Sample Covariance Issue: The sample covariance matrix, though unbiased, often has high estimation errors, especially with more stocks than price points.
Shrinkage Estimation: Combines the sample covariance with a structured target (constant correlation model) to reduce error.
Shrinkage Target: Sets all pairwise correlations to the average correlation.
Empirical Results: Shrinkage improves performance over sample covariance, especially for smaller portfolios.