Glossary
Covariance
A statistical measure of how two assets' returns move together; the raw ingredient behind correlation, beta and portfolio risk.
Covariance measures whether two assets tend to be above or below their own average returns at the same time. Positive covariance means they tend to move together. Negative means they tend to move in opposite directions. Near zero means there is no steady relationship.
From covariance to correlation
The size of a covariance depends on how volatile the two assets are, which makes the raw number hard to read. Dividing by both standard deviations gives correlation, which always lies between −1 and +1.
Variance is an asset's covariance with itself, and standard deviation is its square root.
Why it matters for portfolios
The risk of a portfolio is not the average of its holdings' risks. It depends on every pair's covariance.
A portfolio of many assets needs a covariance matrix, a table with one entry for each pair. A hundred assets have 4,950 distinct pairs. This table is the central input to modern portfolio theory.
Limits
- It is estimated from the past and changes over time. Covariances between risky assets tend to rise sharply in a crisis.
- Many estimates, little data. Large matrices contain so many numbers that errors are unavoidable, and optimizers tend to magnify them.
- It captures only straight-line relationships. Two assets can be unrelated on ordinary days and fall together on extreme ones.
