Cardinality constrained portfolio optimization using Cuckoo search algorithm / by Karl Toni Sawaya.

Abstract

This thesis presents a new heuristic method for solving the cardinality portfolio optimization problem with constraints on the number of assets to invest in. Adding the cardinality constraint makes the problem NP-complete, hence it is not solvable to optimality by simplex polynomial algorithms. As a consequence, heuristics and meta-heuristics approximate methods become the only optimization tools which provide good solutions in a reasonable time frame. Recently, a new meta-heuristic algorithm called Cuckoo Search (CS) was introduced by Yang and Deb (2009) to provide significant improved results to continuous optimization problems. To the authors' knowledge, this thesis presents the first application of the CS algorithm to the cardinality portfolio optimization (CPO) problem. Our implementation involves a master solver, CS, to explore the cardinality search space using Lévy flights, and a slave solver, (a quadratic solver), to find the optimal allocation of investment weights for a chosen set of assets. The hybrid algorithm was tested on a set of data used in the literature by several researchers including Chang et al. (2000), Fernandez and Gomez (2007), and Deng et al. (2012). Our results were compared favorably to previously published results with new found solutions. Further research directions are also reported.