Covariance Matrix Linear Algebra Interview Guide
Covariance matrix linear algebra interview guide for symmetry, PSD constraints, correlations, estimation, PCA, examples, and mistakes.
Candidates connecting portfolio risk, covariance estimation, PCA, and matrix validity.
A covariance matrix stores pairwise co-movement
Diagonal entries are variances and off-diagonal entries are covariances. The matrix should be symmetric when it represents the same variables on both axes.
PSD is a validity condition
A valid covariance matrix is positive semidefinite because any weighted portfolio variance should be nonnegative. Estimated matrices can violate useful numerical properties.
Concrete example
If an estimated covariance matrix produces a negative variance for some weight vector, the estimate or numerical procedure is invalid for portfolio risk use.
Estimation noise matters
Sample covariance can be unstable when assets are many and observations are limited. Shrinkage, factor models, or cleaning may improve usability.
Common mistakes
Candidates often quote covariance formulas without checking matrix properties. Strong answers discuss symmetry, PSD, estimation noise, and interpretation.
Practice the pattern
Use the LeetQuidity curriculum and calibration to turn this topic into a focused practice plan.