The classical mean - variance model is used to study the problem of optimizing the actual return rate of asset, and the problem of the actual return rate can not be well dealt with in real world for the model's extremely sensitivity to input parameters. In order to solve the problem better, a bi - criteria portfolio model based on 1∞ risk function was established. As for the discontinuity of objective function in the model, an improved particle swarm optimization algorithm was chosen to solve it. And considering optimal and suboptimal locations, crossover operations in genetic algorithm were introduced in the algorithm. In simulation experiments, with real data from stocks market it was obtained that the new model has less value at risk than mean - variance model, meanwhile it performs better in the actual return rate.%研究最优化处理资产的实际收益率问题,经典的均值一方差模型对于输入参数过于敏感,不能较好的处理真实世界中的实际收益率问题.为了更好的解决上述问题,建立了一种l∞风险函数的双目标投资组合模型.针对模型中目标函数的不连续性,采用了改进的粒子群优化算法进行求解.改进算法在考虑最优和次优位置的基础上,引入了遗传算法中的交叉操作.在仿真中,运用证券市场中的真实数据分析得出,新模型能够获得比均值一方差模型更小的风险值,同时在实际收益率方面表现更好.
展开▼
