Outperforming ESG stocks portfolio: A machine learning ranking model with catboots regressor

Vittorio Carlei, Donatella Furia, Alessandro Ceccarelli, Piera Cascioli,
Outperforming ESG stocks portfolio: A machine learning ranking model with catboots regressor,
The North American Journal of Economics and Finance,
Volume 80,
2025,
102517,
ISSN 1062-9408,
https://doi.org/10.1016/j.najef.2025.102517.
(https://www.sciencedirect.com/science/article/pii/S1062940825001573)
Abstract: The paper investigates whether outperforming ESG (Environmental, Social, and Governance) stocks can generate alpha over the S&P 500 through a machine learning (ML)-driven simulation. Leveraging advanced ML techniques, particularly the CatBoostRegressor model, the study explores the relationship between ESG factors and financial performance to construct a high-performing, ESG-compliant portfolio. The findings suggest that incorporating ESG criteria not only supports sustainable investing but also enhances return predictability. However, backtesting reveals that portfolios more tolerant of lower ESG scores tend to yield higher short-term returns. A customizable threshold for ESG criteria allows the strategy to accommodate varying risk appetites, demonstrating that strategic flexibility can align ESG values with financial goals. This work highlights the transformative role of machine learning in sustainable investing, illustrating how ethical considerations and profitability can effectively coexist within a robust, ML-powered framework.
Keywords: ESG investing; Alpha generation; Machine learning; Cat boots regressor