Can We Give the Maximum Sharpe Ratio Portfolio a Chance?

This chapter studies the applicability of the maximum Sharpe ratio (MaxSR) portfolio strategy in real-world settings. As shown by Okhrin and Schmidt the plug-in estimated weights show abysmal distributional properties such that it renders an application impossible for financial practitioners. In this chapter we propose a double regularization approach for the MaxSR portfolio strategy based on the bagged pretested portfolio selection (BPPS) algorithm. We show that for certain settings the doubly shrunken portfolio weights strongly mitigate the adverse properties of the plug-in estimated weights and can beat the popular 1/N benchmark strategy.