Why Market-Cap Weighting Is Not Diversification
A cap-weighted index spreads your capital across hundreds of names while concentrating your risk in whatever the last cycle rewarded — and those are two different things.
Owning five hundred names feels like the definition of diversification, and for most investors the reasoning stops there. It shouldn't, because the argument contains a substitution error: it treats the number of holdings as a measure of the number of bets. A cap-weighted index maximises the first while quietly minimising the second, and the gap between them is not a technicality — it is the difference between how the portfolio looks on a statement and how it behaves in a drawdown.
Here is the framing we use: cap weighting diversifies your capital; it does not diversify your risk. Capital diversification is an accounting fact — your dollars are spread across many line items. Risk diversification is a structural claim — that the sources of return behind those line items are meaningfully independent of one another. The first is what the fund fact sheet shows you. The second is what determines whether the portfolio has one large failure mode or many small ones. Cap weighting guarantees the first by construction and makes no promise at all about the second, because the weighting rule was never designed to. It was designed to represent the market, and representing the market means inheriting the market's concentration, whatever that happens to be.
Weight is a record of past appreciation
The mechanism starts with what an index weight actually is. A company's weight is its price times its share count, divided by the total. Price is the compounded record of every past return. So a cap weight is not a neutral measurement of a company's importance — it is a running tally of relative performance. Whatever outperformed over the last cycle occupies more of the index at the start of the next one, automatically, with no trade and no decision. This is the property people praise as self-rebalancing, and the praise is deserved as far as it goes: the index never has to transact to stay true to itself, which is why it is cheap and tax-efficient. But the same property has a second name. A portfolio whose weights grow with relative strength and shrink with relative weakness is, structurally, a momentum position. The index buys nothing, yet it always holds more of whatever just won.
That would be tolerable if winners were randomly distributed across unrelated businesses. They are not, and this is the step where name concentration becomes something worse. The companies that dominate a given era tend to have won for the same reason — a shared business model, a shared macro tailwind, a shared position in whatever technology or credit cycle defined the period. So the index's top weights are not merely large; they are large together, and they are exposed together. Concentration in names is the visible symptom. Concentration in a single return driver is the actual condition.
Three layers of why
The last step deserves dwelling on, because it is where this stops being an argument about indexes and becomes an argument about your portfolio. A common structure for a substantial private portfolio is a cap-weighted core plus a sleeve of individually selected stocks. The selected stocks are usually large, well-covered, high-quality companies — which is to say, disproportionately the same names that dominate the core. The investor believes they hold a diversified base plus some considered bets. What they actually hold is a leveraged version of the index's own concentration. Sterling's Embedded Intelligence treats this as the first check on any portfolio with an index core: measure the overlap between the active sleeve and the top of the index before crediting the sleeve with adding any diversification at all. Sterling's rule of thumb is blunt — until you have measured risk contributions, you do not know how many bets you own, only how many tickers.
The illustration: concentration keeps changing costumes
Treated strictly as history, the pattern is one of the better-documented regularities in markets. In the late 1960s and early 1970s, a group of premier growth companies — remembered as the Nifty Fifty — grew to dominate institutional portfolios and index weight until their de-rating in the mid-1970s bear market. At the end of the 1980s, Japanese equities had compounded to the point where Japan represented an enormous share of global cap-weighted benchmarks — meaning a global index investor held a concentrated Japan position at precisely the moment it was most dangerous, not through any decision, but through arithmetic. At the peak of the late-1990s cycle, technology and telecom names had swollen into a historically large share of the major US indexes; in the mid-2000s, financials had; more recently, a handful of mega-cap technology platforms have.
The lesson is not that any particular group was overvalued — sometimes the dominant group went on to justify its weight, sometimes it did not. The lesson is structural: the index is always maximally exposed to the previous cycle's winner at the moment the cycle turns, because that is what the weighting rule does. Each episode felt different from inside. Each was the same mechanism wearing a different costume. An investor who holds the index without knowing this holds a position whose central risk changes identity every decade while its structure never changes at all.
The strongest case against this framework
The serious objection comes from the arithmetic of active management, and it deserves a fair statement. The cap-weighted portfolio is the only portfolio every investor can hold simultaneously; it is macro-consistent in a way no alternative weighting is. Every deviation from cap weight is a zero-sum trade against another investor, and after costs, the average deviator must underperform the index they deviated from. Alternative weighting schemes that claim to fix concentration — equal weight, fundamental weight — are simply repackaged active bets, usually a tilt toward smaller and cheaper companies, with higher turnover and higher cost. On this view, calling cap weighting undiversified is a rhetorical trick to sell you something more expensive.
We accept nearly all of this and reject the conclusion, because the objection answers a different question. The arithmetic establishes that cap weighting is the right average portfolio — that deviating from it carries no free expected return. It says nothing about whether the portfolio is diversified, because expected return and risk structure are different properties. A portfolio can be impossible to beat on average and still carry one dominant failure mode. The framework here does not tell you to abandon the index; for many investors, on cost and tax grounds alone, holding it is entirely defensible. It tells you to stop describing the index as diversified and start describing it as what it is: the market's current concentration, held at low cost. Once described correctly, concentration becomes a decision — you can accept it knowingly, hedge it, or offset it elsewhere in the portfolio — rather than an accident you discover during the drawdown.
What this changes about how you read your own portfolio
The practical shift is from counting holdings to counting bets. The working questions are: what share of the portfolio's total risk — not weight, risk — comes from the top handful of positions once the index core is looked through? How correlated are those top risk contributors with one another, and is there a single macro variable, such as the rate path or a single technology cycle, that most of them load on? How much of the active sleeve duplicates the top of the core? A portfolio of five hundred names that fails these questions is one bet with excellent paperwork.
How to apply this framework
- Measure risk contributions, not weights. Look through the index core and rank positions by contribution to portfolio variance. If a small group of correlated names dominates the risk, the portfolio's effective bet count is small regardless of the holdings count — and that is the number to manage.
- Check the overlap between your active sleeve and your index core. Individually selected large-cap positions frequently duplicate the index's top weights. Where they do, the sleeve is adding concentration, not diversification, and should be sized as leverage on the core rather than as an independent position.
- Identify the shared driver at the top of the index. Ask what single factor — a rate regime, a technology cycle, a sector's economics — the current top weights load on together. That factor, not any single company, is the index's real central risk, and it is what any offsetting position should be built against.
- Treat concentration as a standing decision, not a discovery. Holding the index is defensible; holding it while believing it is fully diversified is not. Revisit the decision when index concentration is historically elevated, because that is when the gap between capital diversification and risk diversification is at its widest.