HomeLearn
News & Articles
Market
My Account
Tools
AboutNewsletter☕ Buy me a coffee
LearnManage Your Risk › Module 4

Module 4 · Correlation & the Limits of Diversification Core

Diversification is the one genuinely free lunch in investing — and it is also the most over-claimed idea in personal finance. This module is about the difference between owning many things and owning many different things.

~12 min read · Not started

By the end of this module you'll be able to

  • Interpret a correlation coefficient and explain what it does to combined portfolio volatility.
  • Describe the shape of the diversification curve and where the returns to adding holdings flatten out.
  • Distinguish diversifiable idiosyncratic risk from the systematic risk that no number of holdings removes.
  • Identify fake diversification — many tickers exposed to one underlying driver.
  • Explain why correlations converge in a crisis, and what still diversifies when they do.

What correlation measures

Correlation (ρ) measures how two things move relative to one another, on a scale from −1 to +1.

Two points that are constantly misunderstood. First, correlation says nothing about magnitude: two assets can be correlated at 0.9 while one moves three times as far. Second, correlation is not causation and, more practically, it is not stable — a correlation estimated over the last three years is a description of those three years, and the relationships that matter most tend to change exactly when conditions do.

The important number in practice is not −1. Genuine negative correlation between growth assets is rare and usually expensive to obtain. Anything below +1 provides a benefit, and most of what diversification achieves in a real portfolio comes from combining assets correlated at 0.3 to 0.7, not from finding true opposites.

What correlation does to portfolio risk

For two assets, portfolio volatility is:

σp = √( w1²σ1² + w2²σ2² + 2w1w2ρσ1σ2 )

You do not need to compute this by hand, but you should read what it says. The first two terms are each asset’s own contribution. The third term — the one containing ρ — is the interaction, and it is the only place diversification lives. Drive ρ down and that term shrinks; make it negative and the term becomes a subtraction.

A worked case makes the effect concrete. Take two assets each with 20% volatility, held 50/50:

CorrelationPortfolio volatilityReduction vs holding one
ρ = 1.020.0%None
ρ = 0.718.4%8%
ρ = 0.517.3%13%
ρ = 0.014.1%29%
ρ = −0.510.0%50%
ρ = −1.00.0%100%

The row worth staring at is ρ = 0.0: combining two equally risky assets that are merely unrelated cuts volatility by 29% without reducing expected return at all. That is the free lunch, and it is why the phrase gets used. Note also that the benefit is not linear — going from 1.0 to 0.7 buys very little; going from 0.5 to 0.0 buys a lot.

How many holdings actually help

Adding a holding to a portfolio removes some of the risk that is specific to the names you already own. But the benefit falls away quickly, because each new stock is a smaller share of the total and because it brings its own correlation with everything already there.

Portfolio risk falls steeply with the first holdings and then flattens A curve showing portfolio volatility falling sharply from one holding to about fifteen, then flattening toward a floor made of systematic market risk that no amount of diversification removes. systematic risk floor — cannot be diversified away 1 stock ~10 ~20–30: most of the benefit is already here high low Number of holdings → Going from 1 to 10 holdings does far more than going from 30 to 100.
Diversification has sharply diminishing returns. Roughly 20–30 genuinely different holdings captures most of the available benefit; the remaining risk is the market itself.

Two honest caveats on that “20–30” figure, which gets quoted as though it were a law of nature. It assumes the holdings are genuinely different — thirty banks is not thirty holdings in any sense that matters. And it describes the average reduction in volatility, not protection from a single catastrophic name. A 30-stock portfolio still loses 3.3% if one holding goes to zero at equal weight.

The corollary for most people is not to build a 30-stock portfolio at all. A single broad index fund provides more diversification than any hand-assembled list, instantly and at almost no cost. Individual holdings are for when you have a specific reason for a specific company — and Module 6 covers how much of the portfolio those reasons deserve.

Fake diversification

The most common failure is not a lack of holdings; it is holding many tickers that all depend on the same thing. Every one of these is a single bet wearing a diversified costume:

The test to runList your holdings and, beside each, write the one thing that would have to happen for it to fall 40%. If the same sentence appears three times, you own one position in three pieces. Count exposures, not tickers.

Correlations in a crisis

The uncomfortable fact about diversification is that it works best when you need it least. In ordinary conditions, assets move for their own reasons and correlations are moderate. In a genuine panic, investors sell what they can sell to raise cash, sell everything indiscriminately, and dispersion collapses. Correlations across equities converge toward 1.0, and much of the protection you measured in calm markets is simply not there.

In March 2020, equities, corporate bonds, REITs, gold and commodities all fell together for a stretch. In 2022, the classic 60/40 portfolio failed in the way it was least expected to: stocks and bonds fell together, because a single driver — rapidly rising interest rates — was repricing both at once.

The lesson is not that diversification is useless. It is that diversification is a tool for managing the ordinary distribution of outcomes and a weak tool for managing the tail. For the tail, the effective defences are different in kind: not being leveraged (Module 8), holding enough cash that you are never a forced seller, and having written rules that keep you from selling at the bottom (Module 9).

What still diversifies

Ranked roughly by how reliably they have provided genuinely different behaviour:

  1. Cash and short-term government bills. Boring, unglamorous, and the only asset that reliably does not fall in an equity crisis. Cash is not a drag on the portfolio; it is the thing that lets you hold the rest of it.
  2. High-quality government bonds. A good equity diversifier for most of the last four decades — but 2022 demonstrated that in an inflation shock they can fall alongside stocks. Reliable against a growth scare, unreliable against an inflation scare.
  3. Geography. Meaningfully helpful, and less so than it used to be as economies integrated. Still, a Canadian investor holding only Canada carries an enormous and entirely avoidable concentration.
  4. Sector and factor spread. Real, moderate, and the piece most people already have if they own a broad index.
  5. Time. The most underrated diversifier of all. Buying at different times across a cycle removes the risk of a single terrible entry point — which is a genuine risk that no amount of cross-sectional diversification addresses.
💡 Check your own overlap: import your holdings into the Portfolio Tracker to see your sector weights, and read risk & diversification basics if you want the gentler version first.

Educational purposes only; not financial advice. Historical figures are illustrative and past drawdowns are not a forecast of future ones. Always do your own research and consult a licensed advisor.