基于多策略改进人工蜂群算法的投资组合优化应用  

Portfolio Optimization Application Based on Multi-Strategy Enhanced Artificial Bee Colony Algorithm

作　　者：李世豪 魏文红 张宇辉 吴昊 LI Shihao;WEI Wenhong;ZHANG Yuhui WU Hao(School of Computer Science and Technology,Dongguan University of Technology,Dongguan 523808,China)

机构地区：[1]东莞理工学院计算机科学与技术学院,广东东莞523808

出　　处：《东莞理工学院学报》2024年第3期9-16,42,共9页Journal of Dongguan University of Technology

基　　金：国家科技创新2030-“新一代人工智能”重大项目;广东省普通高校“人工智能”重点领域专项项目(2019KZDZX1011);东莞市社会发展科技项目(20211800904722);东莞市科技特派员项目(20221800500052)。

摘　　要：投资组合问题主要侧重风险与收益的平衡,是多维优化问题,需要寻优能力强且具稳定性的算法。人工蜂群算法以其强大的寻优能力和对参数不敏感的优点而著称,因此适用于解决投资组合问题。本文结合反向学习策略、精英策略和Metropolis算法策略三种策略,对参数较少的人工蜂群算法进行改进,以解决投资组合的风险与收益平衡问题。在初始化阶段,利用Logistic映射和反向解来提高收敛速度和精度;在雇佣蜂和跟随蜂工作过程中,利用上一代精英个体的位置信息,让蜂群更好地靠近最优蜜源;最后借鉴模拟退火中Metropolis算法,重新设计侦察蜂机制,提高算法的寻优能力。实验表明,本文提出的人工蜂群EABC算法(Elite Artificial Bee Colony),具有较强的寻优能力和较高的稳定性。不管是在30维还是100维中,EABC算法对6个测试函数的测试都可以取得较高的收敛精度。最后,EABC算法在投资组合夏普比率模型的优化中,可以得到可靠的方案,较高的夏普比率。Portfolio optimization primarily focuses on balancing risk and return and represents a multidimensional optimization problem that requires robust optimization algorithms with strong search capabilities and stability.The artificial bee colony algorithm is known for its powerful optimization capabilities and insensitivity to parameters,so it is very suitable for solving investment portfolio problems.This paper combines three strategies:Reverse learning strategy,Elite strategy and Metropolis algorithm strategy to improve the artificial bee colony algorithm with fewer parameters to solve the risk and return balance problem of the investment portfolio.In the initialization stage,the Logistic mapping and reverse solution are used to improve the convergence speed and accuracy;in the process of hiring bees and following the bees,the location information of the previous elite individuals is used to improve the optimal honey source;finally,the Metropolis algorithm in simulated annealing is used to redesign the scout bee mechanism to enhance the algorithm's search capabilities.Experimental results demonstrate that the proposed Elite Artificial Bee Colony(EABC)algorithm exhibits strong optimization capabilities and high stability.Whether in 30 dimensions or 100 dimensions,the EABC algorithm can achieve high convergence accuracy when testing the six test functions.Moreover,when applied to portfolio optimization,EABC algorithm provides a reliable solution with a high Sharpe ratio.

关 键 词：群体智能优化算法 人工蜂群算法 精英个体 Metropolis算法 夏普比率 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]