Introduction

Traditional ESG scores are static snapshots. Real ESG performance evolves continuously. Machine learning models incorporate dynamic ESG improvements and deterioration into factor models, predicting which improving-ESG companies outperform and which deteriorating-ESG companies underperform.

Dynamic ESG Scoring

Rather than annual ESG scores, track continuous ESG metrics: board changes (improvement signal), policy announcements, controversies, sustainability report releases. Build time-series of ESG trajectories. Identify improving, stable, and deteriorating ESG trends.

Factor Model Integration

Combine value, momentum, quality factors with ESG momentum (changing ESG scores). Models learn: improving ESG + strong value = strong forward returns. Deteriorating ESG + high momentum = reversal risk. Dynamic integration improves factor model predictive power.

Conclusion

Dynamic ESG integration in factor models captures improving companies earlier, enabling alpha generation from ESG trends.