Correlation, VIF, and Multicollinearity Diagnostics

How pairwise factor correlation and VIF help prevent redundant factor exposure.

Category: Portfolio

Correlation triangle

A correlation matrix shows whether factors or sleeves move together. High correlation reduces true breadth.

VIF

Variance inflation factor detects multicollinearity in regression inputs. For a factor j, VIF_j = 1 / (1 - R_j^2), where R_j^2 comes from regressing that factor on the other factors. High VIF means coefficient estimates may be unstable.

Practical use

If two sleeves are highly redundant, the briefer should prefer the cleaner, more validated sleeve rather than double-counting the same exposure.