Machine learning in Esg investing: predictive analysis for stock performance

Abstract

This strategy explores the use of ESG metrics within machine learning frameworks, particularly Gradient Boosting Trees and SHAP-value analysis, to predict stock performance in the Nordic Markets. Focusing on data from 2007-2022, it examines the efficiency of ESG metrics in forecasting Active Return. Utilizing the model, it reveals that long portfolios, especially those informed by SHAP analysis, consistently outperform the OMX Nordic Large cap Index, while short portfolios show underperformance. The study highlights the potential of machine learning in enhancing ESG-focused investment strategies, suggesting the need for broader datasets and diverse market analysis for more robust and comprehensive investment insights.