基于引力搜索算法的证券投资组合问题研究

Study of portfolio optimization problem based on gravitational search algorithm

引力搜索算法是模拟万有引力定律进行搜索的一种新颖的优化算法,已有研究表明该算法比传统的一些优化算法拥有较好的收敛性能,但该算法在局部搜索能力上有所欠缺。提出一种基于惯性递减权重的引力搜索算法(gGSA),该算法能够在局部进行进一步的探索,强化局部搜索能力。该算法应用到基于VaR的证券最优投资组合模型中,解决证券投资组合优化的问题,并以上证50指数中成分股于2012年上半年日收盘价格作为测试数据集进行计算,结果表明改进算法所得到的投资比例能够获得较好的收益率。

Gravitational search algorithm（GSA）is a novel optimization algorithm based on the law of gravity and mass interac-tions.In the algorithm,the searcher agents were a collection of masses which interacted with each other based on the Newtoni-an gravity and the laws of motion.Some study show that GSA can obtain more superior results than PSO in most cases.But the basic gravitational search algorithm was eary to be trapped into local optimum.This paper proposed an improvement weighted algorithm,named gGSA,and strengthed local search capacity of GSA.This algorithm was used to portfolio optimization model based on VaR and solved the portfolio optimization problem.The result demonstrates that the new algorithm is feasible and ef-fective.