Correlation: The Hidden Variable Behind Every Diversified Portfolio

Correlation quantifies how closely two assets' returns move together, on a scale from -1 (perfectly opposite) to +1 (perfectly identical). It is the single most important number behind the theory of diversification, because it determines how much risk-reduction benefit an investor gets from combining two assets — independent of either asset's individual volatility.

How it's measured

Correlations are estimated from historical return data over some lookback window. That immediately raises two subtleties: the number depends on the period you choose, and it depends on the frequency of the data (daily, monthly, annual). A pair of assets can look uncorrelated using monthly data over ten years and more correlated in a specific crisis month. This is why correlation should be read as a rough, historical tendency, not a precise physical constant.

Correlations change — usually when you least want them to

Many assets that appear only loosely correlated in calm markets become much more correlated during a systemic crisis, when investors sell everything at once to raise cash. This is sometimes summarized as correlations go to 1 in a crisis, and it is a genuine limitation of leaning too heavily on historical correlation. The diversification you counted on can weaken at the exact moment you need it most. The practical response is to include at least one holding — often long-term Treasuries or gold — whose crisis behavior has historically differed from equities.

Putting it to work

Despite the caveats, long-run correlation remains a useful starting point. US stocks and long-term Treasuries have historically shown low or even negative correlation over full cycles, which is the central reason the 60/40 portfolio has endured. Gold has historically shown low correlation with both stocks and bonds, which is why it anchors defensively-minded strategies like the Permanent Portfolio and Golden Butterfly. The most diversifying additions to a stock-heavy portfolio are precisely those with the lowest correlation to stocks — even if, on their own, they look unexciting.

See it directly

This site's correlation matrix tool shows these relationships computed from the same historical data used in every backtest, so you can reason about diversification with real numbers. Pair it with the idea of risk parity, which takes low-correlation portfolio construction to its logical conclusion, and with our risk metrics guide to connect correlation to the volatility and drawdown figures shown on every strategy page.