A Hybrid Fuzzy Multi-Criteria Decision-Making and Enhanced Random Forest Approach for Optimizing Stock Portfolio Profit

This paper proposed a hybrid approach that combines fuzzy multi-criteria decision-making (MCDM) and an enhanced random forest algorithm to optimize stock portfolio profit. The motivation stems from the need to address the limitations of traditional portfolio optimization methods, which often rely on simplistic criteria or fail to account for the inherent uncertainty and multi-dimensionality of stock evaluation. The proposed method first employs a fuzzy analytic hierarchy process (FAHP) to rank stocks based on multiple criteria, where fuzzy logic handles the vagueness in decision-making. The weights of the criteria are derived from pairwise comparisons, and the overall score for each stock is computed by aggregating the weighted performance ratings. In the second stage, an enhanced random forest algorithm is introduced to predict future stock returns, incorporating a genetic algorithm-based feature selection mechanism to improve predictive accuracy. The integration of these two stages allows for a more robust and adaptive portfolio optimization strategy, as the ranked stocks from the FAHP are further refined using the predicted returns. The novelty of this work lies in the synergistic combination of fuzzy MCDM and machine learning, which not only captures the subjective preferences of investors but also leverages data-driven insights for better decision-making. Experimental results demonstrate that the proposed approach outperforms conventional methods in terms of both risk-adjusted returns and portfolio stability. This study contributes to the field by providing a comprehensive framework that bridges the gap between qualitative judgment and quantitative analysis, offering practical value for investors and financial analysts seeking to enhance portfolio performance under uncertain market conditions.