Correlation in investing: what it means and how to read it
Correlation measures how closely two assets' returns move together, on a scale from −1 (opposite) through 0 (unrelated) to +1 (in lockstep).
What correlation tells you
Correlation is a statistic that summarises how two series move relative to each other. A coefficient of +1 means they move perfectly in step; 0 means there is no consistent linear relationship; −1 means one rises exactly when the other falls. Most pairs of assets sit somewhere in between: two large banks might show 0.8, while stocks and gold have often been close to zero.
For investors, correlation matters because it determines how much holding several assets actually reduces risk. Owning ten assets that are all highly correlated is not much safer than owning one.
How it is calculated
The usual measure is the Pearson correlation coefficient of the two assets' returns over the same periods:
ρ(X, Y) = Cov(X, Y) ÷ (σX × σY) R² = ρ² (share of variance explained, simple linear fit)
Always use returns — daily or weekly percentage changes — rather than price levels. Two unrelated assets that both happen to trend upward will show highly correlated prices simply because both lines rise, even if their returns are unrelated. The correlation matrix on the site works from returns over the window you choose.
Reading it — a worked example
Suppose Bitcoin and Ethereum have a correlation of 0.85 over the past year, and Bitcoin and gold 0.10. Adding Ethereum to a Bitcoin holding diversifies very little: on most days they move together. Adding gold changes the portfolio's behaviour far more.
Squaring the coefficient gives a feel for its strength: a correlation of 0.5 means only about 25% (0.5²) of one asset's variance lines up with the other's, so 0.5 is weaker than it sounds. A correlation of 0.9 corresponds to 81%.
Limitations and common mistakes
Correlation is not causation, and it only captures linear relationships. It is also unstable: correlations measured in calm periods can jump towards 1 in a crash, when investors sell everything at once. Results depend on the window and the data frequency, so a three-month figure can differ sharply from a five-year one.
Finally, a low correlation doesn't make an asset safe on its own — it can still be highly volatile. Correlation tells you about the relationship, not about the size of the moves.
Frequently asked questions
What is a good correlation for diversification?
The lower, the better for diversification: correlations below about 0.3, or negative ones, add the most risk reduction. Above 0.8, the assets largely move together.
What does a negative correlation mean?
That the two assets tend to move in opposite directions. Perfect negative correlation (−1) is rare; moderately negative correlations can act as a cushion, as government bonds have at times done for stocks.
Why calculate correlation on returns rather than prices?
Because two prices that both trend upward look correlated even when their day-to-day moves are unrelated. Returns strip out the trend and show whether the assets really move together.