A hybrid particle swarm optimization approach and its application to solving portfolio selection problems

Shamshul Bahar Yaakob, Junzo Watada

    Research output: Contribution to journalArticlepeer-review


    In modern portfolio theory, the basic topic is how to construct a diversified portfolio of financial securities to improve trade-offs between risk and return. The objective of this paper is to apply a heuristic algorithm using Particle Swarm Optimization (PSO) to the portfolio selection problem. PSO makes the search algorithm efficient by combining a local search method through self-experience with the global search method through neighboring experience. PSO attempts to balance the exploration-exploitation tradeoff that achieves efficiency and accuracy of optimization. In this paper, a newly obtained approach is proposed by making simple modifications to the standard PSO: the velocity is controlled and the mutation operator of Genetic Algorithms (GA) is added to solve a stagnation problem. Our adaptation and implementation of the PSO search strategy are applied to portfolio selection. Results of typical applications demonstrate that the Velocity Control Hybrid PSO (VC-HPSO) proposed in this study effectively finds optimum solution to portfolio selection problems. Results also show that our proposedmethod is a viable approach to portfolio selection.

    Original languageEnglish
    Pages (from-to)473-478
    Number of pages6
    JournalJournal of Advanced Computational Intelligence and Intelligent Informatics
    Issue number4
    Publication statusPublished - 2011 Jun


    • Genetic algorithm
    • Hybrid particle swarm optimization
    • Modern portfolio theory
    • Particle swarm optimization

    ASJC Scopus subject areas

    • Artificial Intelligence
    • Computer Vision and Pattern Recognition
    • Human-Computer Interaction


    Dive into the research topics of 'A hybrid particle swarm optimization approach and its application to solving portfolio selection problems'. Together they form a unique fingerprint.

    Cite this