The diversification that isn't there: how many companies do you really own?
3 min read by Opthest
Your brokerage account shows three lines: a world ETF, an S&P 500 ETF, maybe a tech fund. Three products, three names, the reassuring feeling of having spread your risk. But a portfolio is not the list of what you bought — it is the distribution of what you own. Look inside the ETFs and the three bets often turn out to be the same bet, written three times.
The number that misleads
“Over a thousand stocks in a single ETF” is true, but it tells half the story; the other half is how they weigh. As of the late-May 2026 factsheets, the top ten holdings of the MSCI World alone account for roughly 28% of the index, and the United States for about 72%. The S&P 500 is even more top-heavy: its top ten exceed 37%, against a thirty-year average of around 25%. Buying “the whole world” today mostly means buying the same ten companies — nearly all American, nearly all tech.
When two ETFs are (almost) the same ETF
A world ETF already contains the S&P 500: adding a dedicated one doesn’t widen the basket, it overweights what was already there. That is overlap, and it is measured stock by stock: a company’s effective weight in your portfolio is the sum, across your ETFs, of how much the ETF weighs for you times how much the company weighs inside the ETF. A textbook example with round numbers: 60% in an ETF where a stock weighs 5%, plus 40% in another where the same stock weighs 7%, makes a real exposure of 5.8% — to a single company, bought twice without noticing.
Look-through: seeing past the wrapper
Serious concentration analysis works like this: decompose each ETF into its underlying holdings, aggregate the effective weights per company, and only then read the real top 5, the sector concentration, the country concentration. That is look-through — seeing past the wrapper. The typical result is surprising: portfolios with five or six lines that, in transparency, hold one big American tech position instead of the diversification they thought they had.
The effective number of holdings
There is an elegant way to say all of this in one figure: the Herfindahl index (the sum of squared weights) and its inverse, the effective number of holdings. Four stocks at 25% each behave like four stocks: effective number 4. Four stocks weighted 70/10/10/10 have an effective number of 1.9: four lines on the account, fewer than two real bets. The same goes for indices: a thousand top-heavy stocks behave like a basket far smaller than its line count.
How to read it like an adult
Overlap, in itself, is not a mistake: overweighting the US or technology is a legitimate position — as long as it is a choice, not a surprise. The problem isn’t concentration; it is unseen concentration, the kind you discover in a drawdown, when all your “different” lines fall together because they were the same thing. Look at effective weights, not the number of products; at the look-through top 10, not the declared one; at aggregated sectors and countries, not per-ETF ones.
What this is NOT
There is no “right number” of ETFs and no magic overlap threshold, and this article doesn’t tell you what to buy or sell: it describes a measurement method, not a recommendation. The decision — and the concentration you knowingly choose to keep — remain yours.
Your portfolio is not what you bought: it is what you own. Until you look through the ETFs, the two can be very different things.