A MODIFIED HARMONY SEARCH ALGORITHM FOR PORTFOLIO OPTIMIZATION PROBLEMS

作者：Tuo, ShouHeng^*

来源：Economic Computation and Economic Cybernetics Studies and Research, 2016, 50(1): 311-326.

• Poroeolio Optimization
• Harmony search Algorithm
• Dimensional-selection strategy
• Cardinality Constrained Mean-Variance Model

For a diversified portfolio problem, building an optimization model is very necessary to make investment return be as large as possible and to make the investment risk be as small as possible. In this work, firstly, the basic mathematic model of Portfolio Optimization (PO) and Cardinality Constrained Mean Variance (CCMV) model are introduced. Then a modified Harmony search algorithm called HSDS based on Dimensional-Selection (DS) strategy and dynamic fret width (FW) strategy is proposed to solve PO problems, in which the DS strategy is for avoiding generating invalid solutions and the FW strategy is to balance global exploration and local exploitation. Finally, Genetic Algorithm, Particle Swarm Optimization, Simulated Annealing and Tabu Search are compared with the HSDS algorithm employing five portfolio problems (HangSeng, DAX 100, FTSE 100, S&P 100 and Nikkei). Experimental results indicate that the proposed algorithm is very effective for solving large scale portfolio optimization problems.

• 出版日期2016
• 单位陕西理工大学