Green Finance Analytics: AI-Based Stock Valuation Using ESG and OHLCV Data

Green finance represents a strategic shift in financial activities, where environmental, social, and governance (ESG) considerations are integrated into investment decisions to foster long-term sustainable development. However, in Indonesia, ESG data is typically presented as a single aggregated score and is rarely integrated with high-frequency market data such as Open, High, Low, Close, and Volume (OHLCV), resulting in limited stock performance analysis. This study develops an ESG-based stock performance scoring system by integrating ESG data with OHLCV using machine learning methods, including Random Forest, XGBoost, and CatBoost. Model performance is evaluated using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), R2, Area Under Curve (AUC), and Spearman rank correlation, with interpretability provided through Shapley Additive explanation (SHAP). The results show that while absolute predictive accuracy is limited by market noise, the models achieve high ranking consistency, with a Spearman correlation of up to 0.855. This indicates that the system is robust in maintaining stock ranking stability. The model is implemented in a Decision Support System (DSS) to support transparent and data-driven investment decisions. These findings demonstrate that integrating ESG and market data using explainable machine learning provides stable and interpretable insights, proving that the proposed system achieves robust ranking stability Spearman = 0.855 and is highly suitable as a decision-support tool in volatile financial environments.