Machine Learning ESG Strategy

The strategy applies machine learning methods—Random Forest, Gradient Boosted Regression Trees, and Robust Elastic Net—to forecast stock returns using detailed environmental data. Based on these forecasts, it forms long portfolios of “green” firms with strong environmental performance and short portfolios of “brown” firms with weak performance, rebalancing periodically to capture return differentials linked to climate concerns.

The rationale rests on the rising financial importance of sustainability. Environmental performance signals provide investors with insights into climate-related risks and opportunities. By combining granular ESG data with machine learning, the strategy enhances predictive accuracy, generates abnormal returns, and aligns investment decisions with the broader transition toward a greener economy.