maximum drawdown −28.0% −22.0%
Expense ratio 0.80% 0.05%
Turnover 70% 5%

The fund led by 0.5 percentage points annually, but took more measured risk, suffered a deeper drawdown, charged more, and traded more. That is not enough information to declare success.

A fuller review would ask:

  • Was the fund’s beta above one?
  • Did its style differ from the benchmark?
  • Was the same manager responsible for the full period?
  • Was the difference statistically distinguishable from noise?
  • Would taxes from distributions and turnover change the outcome in a taxable account?
  • Did the fund remain open and investable at the observed scale?

Every number above is an invented teaching assumption, not a real fund result.

Fees and taxes are part of performance

Expense ratios are deducted from fund assets and therefore reduce reported net returns. Sales loads, account fees, and advisory fees may sit outside the expense ratio.

Taxes can create another gap. A mutual fund may distribute dividends and realized capital gains even when a shareholder did not sell. Turnover can increase the chance of realized gains, but turnover alone does not determine tax efficiency. Losses, creation/redemption mechanics, cash flows, and manager decisions also matter.

Compare after-tax returns when they are available and relevant, but verify their assumptions. Tax outcomes depend on the account and taxpayer.

Detect style drift

Style drift occurs when a fund’s actual portfolio moves away from the mandate or exposure an investor expected. Warning signs include:

  • market capitalization moving far outside the stated category;
  • sector or country concentration rising materially;
  • cash balances becoming unusually large;
  • derivatives changing the economic exposure; or
  • performance behaving unlike the stated benchmark.

Drift is not automatically misconduct or a bad decision. It may result from market appreciation, opportunity, defensive positioning, or a flexible mandate. But it changes the portfolio role and can create overlap elsewhere.

Survivorship bias

A database containing only funds that exist today omits funds that liquidated or merged. If weaker funds disappear more often, the surviving group looks better than the full opportunity set investors faced.

Stephen Brown, William Goetzmann, Roger Ibbotson, and Stephen Ross examined this problem in “Survivorship Bias in Performance Studies”. Their work shows why performance studies need dead funds and a clear rule for mergers and share classes.

Survivorship bias also affects informal lists such as “the ten best funds of the last decade.” The list is formed with knowledge of which funds survived to the end.

Does performance persist?

Mark Carhart’s 1997 study, “On Persistence in Mutual Fund Performance”, found that common factors and expenses explained much of persistence in equity mutual-fund returns. Strong one-year persistence was substantially related to momentum, while persistent poor performance was more evident than reliable superior skill.

Persistence should therefore be tested, not assumed:

  1. Use rolling periods rather than one convenient start date.
  2. Separate manager tenure from the fund’s full history.
  3. Compare multiple appropriate benchmarks.
  4. Examine factor-adjusted return and uncertainty.
  5. Include funds that closed or merged.
  6. Ask whether the strategy had enough capacity to accept new money.

A repeatable scorecard

Area Evidence to collect
Mandate Prospectus objective, strategy, constraints
Portfolio Holdings, concentration, turnover, style
Performance Same-period total return, rolling results
Risk Drawdown, volatility, beta, downside behavior
Attribution Benchmark and factor-model alpha
Cost Expense ratio, loads, advice, account fees
Tax Distributions, turnover, after-tax data
People Manager tenure, process, team continuity
Persistence Results across regimes, not one ranking date

Bottom line

A mutual fund is a process, portfolio, fee structure, and tax vehicle—not just a return series. Begin with what it owns and why, then judge performance against a compatible benchmark over a common period.

Historical alpha can identify a result that a model did not explain. It cannot reveal with certainty whether the cause was skill, omitted risk, luck, or a temporary market condition. A good evaluation makes that uncertainty visible.

For the theory behind that uncertainty, continue with Is the Stock Market Efficient?. For the investor decisions that can create a separate performance gap, read Why Investor Behavior Can Reduce Returns.

Sources