How do you choose a benchmark that can't be gamed?

If you get to pick the yardstick after the race, you always win. Professional performance measurement therefore spends most of its effort on the yardstick, not the arithmetic. Here is the checklist that survives contact with reality.

Match the universe

The benchmark should hold the same kind of assets as the portfolio. A European small-cap value fund measured against a large-cap US index is being measured against a different asset class wearing the word "stocks". When the styles diverge — and they diverge for years at a time — the comparison reports the style gap, not the manager.

Practical version: match asset class, region, size, and style. If the fund says "global infrastructure", the benchmark says global infrastructure too.

Match the return basis

This one quietly decides more arguments than any other. Indices are published in two forms:

• Price return — index level only, dividends ignored.
• Total return — dividends reinvested.

A portfolio's own return almost always includes the dividends it received. Comparing that to a price index is comparing a total-return number to a price-return number. For broad developed-market equity the dividend contribution has historically run in the region of two percentage points a year — enough to flip most "we beat the index" claims. Total return versus total return, always.

Same discipline for fees: gross-of-fee portfolio returns against an index that costs nothing to track is a comparison with a thumb on the scale. Judge net.

Match the currency

A euro-based investor holding a US index earns the index return plus or minus the dollar move. Measure a euro portfolio against a dollar-denominated index and you have measured the exchange rate. Benchmark in the currency the investor actually lives in.

Blend when the portfolio is blended

A multi-asset portfolio needs a multi-asset yardstick. Build it by weighting the component index returns:

Blended benchmark = Σ (target weight × index return)

A portfolio with a 60% equities / 40% bonds mandate, in a year where the equity index returned +18% and the bond index returned −1%:

0.60 × 18% + 0.40 × (−1%) = 10.8% − 0.4% = +10.4%

That 10.4% is what a simple, cheap, rule-following version of the same mandate delivered. Judge the portfolio against that, not against whichever of the two components happened to shine.

Note the phrase target weight. Use the mandate's weights, not the weights the portfolio drifted into — otherwise the benchmark moves with the decisions it is supposed to be judging.

In the data

The S&P 500 index and the fund that tracks it on the same day, each with its traded close and its close adjusted for dividends:

For the index the two figures are identical, on that day and on every other: that is what a price index is, and the constituents' dividends are not in it. For the fund the adjusted figure differs, because the dividends it has paid since are folded back in. Measure a portfolio's total return against the index's line and the gap you report silently contains the index's entire dividend yield.

Try it now

1. Pick any fund and read its stated universe — region, asset class, size — from its record. Name the index that matches it, then name one index that would not be a fair match, and say why in one sentence. One fund's record is below; to pick another, open it in the Terminal (the link starts on SPY; change the symbol).

Open SPY.US — fundamentals in the EODHD Terminal

2. Put the dividend gap on the table. The index below leaves the dividends out; the fund's adjusted line carries them. Both are at full length — Measure 2015 to today on each. Compute both ten-year returns and note how far apart they end up.

3. Build a 60/40 blended benchmark for calendar 2025 from the two tables below, the S&P 500 index and the bond fund over the same window. Compute each year's return from the first and last close, then show the two weighted terms and the sum. For the bond leg, do it once with the traded close and once with the adjusted close, and say which of the two belongs beside a price index, and why neither pairing is quite fair.