CLUSTER ANALYSIS FOR MEAN-VARIANCE PORTFOLIO SELECTION: A COMPARISON BETWEEN K-MEANS AND K-MEDOIDS CLUSTERING
Main Article Content
Abstract
This paper presents the Mean-Variance (MV) portfolio selection using cluster analysis. Stocks are categorized into various clusters using K-Means and K-Medoids clustering. Based on the Sharpe ratio, a stock from each cluster is chosen to represent that cluster. Stocks with the greatest Sharpe ratio are those that are chosen for each cluster. With the guidance of the MV portfolio model, the optimum portfolio is identified. When there are many stocks included in the formation of the portfolio, we may efficiently create the optimal portfolio using this method. For the empirical study, the daily return of stocks traded on the Indonesia Stock Exchange that are part of the LQ-45 index from August 2022 to January 2023 was used to establish the weight of the portfolio, while the fundamental data of LQ-45 stocks for 2022 were used to build clusters. Using K-Means and K-Medoids clustering, this study's results show that LQ-45 stocks are divided into six groups. Additionally, it is obtained that for risk aversion , portfolio performance with K-Means clustering is better than portfolio performance with K-Medoids clustering. In contrast, for risk aversion , portfolio performance with K-Medoids clustering is better than portfolio performance with K-Means clustering.
Article Details
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.