Multiobjective Evolutionary Metaheuristic Approach to the Constrained Portfolio Optimization Problem

Abstract

In this paper, we propose a multi-objective evolutionary metaheuristic approach based on the Pareto Ant Colony Optimization (P-ACO) metaheuristic and the non-dominated genetic sorting algorithms (NSGA II and NSGA III) to solve a bi-objective portfolio optimization problem. P-ACO is used to select the best assets composing the efficient portfolio. Then, NSGA II and NSGA III are separately used to find the proportional weights of the budget allocated to the selected portfolio. The results we obtained by these two algorithms were compared to designate the best performing algorithm. Finally, we performed another comparison between our results and those of an exact method used for the same problem. The numerical experiments performed on a set of instances from the literature revealed that the combination of the ant colony optimization metaheuristic and the NSGA III genetic algorithm that we proposed most often gave much better results than both the combination of the ant colony optimization metaheuristic and NSGA II on the one hand and the iterative approach on the other hand.