Key takeaways
- Over the 50 years to June 2026, equal-weighting the largest 30% of US stocks returned 12.04% a year against 11.93% for cap-weighting, a gap of 0.11 percentage points before any cost.
- Equal weight carried 16.70% annualised volatility against 15.03%, a worst fall of 55.7% against 49.8%, and a lower Sharpe ratio: 0.518 against 0.551.
- The cheapest UK equal-weight tracker charges 0.15% a year against 0.07% for the cap-weighted one, so 0.08 of the 0.11-point gross edge goes on the fee alone.
- Invesco's US equal-weight fund turned over 27% of its portfolio in the year to April 2026. Vanguard's cap-weighted S&P 500 fund turned over 2%.
- Concentration is the real counter-case: ten companies made up 37.56% of the iShares Core S&P 500 in July 2026, with 7.54% in NVIDIA alone.
Equal weight vs market cap: the 50-year answer is 0.11 percentage points
You've probably seen the claim that equal-weighting the S&P 500 beats the ordinary version. It's a real result. It's also much smaller than it sounds, and it hasn't been true for a decade.
Here's the rebuild. Kenneth French's data library publishes a value-weighted and an equal-weighted return series for the same set of US stocks, monthly, drawn from the CRSP database. Take the largest 30% of the market by size, the group that most resembles a large-cap index, and run it from July 1976 to June 2026. Cap-weighting returned 11.93% a year. Equal-weighting returned 12.04%.
Set equal weight against market cap over that period and the whole argument is 0.11 percentage points a year, gross of every fee, spread and tax. A UK investor buying an equal-weight index fund pays 0.15% a year for it, against 0.07% for the cap-weighted index. Most of the historical edge is gone before the first trade settles.
What follows is why the gap is that small, why it wasn't always, and what you actually own when you equal-weight.
Fifty years of American stocks, weighted both ways
French's size portfolios are formed at the end of each June, using June market equity and NYSE breakpoints. They include, in his description, "all NYSE, AMEX, and NASDAQ stocks for which we have market equity data for June" of that year. The top 30% group held 518 companies at the June 2026 formation and 469 at the July 1976 one, so it is a large-cap universe of roughly index size.
Run both weighting rules across the full 50 years and this is what comes out.
- Cap-weighted: 11.93% a year, 15.03% annualised volatility, a worst peak-to-trough fall of 49.8%, and a Sharpe ratio of 0.551.
- Equal-weighted: 12.04% a year, 16.70% volatility, a worst fall of 55.7%, and a Sharpe ratio of 0.518.
Sharpe ratio here means return above cash per unit of volatility, calculated with French's own risk-free series. On that measure the cap-weighted portfolio won, because equal weight's extra return didn't cover its extra bumpiness. A pound in the cap-weighted series grew 280.6 times over the 50 years. In the equal-weighted one it grew 294.8 times.
That is the honest headline, and it is not the one the strategy is usually sold on.
The 0.11 points hide five very different decades
An average this thin is made of much larger pieces. Split the 50 years into decades and the equal-weighted minus cap-weighted gap runs like this.
- July 1976 to June 1986: cap-weighted 14.48%, equal-weighted 16.50%. Equal weight ahead by 2.02 points a year.
- 1986 to 1996: 13.64% against 13.68%. Ahead by 0.04.
- 1996 to 2006: 8.28% against 9.51%. Ahead by 1.23.
- 2006 to 2016: 7.62% against 8.33%. Ahead by 0.71.
- 2016 to 2026: 15.90% against 12.39%. Behind by 3.51.
The chart above runs the same thing as a total. It divides a pound compounded in the equal-weighted portfolio by a pound compounded in the cap-weighted one, from a base of 100 in June 1976. That ratio reached 119.2 by 1986, 133.8 by 2006 and 142.9 by 2016. By June 2026 it was back to 105.1. Forty years of accumulated advantage, and most of it handed back in ten.
Note what that last figure does to the reading of the first four. If you had run this study in 2016 you'd have found equal weight ahead in every decade since 1976, and you'd have been describing a pattern that was about to reverse hard.
That is the standing hazard with a rebuilt series. The result you get depends on where you stop.
What you buy with equal weight is a small-company tilt
The mechanism isn't mysterious. Equal-weighting takes money away from the biggest companies in an index and hands it to the smallest ones in the same index. The portfolio's fate then leans on the gap between small and large.
French's size factor measures that gap directly. Over the same 50 years to June 2026 it compounded at 1.07% a year. Over the decade to June 2026 it compounded at -2.00% a year. One sign flip, and most of the decade table above is explained. Our piece on the size premium after Banz follows where the rest of that premium went.
A valuation tilt rides along with it. In July 2026 the iShares equal-weight fund carried a price-to-earnings ratio of 23.53 and a price-to-book of 3.13. Its cap-weighted sibling carried 29.54 and 5.49. The same 500 companies, weighted differently, produce a measurably cheaper book. That's why the equal-weighted record tracks the value premium more closely than a cap-weighted one does.
Neither tilt is hidden and neither is free. You're taking a factor bet, and factor bets go quiet for a decade at a time.
Equal weight vs market cap is a cost argument before it's a returns argument
Equal weights don't hold themselves. Prices move, weights drift, and something has to trade them back. The S&P 500 Equal Weight Index resets on a schedule: Invesco's prospectus describes an index that "assigns each component security the same weight at each quarterly rebalance." As of June 2025 that index held 503 constituents, with market values running from $5.2 billion to $3.9 trillion.
