The risk you get paid for

A friend told me he doesn't need fifty stocks.

Six names. All researched, all high conviction. And I get the instinct. Fifty stocks can feel like admitting you don't know which ones will work.

But here is the part the confidence hides. Every stock you hold carries two completely different risks, and only one of them ever pays you.

The first is the company's own risk. The factory fire, the CEO who quits, the product that flops, the fraud nobody saw coming. Roughly 80 percent of a single stock's volatility is this kind, specific to that one name. The second is market risk, the tide that moves every boat, the thing you cannot escape no matter how many stocks you own.

Here is what took me years to really absorb. The market hands you an expected return for holding the second risk. It pays you nothing for the first.

And the first one is free to get rid of. Watch what happens as you add names to an equal-weight basket. One stock sits near 40 percent volatility. Five stocks, 24. Ten, 21. Thirty, 19. Then it just stops, resting on a floor around 18 that is the market itself. One name's bad quarter cancels another's good one, and the company-specific risk quietly averages away. That drop from 40 to 19 cost nothing in expected return. You shed it for free.

So my friend's six researched names are carrying a pile of risk the market was never going to reward. Not because the research is bad. Because concentration keeps the unpaid risk on the books.

Oldest idea in portfolio theory, and still the most ignored. Diversification is not caution. It is refusing to hold the risk that was never going to pay.

Where the 40, 24 and 19 come from

Those numbers are not magic. They fall out of one short formula, and once you see it you can redo the whole curve on paper.

Take an equal weight basket of N stocks. Give each stock the same volatility, call it s, and the same average correlation with every other stock, call it r. The basket's volatility is s times the square root of (r plus (1 minus r) divided by N). With s at 40 percent and r at 0.20, one stock gives 40. Five give 40 times the square root of 0.36, which is 24. Ten give about 21.2. Thirty give about 19.0. And as N gets very large, the second term goes to zero and you are left with 40 times the square root of 0.20, about 17.9. That is the floor.

The same formula explains the 80 percent. For a single stock, the share of its variance that is company specific is 1 minus r. With r at 0.20, that is 80 percent.

Six names against thirty, in rupees

Put Rs 6,00,000 into six stocks, Rs 1,00,000 each. On the illustration's numbers the basket runs at about 23.1 percent volatility. A one standard deviation move, which on a normal curve is matched or exceeded roughly one year in three, is about Rs 1,38,600.

Now spread the same Rs 6,00,000 across thirty names, Rs 20,000 each. Volatility drops to about 19.0 percent. The same one standard deviation move is about Rs 1,14,300. Rs 24,300 less swing, and the expected return has not changed, because each stock was assumed to carry the same expected return in both baskets.

Run your own numbers below. Put in 50 names and the basket is already within a whisker of the floor. Push the correlation up to 0.5 and watch the floor rise: when stocks move together, there is less to diversify away.

At six names, 40 percent of the basket's variance is still company specific. That is the unpaid pile. At thirty it is about 12 percent.

What the floor looked like this year

The floor is not a fixed 18. It is whatever the market itself is doing. Over the twelve months to 30 Sep 2026 I measured the Nifty 50 from its daily closes.

That is the risk no amount of diversification inside Indian equity would have removed. A thirty stock basket of Nifty names would have felt most of that 15.2 percent fall. What it would not have felt is the extra damage from any one of its names having a terrible year on its own. That second kind of damage is exactly what six names keep.

The free Learn module on modern portfolio theory builds this from the ground up, with correlation, the efficient frontier and a quiz. The Sharpe and volatility calculator lets you test the return you get per unit of the risk that is left. And the post on single stock skew shows why a few names also make the rare big winners easier to miss.