01 · Briefing · what happened

Diversification: the one free lunch in investing

Personal Money 4 min 16 sources

Spreading money across things that don't move together cuts risk without cutting your expected return - the rare gain economics says shouldn't exist.

14%

combined swing

two unrelated funds that each swing 20% alone

8%

return, unchanged

the same at every level of correlation - that's the free part

20-30

different companies

captures most of the risk spreading can remove

~5%

per-holding guardrail

a rough cap so no single bet can sink you

At a glance

Diversification is the one 'free lunch' in finance: spread across things that don't move together and risk falls while expected return stays put.
The engine is correlation - it works only when holdings zig and zag at different times, not just when you own a lot of them.
Two funds that each swing 20% but move unrelated combine into a ride that swings only about 14% - for the same 8% average return.
Roughly 20-30 different companies capture most of the benefit; an index fund buys a whole market at once for one low fee.
Overdo it and you get 'diworsification' - overlapping funds that cancel out while you pay fees on all of them.
The catch: in a crash, normally-unrelated things fall together, and the protection thins right when you need it most.

Forces in play

Holdings that differ Steady→

owning things that fall at different times is what actually smooths the ride

Market concentration High↑

US markets lean on a few giant names more than any time since the 1930s, so 'spread out' can be an illusion

Crash correlation Building↑

in a severe sell-off things fall together and the protection thins

In play Harry Markowitz — Nobel economist credited with the 'only free lunch' idea Peter Lynch — fund manager who coined 'diworsification' for overdoing it Warren Buffett — called diversification 'protection against ignorance'

Where this points

Watch whether market gains keep broadening beyond a few giant stocks - if they narrow again, a 'diversified' portfolio may quietly be riding the same handful of names.

Full briefing

Economists have a saying: there is no such thing as a free lunch. Every gain has a price. Diversification is the famous exception - the closest thing investing has to a genuinely free gain [2].

Spread your money across investments that don’t rise and fall in step, and the ride gets smoother without your expected return getting smaller [1][2]. Two funds that each swing wildly on their own, but zig and zag at different times, combine into something calmer than either alone [3]. You don’t pay for that calm in returns. That is why it is called a free lunch.

The trick is not owning many things. It is owning things that move differently [3][4]. Ten stocks that all crash together is not diversification - it is one bet in ten costumes.

How the smoothing works

Every investment carries two kinds of risk. One is specific to that company or sector - a factory fire, a failed product, a scandal. The other is market-wide - a recession, a rate shock, a crash - and hits almost everything at once [5].

Diversification only kills the first kind. Spread across enough different companies and one firm’s disaster gets diluted by the others still doing fine. The market-wide risk stays; no amount of spreading escapes a crash that drags down everything [5].

The engine is correlation - a measure of how closely two things move together, from +1 (locked in step) through 0 (unrelated) to -1 (perfect opposites) [4]. Say two funds each swing about 20% in a typical year, and each earns 8% on average. If they move in lockstep, holding both still swings the full 20% - you gained nothing. If they are unrelated, the blend swings only about 14%. If they moved as perfect opposites, the blend would barely swing at all - yet still earn 8% [3]. Same return, far less white-knuckle. That gap is the free lunch, and correlation is what sets its size.

How much is enough

You do not need hundreds of holdings. Classic studies found that a stock portfolio of roughly 20 to 30 well-chosen, different companies removes most of the risk that spreading can remove [6]. Beyond that, each new stock barely moves the needle.

Some argue even 30 is too few if they are all large US technology names that rise and fall together [7]. That is the real point again: what matters is that the holdings move differently, not that there are many of them.

The cheapest way most people get there is an index fund - a single fund that holds a whole market at once. Buy one tracking the S&P 500 and you own small slices of 500 companies for one low fee [14]. That is instant diversification, no stock-picking required.

When more becomes worse

You can overdo it. Legendary fund manager Peter Lynch coined a term for overdoing it: diworsification [9][8]. Pile on so many funds that they overlap, and they cancel out and drag performance while you pay fees on all of them. Owning five funds that all hold the same tech giants is not five bets; it is one bet, five times, minus the fees.

A rough guardrail some use: no single holding worth more than about 5% of the pot, so no one bet can sink you [13]. The aim is enough different things to be safe, not so many that you own the whole market twice.

The catch: crashes break it

Diversification’s weak spot shows up exactly when you want it most. In a severe sell-off, things that normally move differently start falling together - correlations rush toward +1, and the protection thins right when the storm hits [10]. The classic stock-and-bond mix that smoothed returns for decades has offered less shelter in recent shocks [10].

It still works over time. Diversification had its best year since 2009 in 2025, and kept paying off through 2026 as gains spread beyond a handful of giant stocks [11][9]. But US markets are now more concentrated in a few names than at any point since the 1930s [12]. So a portfolio that looks spread out can secretly ride on the same few companies [16].

