High-Dimensional Portfolio Selection through a Robust Glasso Approach

Proceedings of The 14th International Conference on Management, Economics and Humanities

Year: 2023



High-Dimensional Portfolio Selection through a Robust Glasso Approach

Wenliang Ding, Lianjie Shu, Xinhua Gu



The Glasso applied to portfolio selection in Goto & Xu (2015) achieves significant risk reduction and boosts certainty- equivalent returns (CER) through sparse estimators against hedge trades. However, the sample covariance matrix used as input for their Glasso is susceptible to data outliers. This input is replaced in our Glasso by a Kendall-type robust estimator (Glasso-K). The new Glasso inherits the original Glasso’s risk reduction advantage while dealing well with data contamination. The Glasso-K is found to perform better than the Glasso in every aspect, especially in the CER because of its induced better- conditioned covariance, more-diversified portfolios, and less-frequent turnover. The robust Glasso also performs better than many non-Glasso strategies well established in the literature, and its superior performance consists in complete removal of sample means from covariance estimation.

keywords: Data contamination, High dimension, Hedge regression, Portfolio selection, Robust covariance matrix estimation