基于双种群交流粒子群算法的离散投资组合模型

Discrete portfolio selection model based on particle swarm optimization algorithm with two particle swarms communicating with each other

针对标准粒子群算法易陷入局部最优的缺陷,提出一种双种群交流的新型粒子群算法,利用速度变异成功地解决了上述问题;综合考虑了我国股票市场上的交易费用、整数手数投资、不允许买空卖空等问题,建立了符合我国股票市场的投资组合模型,并将双种群交流的离散粒子群算法应用于其求解过程中,给出最优投资组合。

Considering that standard Particle Swarm Optimization（PSO）has the severe problem of being stuck in local optimums, this paper puts forward an improved particle swarm optimization with two particle swarms communicating with each other on the basis of velocity mutation, leading to the problem above to be resolved. In addition, taking into account the existence of transaction cost and the restraints of long sale, short sale, integral number of transactions and so on in China,this paper builds a portfolio selection model which totally reflects current status in our country. It applies the improved PSO above to deal with the model. The result comes out that the model is integrated and effective in our country, and also the PSO with two particle swarms communicating with each other is proper and efficient.