Correlation matrix: asset classes, sectors and countries

What this data shows

Fixed income and Real estate have the highest correlation (0.46). Fixed income and Broad commodities have the lowest (-0.42) across the covered asset classes.

Higher positive values indicate closer co-movement. Values near zero indicate a weak linear relationship; negative values indicate opposite co-movement. Low historical correlation does not guarantee protection in a downturn.

Correlation data

What is correlation?

Correlation measures the strength and direction of a linear relationship between two return series. Pearson correlation ranges from −1 to +1: +1 means perfectly aligned movements, zero means no linear relationship, and −1 means perfectly opposite movements. Zero correlation does not establish independence.

Each heatmap cell compares two peers within the selected universe—for example Japan with China, Financials with Consumer Discretionary, or equities with bonds. It does not measure every category against a single benchmark.

Methodology and data coverage

We use named US-listed ETF proxies rather than calculating a bespoke index from every underlying constituent. Country ETFs represent their respective investable equity markets; sector ETFs represent US equity sectors. Asset-class proxies represent specific exposures, not every instrument in the category. ETFs and mutual funds are wrappers, so they are not separate economic asset classes here.

Returns use provider-adjusted closing prices in USD. Country results therefore reflect the US-listed ETF’s trading session, fees and currency exposure, not local-market closing times or local-currency national indices. Adjusted prices incorporate the provider’s split and distribution adjustments; historical corrections can revise previous results.

The existing regional bulk-price refresh includes these proxies in its US-market run. Charts use the latest successfully stored completed closes; they are not live intraday prices. Monthly ranking snapshots remain separate. Missing or stale histories are excluded and disclosed above. Comparing today’s proxy selection historically does not recreate historical constituent membership.

Default sampling matches the platform: 3M, YTD, 1Y and 2Y use daily observations; 5Y uses weekly observations; MAX uses monthly observations and appears with meaningful history beyond five years. You can change sampling frequency independently. Weekly observations end on Friday and monthly observations at month-end; incomplete periods are excluded.

All displayed peers use a common observation window. Missing values are not filled with zero returns. We require at least 20 daily, 8 weekly or 6 monthly common return observations before annualising or publishing correlations. Short periods at monthly frequency can therefore be unavailable. MAX uses the history shared by eligible proxies.

Pearson correlation is the covariance of two aligned periodic return series divided by the product of their standard deviations. It is calculated from periodic returns, not prices or cumulative returns. A series with no return variation has an undefined correlation. Correlation can change with the period, sampling frequency and market conditions.

Source: EOD Historical Data adjusted ETF prices; US Treasury for the reference yield. Historical comparisons are descriptive and do not establish an efficient frontier or recommend an investment.

Which proxies are included?

Explore more market intelligence

Compare risk and return · Market performance rankings · Country valuations and returns · Sector intelligence · Calculation methodology · Volatility definition