Constrained portfolio selection using multiple swarms

Markowitz mean-variance model is one of the best known models that has been heavily studied in modern world of finance. However, the model is considered to be too basic in practice, as it ignores many of the constraints that real-world investors have to face with. In this paper we focused on a complex constrained portfolio selection model with additional constraining factors including the transaction fee, the minimal transaction unit, the maximal transaction quantity of every assets and the minimum/maximum of the investment. When taken these complex constraints in to account, the process became a high-dimensional constrained optimization problem. In our study, based on the study of symbiosis phenomenon in natural ecosystem, a multiple-swarm approach (SMPSO) was proposed to solve the resulting model. A numerical experimental study of a portfolio selection problem was conducted to illustrate our proposed method. The simulation results demonstrated that our proposed method is more efficient than PSO based method in solving the complex constrained portfolio selection problem.