2. Weighting by capitalization: what you are really buying

Level 11: The structure of the index

You now know who picks the members. Now for the second half of the question, which matters just as much and is examined far less: once the 500 have been picked, how much of each do you buy? The S&P 500's answer is by free-float-adjusted capitalization: each company weighs what its listed capital in circulation is worth, divided by the sum of all of them.

What that rule means in practice

If one company is worth twenty times another, you hold twenty times as much of it. Put less kindly but just as accurately: you hold most of whatever has risen most, because rising is literally how a company gains weight in the index. There is no opinion behind it. It is arithmetic.

Out of that comes a criticism you will hear a lot: 'capitalization indexing buys dear.' It is a half-truth and it is worth dismantling carefully. The index does not buy when something rises. Precisely because it weights by capitalization, when prices move nothing has to be traded: the weight readjusts on its own. If a position rises 40% and the others do not move, its percentage in your portfolio rises exactly as much as it should without anyone executing a single order. What does happen is that your exposure drifts toward whatever has risen, and that drift is not free.

The virtue almost nobody values enough

That property of requiring no trades is enormous, and it is the deep reason capitalization weighting ate the world:

• Minimal turnover. You only trade when a company's composition or capital changes, not when its price changes. Fewer trades, less cost, less market impact.
• Practically unlimited capacity. You can put gigantic sums in without overwhelming the strategy, because the size of each position is proportional to the size of that share's market.
• Aggregate coherence. It is the only weighting every investor in the world can hold at the same time. Any other requires somebody to take the other side.
• Honesty of costs. It is the reason an index fund can charge minuscule fees sustainably.

This is exactly the argument John Bogle made for decades, and it was not an argument about being right: it was an argument about not paying. The average return before costs is distributed by the market; what you can genuinely control is how much is lost along the way.

The risk: concentration grows on its own

The flip side is that the same property that avoids trading lets the portfolio concentrate unchecked. Nobody corrects anything, so if a few holdings rise far more than the rest for years, you end up with an index that looks less and less like what its name suggests.

The magnitudes are eloquent. In the mid-2010s, the ten largest constituents weighed around 18-20% of the index. In 2024 that weight passed 35% and came close to 40%, the highest level in several decades. And there is a derived figure that is more frightening still: the effective number of holdings, which corrects the count for actual weights rather than counting names, falls to a few dozen. You hold 500 lines and a diversification equivalent to considerably fewer.

The second consequence is about where the result comes from. In years such as 2023, a very short handful of companies explained on the order of two thirds of the index's rise. That means the return appearing on your statement is not that of 'five hundred American companies': it is above all that of a handful of them, wrapped in four hundred and ninety-odd others.

Comparing with equal weighting

The natural contrast is the same universe of companies with equal weights: each around 0.2%. Put like that it sounds more sensible, and sometimes it is, but it is worth seeing the full trade-off:

• Equal weighting forces you to trade every quarter, because the weights drift on their own. It trims what has risen and adds to what has fallen. That has a cost and it does not always pay.
• Its implicit exposure tilts toward smaller and less glamorous companies within the index, so it inherits the behavior of those tilts, for better and for worse.
• Its capacity is lower, because it requires buying as much of number 480 as of number 3.
• In narrow market periods, where a few giants call the tune, it falls a long way behind. In 2023, with the normal index up on the order of 26%, the equal-weighted version came in close to 14%.

Neither weighting is 'correct'. They are two different bets: one says the market already knows what everything is worth and that the cheap thing to do is not to trade; the other says size should not determine exposure and accepts paying turnover to correct it. Choosing without knowing which one you are choosing is the only thing that is clearly wrong.

Real case: March 2000 and the arithmetic of concentration

At the March 2000 peak, the technology sector came to weigh around a third of the S&P 500, after years in which exactly those holdings had been the ones that rose most. Nobody took the decision to have a third of the portfolio in technology: capitalization weighting did it on its own, without a single order.

What came next is the lesson. Between 2000 and 2002 the index lost on the order of 40% cumulatively, dragged down above all by that concentration, and it did not recover its previous high until 2007. The equal-weighted version of the same universe, with the same 500 companies, gave up a fraction of that over the same period, because its exposure to the holdings that had inflated the weight was far smaller.

Hold on to the idea without turning it into a prediction: capitalization weighting is not telling you those holdings are going to fall. It is telling you that, if they do fall, your portfolio will feel it far more than the word 'five hundred' suggests. This is education, not a recommendation about any security or about any way of weighting.