PSO for CVaR-based Portfolio Optimization: the case of Cordusio SIM

Abstract

The purpose of this dissertation is to obtain an optimal asset allocation for replicating the performances of a portfolio offered by Cordusio SIM, a firm specialized in wealth management and financial advisory. The idea is to develop the whole portfolio selection problem in quantitative terms. In particular, the latter is formulated in a manner that it minimizes a coherent risk measure, the Expected Shortfall, subject to a system of constraints which tries to make the problem as real as that selected by Cordusio SIM. Subsequently, due to its complexity, the portfolio selection problem is solved through a modern optimization technique, called Particle Swarm Optimization (PSO), which belongs to the family of metaheuristic algorithms. In the final part of the work, an attempt using the Median Shortfall as objective function of the portfolio selection problem is made.