Clustering Stock Exchange data by Using Evolutionary Algorithms for Portfolio Management
In present paper, imperialist competitive algorithm and ant colony algorithm and particle swarm optimization algorithm have been used to cluster stocks of Tehran stock exchange. Also results of the three algorithms have been compared with three famous clustering models so called k-means, Fcm and Som. After clustering, a portfolio has been made by choosing some stocks from each cluster and using NSGA-II algorithm. Results show superiority of ant colony algorithms and particle swarm optimization algorithm and imperialist competitive to other three methods for clustering stocks. Due to diversification of the portfolio, portfolio risk will be reduced while using data chosen from the clusters. The more efficient the clustering, the lower the risk is. Also, using clustering for portfolio management reduces time of portfolio selection.