Portfolio optimization primarily focuses on balancing risk and return and represents a multidimensional optimiza-tion problem that requires robust optimization algorithms with strong search capabilities and stability. The artificial bee colony algo-rithm 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 accu-racy;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 algo-rithm 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
Key words
swarm intelligence optimization algorithm/artificial bee colony algorithm/elite individuals/Metropolis algorithm/Sharpe ratio
维普
万方数据