Robust portfolio optimization with second order stochastic dominance constraints

Abstract

A portfolio optimization problem equipped with stochastic dominance constraints creates optimal portfolio ideal for rational and risk-averse investors. This paper proposes a robust portfolio optimization model involving second-order stochastic dominance in constraints. The input returns of the assets are considered as uncertain parameters and are varied in symmetric and bounded intervals to construct an optimal robust portfolio. Although the resulting optimization model is a linear program, it involves a large number of constraints, thereby urging us to apply the cutting plane algorithm. We experimentally examine the performance of our model on datasets drawn from S&P 500, S&P BSE 500, Nikkei 225, S&P Global 100, FTSE 100, and BOVESPA index, and compare it with the corresponding non-robust counterpart model. The optimal portfolios from the proposed model are shown to yield better performance on several performance measures, including Sharpe ratio, STARR ratio, standard deviation, worst return, violation area in SSD, value at risk, and conditional value at risk. The results demonstrate the effectiveness and efficiency of the presented robust model.