How to Read a Correlation Matrix (and Heatmap)
Updated 2026-06-11 · 5 min read
When you compare more than two ETFs, a single number isn't enough — you want every pair at once. A correlation matrix puts all of those values into one grid, and a heatmap colors them so patterns jump out instantly. This guide shows you how to read both.
What's in the grid
A correlation matrix is a square table with the same ETFs listed across the top and down the side. The cell where row A meets column B holds the correlation between A and B. Two things are always true: the diagonal is all 1s (every fund is perfectly correlated with itself), and the matrix is symmetric — the value for A-vs-B equals B-vs-A, so the top-right triangle mirrors the bottom-left.
Reading the heatmap colors
The heatmap maps each value to a color so you don't have to scan numbers. In this tool, green marks strong positive correlation (near +1), grey marks values near 0, and red marks negative correlation (toward −1). A block of green across several funds means they all move together — a concentration warning. Patches of red or grey are the diversifiers: holdings that behave differently from the rest.
What to look for
Scan for three things. First, clusters: groups of funds that are all green with each other usually share a driver (e.g. all broad A-share equity). Second, outliers: a row that's grey or red against everything else is your best diversifier. Third, the weakest links: the lowest values in the grid point to the pairs that combine best for a steadier portfolio. A typical China basket shows a green equity cluster, with the bond and gold rows standing out in grey or red.
Mind the sample window
Every value in the matrix is computed over the same lookback period you selected, using daily returns. A short window (1 month) reacts fast but is noisy; a long window (3 years) is steadier but can hide a recent regime change. If a relationship matters to you, check it over more than one window. And remember the values describe the past — they're an informed guide, not a forecast.
FAQ
Why is the diagonal always 1?
Each diagonal cell compares a fund with itself, and any series is perfectly correlated with itself, so the value is always exactly 1.
Why does the matrix look symmetric?
Correlation has no direction — the correlation of A with B is identical to B with A. So the grid mirrors across the diagonal, and you only really need to read one triangle of it.
How many ETFs can I put in one matrix?
The tool compares up to eight ETFs at once. Beyond that the grid gets hard to read; for larger studies, compare a few baskets and look at the standout rows in each.