Portfolio selection based on the global minimum variance (GMV) model remains a significant focus in financial research. The covariance matrix, central to the GMV model, determines portfolio weights, and its accurate estimation is key to effective strategies. Based on the decomposition form of the covariance matrix. This paper introduces semi-variance for improved financial asymmetric risk measurement; addresses asymmetry in financial asset correlations using distance, asymmetric, and Chatterjee correlations to refine covariance matrices; and proposes three new covariance matrix models to enhance risk assessment and portfolio selection strategies. Testing with data from 30 stocks across various sectors of the Chinese market confirms the strong performance of the proposed strategies.
Sun et al. (Wed,) studied this question.