Portfolio weights solely based on risk avoid estimation errors from the sample mean. but they are still affected from the misspecification in the sample covariance matrix. To solve this problem. we shrink the covariance matrix towards the Identity. https://tonerparts.shop/product-category/ribbon/