Portfolio Selection During Crises Using Principal Component Analysis

Abstract

The current Covid-19 crisis, that debuted as an exogenous shock, has determined overreaction and herding among investors, as well as efficiency breaches and alterations within the Gaussian characteristics of returns’ distributions by generating strong asymmetries. Under such circumstances, the – econophysics approached – correlation structure within the market has been affected to an uncertain extent. Within these conditions, the problem of optimal portfolio selection becomes a subject of interest even for professional investors who tend to seek refuge towards developed, mature economies and quality assets. The Principal Component Analysis manages to offer a considerably useful tool for the allocation problem, by reducing dimensionality and, at the same time, by being able to account for the “unobservable risk” hidden within the market. The method is able to provide well-diversified portfolios, by explaining the systematic risk component within the analyzed returns series. The paper analyzes the ability of the PCA method, used aside with a KMO Test, to provide high rates of return even during turbulent, extremely volatile periods, such as the current one. The results show that the PCA weighted portfolio manages to achieve a rate of return higher than the one generated by an equally weighted “market portfolio”, proving therefore the robustness of the method.