ESG Scores and Stock Performance Prediction: A Quantitative Study Using Random Forest Model

Environmental, Social, and Governance (ESG) factors have become increasingly relevant in financial investment decisions. Although previous research has focused on ESGs long-term financial impact, the predictive power of ESG scores on stock returns remains uncertain. This study employs a machine learning approach, utilizing a Random Forest model to investigate whether ESG scores can predict stock performance. Historical stock returns and ESG scores for the S&P 500 companies dataset are used, originally extracted from sources like Yahoo Finance. The dataset is used to train multiple machine learning models, including Random Forest, Logistic Regression, Decision Tree, and Support Vector Machine (SVM), distinguishing between high- and low-return stocks based on ESG metrics. The correlation analysis and feature importance analysis are carried out to examine the real impact of the ESG scores on stock performances. The findings suggest that ESG scores exhibit minimal to no predictive power in forecasting stock performance, challenging the notion that ESG-driven investment strategies yield superior returns. These results contribute to the growing debate on the financial relevance of ESG factors.