Key Takeaways
- Co-movement is the core concept: +1 indicates perfect positive linear correlation, -1 perfect negative correlation, and 0 no linear correlation.
- Diversification depends on imperfect correlation: Combining assets that do not move identically can reduce portfolio volatility.
- Causation requires separate evidence: Two assets moving together does not prove that one caused the other to move.
- The number can change: Correlations depend on the time period, return frequency, market regime, and assets being measured.
- One statistic is not enough: Nonlinear relationships, tail risk, liquidity, and valuation are not captured by a simple correlation coefficient.
How to Read a Correlation Coefficient
| Coefficient | Interpretation |
|---|---|
| +1.0 | Perfect positive linear relationship |
| Between 0 and +1 | Tendency to move in the same direction |
| Near 0 | Little or no linear relationship |
| Between 0 and -1 | Tendency to move in opposite directions |
| -1.0 | Perfect negative linear relationship |
A value of 0.70 does not mean the assets move together 70% of the time. Correlation is a statistical measure of the relationship between paired observations, not a probability or percentage of matching days.
Why Correlation Matters for Diversification
Portfolio risk depends not only on the volatility of each holding but also on how those holdings move together. Two volatile assets can still provide diversification if their returns are sufficiently different across time.
Combining assets with lower correlations can therefore reduce total portfolio variability relative to holding assets that behave almost identically. The benefit does not require negative correlation; even a positive correlation well below +1 can provide some diversification.
Individual risk measures such as standard deviation provide additional context.
A Simple Investment Example
Example: Portfolio A holds two funds that have historically moved almost in lockstep, with a correlation near +0.95. A second portfolio pairs one of those funds with an asset whose historical correlation is +0.30.
All else equal, Portfolio B has more potential diversification because the second holding has behaved less like the first. The actual risk reduction still depends on each asset’s volatility and portfolio weight.
Correlation Changes Over Time
Historical correlation is not a permanent property. Different sample periods can produce different values, especially when markets move from calm conditions into recessions, inflation shocks, liquidity stress, or other regime changes.
Return frequency matters too. Daily, monthly, and annual data can show different relationships. Currency treatment, hedging, and the benchmark used for each asset can also change the result.
Correlation Does Not Mean Causation
Two variables can move together because one affects the other, because both respond to a third factor, or by coincidence. Correlation alone cannot distinguish among those explanations.
Investment analysis therefore needs an economic reason behind the statistic. A persistent relationship tied to interest rates, inflation, commodity exposure, or business-cycle sensitivity may be more informative than a high coefficient with no plausible mechanism.
Correlation vs. Covariance
Covariance also measures whether two variables tend to move together, but its magnitude depends on the units of the variables. Standardizing covariance by the variability of each series produces a dimensionless correlation coefficient between -1 and +1.
That standardized scale makes correlation easier to compare across different asset pairs.
How Investors Can Use Correlation Responsibly
- Use total-return data that matches the decision. Compare consistent return frequencies and time periods.
- Look at more than one window. Rolling or multi-period views can reveal unstable relationships.
- Combine correlation with volatility. Portfolio risk depends on both.
- Stress-test difficult periods. Check whether the relationship changed during major selloffs.
- Understand the economic exposure. Economic logic makes the statistic more useful when the drivers behind it make sense.
Frequently Asked Questions (FAQs)
What is a good correlation for diversification?
No single cutoff works for every portfolio. Lower correlation generally improves diversification potential, but volatility, expected return, weight, liquidity, taxes, and investment objective also matter.
Can two investments have zero correlation and still be related?
Yes. Pearson correlation measures linear relationships. A strong nonlinear relationship can exist even when the coefficient is near zero.
Why can correlation rise during a crisis?
Common shocks, forced selling, funding stress, or broad changes in risk appetite can push different assets in the same direction at the same time.
How often should investors recalculate correlation?
There is no universal schedule. Reassessment is most useful when portfolio holdings, market conditions, or the investment horizon changes materially.