<![CDATA[This study aims to optimize stock portfolio allocation through a data-driven approach by integrating the K-Means Clustering algorithm and the Markowitz Model. The dataset includes technical and fundamental indicators of LQ45 index stocks from 2019 to 2024. The process begins with data normalization and feature extraction, followed by stock clustering using the K-Means algorithm. From the four resulting clusters, the top-performing stock with the highest average return is selected from each. Portfolio weights are then optimized using the Markowitz Model under a mean-variance framework without short selling. The optimization results allocate the largest weights to ARTO, BRPT, and ISAT. Performance evaluation through a backtest simulation in 2024 shows that the portfolio experienced only an 8.02% decline, outperforming the LQ45 index which dropped by 15.60%. These findings underscore the potential of integrating data mining and quantitative optimization methods to improve diversification efficiency and strengthen portfolio resilience during market downturns.]]>
Copyrights © 2025
