Investing in the S&P 500: index, equal-weight or factors

• Cap-weighted: the index as everyone knows it. Each company weighs according to its market value, so the largest ones dominate. It is what a plain S&P 500 index fund tracks. Here it is measured by SPY, a US exchange-traded fund that has existed since 1993 and charges 0.09% a year.
• Equal-weight: the same companies, in equal parts. Each one weighs the same and the mix is rebalanced periodically, so mid-sized companies weigh more and giants less. Here it is measured by an index the site's own engine computes from every company that was in the S&P 500 each month; the real fund that tracks it, RSP, has existed since May 2003 and charges 0.20%.
• Factors: a selection made with a rule. The ~500 are scored on momentum, value, quality, growth or a blend of the four, and only the top-scoring ones are held. Here it is measured by the site's twenty model portfolios, rebuilt month by month.

SPY and RSP are named here as yardsticks, because they are the funds the site measures against, not as a suggestion to buy them. They are US-domiciled funds and, under European rules on product disclosure documents, a retail investor in Spain normally cannot buy them directly; the same idea exists in other vehicles.

The monthly column is the only one where all four rows are measured the same way. It reads shallower than the daily one — whatever falls and recovers within a month does not show — but it is fit for comparison. All four ways lost more than half in the 2008-2009 crisis. None of them avoided that.

Three readings, each with its caveat:

• Equal-weight beat the regular index by 1.7 points a year over this window, with a somewhat deeper drawdown. But it is a paper index, before costs: where both exist, the calculation beats RSP by 0.57 points a year, which is what the real fund charges and spends on trading.
• The equal-weight factor portfolio beat equal-weight by 1.0 point a year, with more volatility (25.35% against 20.27%) and a deeper drawdown.
• The cap-weighted factor portfolio beat SPY by 5.0 points a year, the figure on the site's home page. But weighting 20 companies by size, with no cap, concentrates: the largest position weighed 29.68% at the median rebalance and reached 67.99% on March 20, 2025, and in half the months the money was really spread over fewer than seven positions.

Four periods each. The regular index won in the dot-com bubble, fell less in the financial crisis, won in the COVID rebound and won over the last four years, the period the site itself calls “mega-cap concentration”. Equal-weight won the other four, and made most of its lead between 2000 and 2007: in the bust and in the bull market that followed.

That is why any comparison depends on where it starts and where it ends. Someone who looks only at the last four years will see the opposite of someone who looks at 2000 to 2007, and both are true.

The site publishes twenty factor portfolios — five rules in four sizes. Held in equal parts, 13 of the 20 beat equal-weight over this window, and 17 of the 20 returned more than SPY; but 15 of the 20 fell deeper than SPY. The 20-stock ones, one per rule:

A portfolio can end up ahead without having done anything but noise. The site puts the best of these through a test and publishes the result: multi-factor 20 beats equal-weight by 0.10% a month on average and wins 187 of 331 months, but with a block bootstrap the annual edge lies between −4.2 and +6.3 points at 90% confidence, and comes out negative in 37% of the resamples. Zero is inside. And over one-year windows it lost to the index 42% of the time; over five-year windows, 55%.

All twenty, with their curves and holdings, are in Model Portfolios; how each factor is computed and every test, in the methodology. To see what changes when you alter a single decision — the number of stocks, the weighting — use the simulator.

• The currency. Every curve on this site is computed with dollar prices and is not converted to euros. For someone investing in euros, the return also depends on what the exchange rate does, which can add or subtract.
• Currency conversion on every trade. A portfolio that rotates every month trades a lot: multi-factor 20 makes about 311 trades a year. Paying for currency conversion on each one would cost it 269 basis points a year. An index fund barely trades.
• Commissions and spread. With 100,000 and €0.30 per trade, and assuming a 3-basis-point spread, multi-factor 20 would net 11.03% against 10.18% for equal-weight after RSP's fee. With the spread our two estimators infer from each day's highs and lows — an upper bound — the net falls to between 8.32% and 8.94%: below the index. Costs can eat the whole edge.
• Taxes. A portfolio rebalanced monthly realizes gains every month; an index fund defers them. The vehicle matters too: in Spain, for instance, the tax-free transfer regime between investment funds does not, in general, apply to exchange-traded funds. Taxes depend on your case, and no figure on this site includes them.

• How much of a fall could you sit through? All four rows of the table lost more than half in 2008-2009.
• How long could you stand trailing the index? By period, multi-factor 20 went up to 66 months in a row without regaining its best point against SPY in the cap-weighted version, and 314 against equal-weight in the equal-weight version.
• How many trades do you want to make? An index fund asks for none; a factor portfolio, hundreds a year.
• What happens with taxes and currency in your case? No website that does not know your situation can say, and this one does not know it.

If you already hold stocks and want to see how they did against buying the index and sitting still, My Portfolio measures them with the same metrics, without telling you what to do with them.

• Except SPY, they are calculations and simulations. Nobody traded these portfolios in 1998: it is the rule applied month by month with the data that existed at each date.
• No costs, no taxes, no currency. Real funds carry their fee inside the price, so in the comparison with SPY the asymmetry slightly favors the simulated portfolios.
• Survivorship bias nearly closed. The history includes companies that left the index; the price is missing for around 0.1% of membership slots.
• One window and one market. 27.6 years of the 500 largest US companies. Another start date, another country or another era would give other numbers.