Multiobjective portfolio optimization via Pareto front evolution

Abstract Portfolio optimization is about building an investment decision on a set of candidate assets with finite capital. Generally, investors should devise rational compromise to return and risk for their investments. Therefore, it can be cast as biobjective problem. In this work, both the expected conditional value-at-risk (CVaR) are considered objectives. Although objective CVaR optimized existing techniques such linear programming optimizers, involvement practical constraints induces challenges exact mathematical methods. Hence, we propose new algorithm named F-MOEA/D, which based Pareto front evolution strategy decomposition multiobjective evolutionary algorithm. This involves two major components, i.e., constructing local fronts through methods picking best one via approaches. The empirical study shows F-MOEA/D obtain better approximations test instances against several alternative algorithms same time budget. Meanwhile, large 7964 9090 assets, still performs well given that method does not finish in 7 days.