Predicting Firms Market Value in Italy: The Role of Environmental Scores

This study investigates the ability of two tree-based Machine Learning models (Random Forest and XGBoost) to predict the market value of Italian listed firms. The study employs a dataset comprising 71 non-financial companies observed between 2019 and 2023. The adopted models are benchmarked against linear regression in terms of predictive performance; they are interpreted using SHAP values and Rank Graduation Explainability. The results show that the predictions are primarily driven by financial metrics, with environmental scores playing a marginal role.