How Many Stocks Do You Need to Diversify?

If the holdings are genuinely independent of one another, most of the available risk reduction arrives by roughly six to eight positions, and additions beyond that contribute very little. If the holdings are not independent, there is no number that works, because the eleventh position that shares a driver with the other ten is adding commission rather than diversification.

That distinction, between counting tickers and counting drivers, is the whole of the answer. The arithmetic behind the first half is simple enough to do by hand, and worth doing once.

Why the curve flattens

Take a simplified case: several positions, each with the same volatility, each genuinely independent of the others. Combine them in equal weight and the spread of outcomes for the portfolio falls with the square root of the number of positions, not in proportion to it.

Work through what that means.

• One position to four: risk falls to about half.
• Four to sixteen: halves again.
• Sixteen to sixty-four: halves again.

Each halving costs four times as many positions as the last. The first few additions do almost all the work, and by the time you are adding a ninth or tenth independent holding, the improvement is small enough that it is swamped by the costs of holding it: commission, spread, and the attention you no longer have to give the other nine.

That is the honest case for diversification and the honest case against overdoing it, and both come from the same line of arithmetic.

The assumption that breaks it

Every number above depends on the positions being independent. Real holdings rarely are.

Six semiconductor firms are one position with six tickers. So, less obviously, are a homebuilder, a mortgage lender, a furniture retailer and a cement producer: the shared driver is the rate cycle, and it does not care that the companies sit in four different sectors on a screener.

The practical test is not statistical. Write one sentence for each holding saying why it should rise. Positions whose sentences restate each other belong to the same bet. Our guide to correlation and portfolio heat works through what that does to a book that looks spread, and the portfolio correlation checklist is the version to run before adding anything.

Counting drivers instead

Rework the question as drivers and the answer becomes more useful.

Six to eight independent drivers captures most of the benefit. Those might be a rate story, a commodity story, a regulatory story, a demand cycle and a company-specific catalyst or two. How many tickers express them matters far less than how many there are.

This also reframes what a long watchlist is for. Twenty candidate positions across four drivers does not give you a choice of twenty bets. It gives you four, with five ways to express each, and the real decision is which expression is cleanest rather than how many to hold.

Three or four holdings across three or four genuinely separate drivers is better spread than twenty across two, and it is a great deal easier to follow closely enough to notice when one of the stories changes.

What more positions actually cost

Each addition carries costs that do not fall with the square root of anything.

Attention. Every holding needs its thesis checked, its catalyst dates known and its exit defined. Past a certain count, positions start being held because they are already open rather than because they still make sense.

Execution. Spread and commission are paid per position. On small accounts, spreading the same capital across twenty holdings can make the per-position cost a meaningful fraction of the expected gain, which is covered in trading costs.

Dilution of the good ideas. A twentieth position sized the same as the first means the best idea in the book carries five per cent of it. Whether that is sensible depends entirely on whether the ideas are actually of similar quality, and they usually are not.

The number that matters more

For a trader rather than a long-term investor, position count is the second question. The first is total risk: what a bad week does to the account when several holdings move against you at once.

Eight positions each risking one per cent across eight independent drivers is an eight per cent worst case that is unlikely to arrive all at once. Eight positions each risking one per cent across two drivers is a potential eight per cent that can arrive on a single afternoon. Identical position count, identical per-trade risk, completely different exposure.

Setting a cap per driver and splitting risk within it, rather than assigning full risk to each ticker, is what makes the count meaningful. The position size calculator handles the split, and expectancy and sample size covers why keeping that total survivable matters more to long run results than picking well does.

Equal weight is doing quiet work in that arithmetic

The square root relationship assumes the positions are equally weighted. Real books rarely are, and the gap matters.

If one holding is four times the size of the others, the portfolio’s behaviour is dominated by that holding regardless of how many names sit beside it. The count says eight positions; the risk says roughly one position with seven small hedges. Concentration arrives through weight as easily as through correlation, and it is easier to miss because every line on the screen looks like a separate decision.

This is also how winners create concentration without anyone deciding to concentrate. A position that has tripled now carries three times its original share of the account, and if it sits beside another winner in the same cluster, the book has drifted toward one story while the position count stayed flat.

The practical check is to rank holdings by current value rather than by entry size, then ask whether the top two or three would be an acceptable loss if their shared driver turned. If the answer is no, the portfolio is more concentrated than its length suggests, and the fix is trimming rather than adding.

A workable answer

Hold as many positions as you have genuinely separate reasons to hold, up to around six to eight, and size them so the total loss from any one driver is something you would accept knowingly.

Below three or four drivers, single-story risk dominates and one piece of news can take an uncomfortable share of the account. Above eight, you are paying costs and attention for a benefit the arithmetic says is close to exhausted — and in practice most portfolios that look like twelve or twenty positions are running on three or four drivers anyway.