Open the average retail portfolio and you will often find the same story: ten or fifteen positions, the owner convinced they are diversified because they hold more than one thing. But if those positions are Nvidia, AMD, Microsoft, Apple, Meta, Google, and a handful of other large-cap technology names, you do not own a diversified portfolio. You own one big bet on technology, split across ten tickers. When the sector sells off, they all fall together, and the diversification you thought you had disappears exactly when you need it.
Diversification is not about how many positions you hold. It is about how differently they behave. The tool that measures "how differently" is correlation, and understanding it changes how you build a portfolio and how you size risk across it.
What Correlation Actually Measures
Correlation measures how two assets move relative to each other, on a scale from -1 to +1. A correlation of +1 means they move in lockstep: when one rises 2%, the other rises 2%. A correlation of -1 means they move in perfect opposition: one rises 2%, the other falls 2%. A correlation of 0 means their movements are unrelated — knowing what one did today tells you nothing about the other.
You do not need the underlying math to use the concept. The intuition is what matters, and it is blunt: two positions with a correlation near +1 are effectively one larger position. If Nvidia and AMD move together 90% of the time, holding both is not two independent bets. It is roughly one bet in a slightly larger size. That is the whole idea, and most portfolios ignore it.
Reading a Correlation Value
| Correlation | What It Means | Effect on a Portfolio |
|---|---|---|
| +1.0 | Move identically | No diversification — effectively one position |
| +0.7 to +0.9 | Strongly together | Minimal diversification benefit |
| +0.3 to +0.7 | Loosely related | Partial diversification |
| 0.0 | Independent | Full diversification benefit |
| Negative | Move opposite | Positions hedge each other |
Diversification Comes From Low Correlation, Not Position Count
Here is where intuition fails most traders. Adding more positions only reduces risk if those positions are not highly correlated. Pile on ten names that all move together and you have barely reduced your risk at all — you have just spread the same bet across more line items.
The math behind this is a single clean formula. For a portfolio of N equally weighted positions with an average pairwise correlation of ρ (the Greek letter rho), the portfolio's volatility relative to a single position is:
Relative volatility = square root of ( 1/N + (1 - 1/N) × ρ )
The part that matters is ρ. As you add positions, the 1/N term shrinks toward zero, but the (1 - 1/N) × ρ term does not — it converges on ρ itself. That sets a floor on how much risk you can diversify away. At high correlation, that floor is high, and adding names past a handful stops helping almost entirely.
Portfolio Volatility as a Percentage of One Position's Risk
| Positions (N) | Same Sector (ρ = 0.8) | Diversified (ρ = 0.2) |
|---|---|---|
| 1 | 100% | 100% |
| 2 | 95% | 77% |
| 5 | 92% | 60% |
| 10 | 91% | 53% |
| 25 | 90% | 48% |
| Floor (∞) | 89% | 45% |
Look at the ρ = 0.8 column — the ten-tech-stocks case. Going from one position to twenty-five cuts your volatility by roughly 10%, and most of even that tiny gain is gone by the fifth position. The floor sits at 89% of single-position risk no matter how many names you add. Now look at ρ = 0.2, genuinely different assets: ten positions cut volatility roughly in half, and the floor drops to 45%. Same number of positions, completely different outcome. Correlation, not count, did the work.
Plotted against the number of positions, the two paths separate immediately and never reconverge:
Relative Portfolio Volatility vs Number of Positions
Crypto Shows This in the Extreme
Crypto is the cleanest live demonstration of correlation eating diversification, because the intra-asset correlation is notoriously high. Below are the actual pairwise correlations of daily returns for four of the most-traded assets — Bitcoin, Ethereum, Solana, and Chainlink — computed from roughly 179 daily returns over the six months to July 2026.
Daily-Return Correlation Matrix (6 months to July 2026)
| Asset | BTC | ETH | SOL | LINK |
|---|---|---|---|---|
| BTC | 1.00 | 0.91 | 0.88 | 0.89 |
| ETH | 0.91 | 1.00 | 0.90 | 0.93 |
| SOL | 0.88 | 0.90 | 1.00 | 0.92 |
| LINK | 0.89 | 0.93 | 0.92 | 1.00 |
Every pair sits between 0.88 and 0.93. Even the weakest relationship in the table, Bitcoin and Solana at 0.88, is far above the "strongly together" threshold. A trader holding all four of these feels diversified across four different projects with four different use cases. In risk terms they are holding roughly one position. When crypto has a bad week, it does not have a bad week in Bitcoin only — the whole complex moves as one. Holding more coins from the same market does almost nothing to change that. This is the ρ = 0.8 column made real, and if anything it understates it.
Correlation Is Not Static — It Spikes in a Crisis
The uncomfortable part is that correlation is not a fixed number. The values you measure in calm markets understate what happens in a panic. In a broad selloff, assets that normally drift apart suddenly plunge together as investors sell everything liquid to raise cash. Stocks, credit, commodities, and crypto that looked loosely related all week become tightly linked in a single afternoon.
In a selloff, the one thing that reliably goes up is correlation.
This is the cruel timing of it. Diversification is supposed to protect you in exactly the moments it stops working. A portfolio that looks well spread using normal-market correlations can behave like a single concentrated bet during the crash — the one time the protection actually mattered. Plan for crisis correlation, not the comfortable numbers you see when nothing is wrong.
What This Means for Position Sizing
Correlation quietly breaks the risk math most traders rely on. The 1% rule — risk no more than 1% of your account per trade — assumes each trade is an independent bet. Put on five positions at 1% each and you tell yourself you are risking 5% across five separate ideas. But if those five positions are highly correlated, they are not five ideas. They are one idea in five pieces, and a single adverse move can take all five stops out together. That is not five 1% bets. It is closer to one 5% bet.
The fix is to size correlated positions as a cluster. If five names all depend on the same driver, treat their combined risk as your unit of exposure and keep the total near your single-trade limit, not five times it. This matters even more with borrowed money: correlated leveraged positions amplify each other, so a coordinated move against a cluster can trigger stacked margin calls across the whole group at once.
How to Actually Diversify
Real diversification comes from holding things driven by different forces, not from collecting more tickers. Before adding a position, ask what actually moves it. Ten "different" stocks that all live or die on interest rates, or on AI capital spending, are one bet wearing ten costumes.
- Diversify across asset classes, not just tickers — equities, bonds, commodities, and cash respond to different pressures.
- Spread across sectors and geographies so a single sector shock or one country's policy does not sink everything at once.
- Check the shared driver. Even a broad index is a correlated basket: because the largest companies dominate the weighting, a position in the S&P 500 is not 500 independent bets but a heavily mega-cap-tech-weighted one.
- Size correlated groups as one position, and stress-test the portfolio assuming correlations jump toward 1 in a crisis.
Key Takeaways
Owning ten of anything is not a strategy. Owning ten things that move differently is. Correlation is the difference between the two, and it is measurable, so measure it before you assume you are protected.
Diversification is not about how many positions you hold. It is about how differently they behave when it matters most.
Disclaimer: This content is for educational purposes only and does not constitute financial advice. Trading involves substantial risk of loss. Past performance does not guarantee future results.