Warren Buffett once called diversification “protection against ignorance” - unnecessary if you truly know what you are doing, sensible if you admit you don’t [15]. For almost everyone who is honest about that, spreading across things that move differently is the one gain finance hands out for free [16].

02 · Lesson · why it matters

The one gain that economics says shouldn't exist

Spread money across things that fall at different times, and the ride smooths without the reward shrinking - a bargain with no bill attached.

How it works

Every holding carries company-specific risk plus market-wide risk
Spreading dilutes the company-specific part; the market-wide part stays
How much it smooths depends on correlation - how differently things move
Unrelated or opposite movers cancel each other's swings
The average return survives; only the wild swings shrink

The twist

The free lunch isn't in owning many things - it's in owning things that fall at different times, so one bad bet never sinks the whole boat.

Where you've seen this

Farming

planting several crops so one blight doesn't wipe out the whole harvest

A career

skills that pay off in different conditions, so one industry's slump isn't your ruin

A supply chain

sourcing a part from several suppliers so one factory fire doesn't stop the line

The catch

It breaks in a crash: when everything falls together, correlations rush toward one and the protection thins exactly when you want it most.

Full lesson

The saying that has one exception

There is no such thing as a free lunch. Every gain costs something. Want more reward, take more risk. Want less risk, accept less reward. That trade sits under almost every money decision you will ever make.

Diversification is the one place the trade breaks. Spread your money across investments that do not rise and fall in step, and something odd happens. The wild swings shrink, but the average reward stays put. You get calmer without getting poorer. Economists, who distrust anything that sounds too good, call it the closest thing to a free lunch that finance offers.

Why “different” beats “many”

The instinct is to think diversification means owning a lot of things. It does not. It means owning things that move differently.

Picture two funds. Each has good years and brutal years, each swings hard on its own. But they swing at different times - when one is having a bad month, the other tends to have a decent one. Hold both, and their bad patches partly cancel. The blend is steadier than either piece.

Now picture ten funds that all rise and crash together. That is not ten bets. It is one bet wearing ten costumes. When the bad day comes, all ten fall at once and the “spreading” saves you nothing. The number of holdings was never the point. Whether they move together was.

It is not the same as insurance

This is worth pausing on, because it is easy to blur with a different idea: pooling. Insurance pools risk across people. A thousand strangers each pay a small premium, one of them has the house fire, and the pool covers it. The magic there is in the crowd - one person’s disaster is unpredictable, but a thousand people’s is nearly a fact.

Diversification is not that. There is no crowd of people. It is your money, spread across things that behave differently. Insurance makes one person’s rare catastrophe survivable by sharing it. Diversification makes your own fortunes smoother by never letting a single bet carry the whole load. Both are ways of not putting everything in one place - but one spreads across people, the other across bets, and the engines are not the same.

The trap of overdoing it

Because it feels safe, people push it too far. They buy five funds, then ten, thinking each new one adds protection. But if all ten quietly hold the same handful of giant companies, they are not spreading anything. They are buying the same bet again and again, paying a fee each time.

One fund manager gave this a name: diworsification. Past a point, more holdings do not add safety - they add overlap, cancel each other out, and cost you in fees. The goal was never maximum stuff. It was enough genuinely different things that no single one can sink you.

The catch that shows up at the worst moment

Here is the honest part. Diversification’s protection is weakest exactly when you most want it. On an ordinary day, your holdings move at their own paces and the smoothing works. But in a real panic - a crash, a shock that frightens everyone - people sell everything at once. Things that normally have nothing to do with each other start falling together. The very differences you were counting on collapse, and for a while almost everything drops as one.

So it is not a shield against catastrophe. It is a shield against the ordinary bad luck of any single bet going wrong. That is a smaller promise than “you cannot lose,” and it is the true one.

What the free lunch is really telling you

Step back and the lesson is bigger than money. Almost every hard thing you rely on rests on one question. Your income, your skills, the businesses in a town, the crops in a field - is each many separate bets, or one bet in disguise? A person with three income streams that all depend on the same employer has one income. A country that grows several crops that all fail in the same drought grows one crop.

The comfort of “I have spread things out” is often false, because we count holdings and forget to ask whether they move together. And the deepest part is that we can rarely see the hidden threads that tie our supposedly separate bets into one. The shared employer, the shared market, the shared bad day - they bind us without our noticing. Diversification is not a trick for beating that uncertainty. It is a humble admission of it. Since we cannot know which bet will fail, we make sure no single one carries everything. And we hold even that plan loosely, knowing a storm can tie it all together anyway.

03 · Lab · your turn

Two Funds, One Dial

Rehearse how correlation, not the number of holdings, is what smooths a portfolio's ride for the same return.

04 · Hope · carry this

There is at least one bargain the world hands out for free, and it rewards not cleverness but the honest admission that no single one of us can see the whole board.