Bayesian Portfolio Selection: Application to Tactical Asset Allocation

Abstract

This article discusses the portfolio selection problem from a Bayesian perspective. In doing so, I first provide an overview of the portfolio problem and motivate the decision-making process from an expected utility point of view. Then, I demonstrate the analytical solution to the problem and stress the intuition behind the Bayesian application. In particular, in the case of risky assets, the optimal portfolio corresponds to three funds. The first is the minimum variance portfolio, whereas the others denote two self-financing portfolios corresponding to the mean returns. The combination between the first and the second funds is consistent with the conventional mean-variance portfolio. Furthermore, with the inclusion of the third fund, the portfolio incorporates the priors/beliefs of the decision-making into the portfolio selection. Based on the analytical insights, I conduct a small empirical experiment using two ETFs. The experiment emulates a tactical asset allocation problem. All the empirical analysis is conducted using R.