You can read the bill straight off the filings. Invesco's S&P 500 Equal Weight ETF (RSP) reported portfolio turnover of 27% for the year to April 2026. Vanguard's S&P 500 ETF (VOO) reported 2%. The fee tables have the same shape: 0.20% a year against 0.03%.
That's a 0.17-point cost gap set against a 0.11-point gross edge over half a century. On the American pairing, the arithmetic is negative before anything else happens.
UK buyers face a narrower spread. The iShares S&P 500 Equal Weight UCITS ETF (EWSP) charges 0.15%; the iShares Core S&P 500 UCITS ETF (CSPX) charges 0.07%. Both are ISA eligible, both were holding 504 positions in July 2026. The 0.08-point gap still absorbs roughly three quarters of the 50-year edge, and the turnover the fund pays on your behalf sits underneath it, unbilled. Our breakdown of turnover drag puts a number on that second layer.
Two independent datasets agree the last decade was brutal
Invesco's audited annual report puts the S&P 500 Equal Weight Index at 11.96% a year over the ten years to April 2026, against 15.26% for the S&P 500. That's a gap of 3.30 points a year. Over five years it is 8.50% against 13.14%, a gap of 4.64.
French's series, built from a different universe under a different membership rule, put the same decade at 15.90% for cap weight and 12.39% for equal weight. Two datasets that share no methodology land within half a point of each other on the size of the gap. When independent sources agree that closely, the result is usually about the market rather than the measurement.
The UK funds show it year by year. On benchmark returns, the S&P 500 returned 25.67% against 13.17% for its equal-weighted twin in 2023, 24.50% against 12.37% in 2024, and 17.43% against 10.80% in 2025. Gaps of 12.50, 12.13 and 6.63 points.
Three consecutive years like that is why the question keeps getting asked, and why it gets answered so loudly in both directions.
The strongest objection has nothing to do with returns
Here is the case for equal weight that survives all of the above. Cap-weighting is not a neutral default. It is a rule that puts the most money into whatever has already risen most, and right now it produces a portfolio very few people would assemble deliberately.
In July 2026, ten companies made up 37.56% of the iShares Core S&P 500 fund, with NVIDIA alone at 7.54%. Vanguard's 500 Index Fund reported 34.4% of net assets in information technology at the end of 2025. The equal-weighted fund's ten largest positions came to 2.77% of the portfolio, and one of those ten was 0.46% in cash. Its biggest company holding was PayPal, at 0.27%. In Invesco's US fund the top name in April 2026 was Intel, at 0.38%.
Those are different products wearing the same index name. A Sharpe ratio computed over a period when concentration mostly paid does not capture the difference, and neither does a return gap. If you hold a tracker alongside a US technology fund, the overlap is larger than the two labels suggest, which is the arithmetic our note on NVIDIA's index weight works through.
The reply to that objection is not that concentration is fine. It is that the insurance has a posted price. In the UK pairing it costs 0.08 points a year in fees, plus turnover, and over the last decade it cost 3.51 points a year in return. Whether that price is worth paying rests on a forecast that neither dataset here supplies.
What this rebuild cannot tell you
Start with what the 50-year series is not. French's top-30% group is not the S&P 500. It applies NYSE breakpoints across NYSE, AMEX and NASDAQ, so membership is mechanical where S&P's is set by a committee. It held 518 companies in June 2026 against the 508 positions in Invesco's fund. Close is not the same.
The equal-weighted series is an arithmetic average of each month's returns, which is the same thing as a portfolio reset to equal weights every month. Real equal-weight funds reset quarterly. So the backtest rebalances far more often than any fund does, and pays nothing for the privilege. Both series are gross: no fees, no spreads, no tax, no gains realised along the way.
The sample is one country. Every figure here is American. It is also one path through history, 50 years that happened to contain a long decline in inflation and two of the largest bull markets on record. A backtest is a record, not a forecast, and a different 50 years could reorder every row.
The fund figures cover short windows too. The Invesco record ends in April 2026, and the UK calendar-year comparison covers 2023 to 2025. Three years is a sample of three, whichever way it points.
What would change the conclusion
If the cost gap closed. Everything above turns on 0.08 to 0.17 points of annual cost set against 0.11 points of gross edge. Equal-weight funds are already cheaper than they were, and if that spread narrowed towards zero the historical case would stop cancelling itself out. Our comparison of the cheapest index fund against tracking difference shows how far a headline fee can sit from the number you actually pay.
If the size factor turned. It compounded at 1.07% a year over 50 years and at -2.00% over the last decade. If it reverted to its long-run sign, the decade table would read differently and the last ten years would look like the anomaly rather than the trend. Nothing in either dataset says when that happens, or whether.
If the account is taxable. Portfolio turnover of 27% against 2% is not only a trading cost. In an unsheltered account it is a stream of realised gains arriving on a schedule you don't control. Inside an ISA or a SIPP both funds face the same tax, which is none, and the comparison stays the clean one above.
The number worth watching isn't the return gap. That is noisy, and it mostly reports what already happened. It is the top-ten weight. LedgerTouch shows that figure across your own holdings, which is where the overlap between a tracker and a sector fund tends to hide. At 37.56% it is the reason anyone is asking about equal weighting at all, and if it falls back, the question answers itself without a single trade.