Anticipatory Stochastic Multi-Objective Optimization for Uncertainty Handling: A Case Study in Portfolio Selection

This paper deals with how to incorporate anticipatory behavior in multi-objective metaheuristics. We propose a principled methodology to track the stochastic dynamics of the population of candidate Pareto optimal solutions in the objective space by incorporating the Kalman Filter (KF) estimation into the Anticipatory S-Metric Selection Evolutionary Multi-Objective Algorithm (ASMS-EMOA). The proposal is assessed for real-world cardinality constrained portfolio selection. Preliminary results suggest that we were able to reduce the sum of squared errors prediction of the portfolios ex-post return and risk estimation in out-of-sample investment environments.