Parameter tuning of hybrid nature-inspired intelligent metaheuristics for solving financial portfolio optimization problems

Abstract

In previous studies, nature-inspired algorithms have been implemented in order to tackle hard NP-optimization problems, in the financial domain. Specifically, the task of finding optimal combination of assets with the aim of efficiently allocating your available capital is of major concern. One of the main reasons, which justifies the difficulties entailed in this problem, is the high level of uncertainty in the financial markets and not only. As mentioned above, artificial intelligent algorithms may provide a solution to this task. However, there is one major drawback concerning these techniques: the large number of open parameters. The aim of this study is twofold. Firstly, results from extended simulations are presented regarding the application of a specific hybrid nature-inspired metaheuristic in a particular formulation of the financial portfolio optimization problem. The main focus is on presenting comparative results regarding the performance of the proposed scheme for various configuration settings. Secondly, it is our intend to enhance the hybrid scheme's performance by incorporating intelligent searching components such as other metaheuristics (simulated annealing). © 2012 Springer-Verlag.