Robust Portfolio Selection with Near Optimal Centering

Abstract

Quantitative asset allocation models have not been widely adopted by practitioners because they suffer from two problems: the lack of robustness and diversification of portfolios obtained through these models. To solve these problems, We developed a new portfolio selection method that can be applied to any convex risk measure. The procedure begins selecting an optimal portfolio in the efficient frontier, then we define a near optimal region and finally we define the analytic center as the new optimal portfolio. We compare 30 portfolio optimization models for 4 asset samples, and the results suggest that the new method overcomes traditional methods in robustness and diversification.

Keywords: Mean Variance Portfolio, MAD Portfolio, CVaR Portfolio, Robust Optimization, Portfolio Selection, Near Optimal Portfolios

JEL Classification: C61, G11