Diversification
Diversification means holding assets that do not fail for the same reason, so a single adverse event cannot damage all of them. It is measured by how differently the holdings behave rather than by how many there are, and correlations tend to converge in the falls it is meant to protect against.
Diversification is widely agreed with and rarely implemented, because the implementation requires owning something that is currently doing badly and continuing to own it. The principle is easy; the consequence is not.
How it works
Combining assets whose outcomes are not driven by the same thing reduces the range of what the portfolio can do, because a bad outcome in one is not automatically a bad outcome in another.
The mechanism is difference, not count. Two genuinely unrelated holdings diversify more than twenty related ones, and the arithmetic does not care how many lines are on the statement.
What it buys is a narrower range, not a higher expected return. That distinction is the source of most disappointment with it.
A worked example
Take a portfolio of twenty companies, each around 5% of the total.
If they all sit in one sector, the portfolio has one exposure. A regulatory change, a rate move, or a shift in that industry’s economics hits every position at once, and the twenty lines provide no protection at all.
Measured through beta, the effect is visible. A portfolio of 50% at a beta of 1.10, 30% at 0.85 and 20% at 1.40 has a portfolio beta of 1.0850 — so a 10% market fall implies a 10.85% portfolio fall, regardless of how many holdings produced those weights. The workings are in the portfolio beta calculator.
The count told you nothing. The weighted exposure told you everything.
The failure that matters
In ordinary conditions, holdings behave independently enough for the arithmetic to work. In a severe fall they converge — everything is sold at once because people need cash, not because the businesses became related.
So the protection is weakest in the event it exists for. That is not a reason to abandon it; it is a reason not to rely on measured correlations from calm periods when sizing a portfolio.
Measured on the S&P 500 since 1950, 93% of trading days ended below the previous record, the deepest fall was 56.8%, from 2007 to 2009, and the longest wait for a new record was 7.5 years, from 1973 to 1980 — and the index still went on to new records. The S&P 500 numbers come from every daily close since 1950, price only, through September 2026.
What it costs
A properly diversified portfolio always contains something that is doing badly. That is what diversification looks like from the inside, and it is why it gets abandoned — the laggard is visible every month and the protection is invisible until it is needed.
It also guarantees you will not have the best return available. Owning everything means owning the average, and the average is never the top of the table.
The willingness to hold the disappointing part is the entire discipline. Everything else is arithmetic.
Where the limits are
Adding holdings has sharply diminishing returns. Moving from one holding to ten removes most of the company-specific risk; moving from thirty to three hundred removes very little further, because what remains is market risk that no amount of spreading can touch.
Which means over-diversifying is a real thing, and it usually shows up as several funds that hold substantially the same companies at a higher combined fee.
Costs
Assembling diversification from individual positions costs a spread on each. Every round trip pays the bid-ask spread, and on a thinly traded holding that spread can be a real share of a normal day’s move.
A single broad index fund delivers more of it, for less, than a hand-assembled portfolio of twenty companies — which is the practical form the principle usually takes.
The original data
Of the 24,971 unique videos in the search study behind this site, 2 have a title about diversification, at a median of 14,310 views across 2 channels — and 50% of those titles are instruction-shaped. Asset allocation appears in 4 videos at 5,289 and sector funds in 2 at 1,453.
2 videos at a 14,310 median. The most universally agreed principle in investing has almost no coverage, which is consistent with everything else here: the operational subjects get made and the foundational ones do not.
The question: One holding has carried the portfolio for years. Concentrate?
The answer is that concentration is how the portfolio got here and it is also how it could leave. A holding that has grown to dominate has already concentrated the portfolio without a decision being made. Deliberately adding to it is choosing to run the same risk on purpose — which is a legitimate choice, and one that should be made knowingly rather than by drift.
When it fails
The characteristic failure is a portfolio that is diversified by holding count and concentrated by driver, and it looks correct on every statement. Twenty positions across four sectors can share one sensitivity — to rates, to a currency, to a single end market — and nothing in the portfolio view reveals it. The owner believes the risk is spread, sizes accordingly, and discovers the truth in the one week when every holding falls together for the same reason.
The second failure is abandoning it because something lagged. The laggard is the evidence it is working.
A third is over-diversifying into overlapping funds. Three funds holding the same companies is one fund at three fees.
A fourth is relying on calm-period correlations. They converge when it matters.
A fifth is treating it as a return strategy. It narrows outcomes; it does not raise them.
And a sixth is diversifying only within one asset class. Twenty equities is still equities.
Related
Asset allocation turns the principle into weights. Sector funds is the most common way it is accidentally undone. And international stocks is the axis most portfolios under-use.
The useful check is to ask what single piece of news would hurt every position at once. If you can name one, the portfolio is less diversified than the holding count implies — and one can usually be named, which is the uncomfortable part.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.