Implementation of iterative k -means-+ and ant colony optimization (ACO) in portfolio optimization problem

Abstract

Portfolio optimization aims to provide investors with the highest returns and the least amount of risk. To that end, investors diversify to improve a portfolio's effectiveness by reducing risks. In this research, we used the iterative k-means -+ algorithm as a clustering method and ant colony optimization (ACO). Clustering was used to diversify a portfolio based on the financial ratio of each stock. Iterative k-means -+ improves the solution obtained using k-means by removing 1 cluster (minus), dividing another cluster (plus) and re-clustering with each iteration. After clustering, some of the stocks are chosen and their weights are determined using a metaheuristic method, ant colony optimization (ACO). The numerical result of this method is evaluated with the data. This research yielded the results in which the performance of the iterative k-means -+ and ACO methods yields better returns and Sharpe ratios compared to those of the S&P 500 index data used.