A Multivariate Analysis and Machine Learning Approach to ESG-Informed Portfolio Optimization A Case Study of Multiple Companies

Environmental, Social, and Governance (ESG) factors have steadily gained prominence in portfolio optimization in recent years. Through its offering of both financial returns and sustainability benefits. The integration of ESG metrics into financial models presents clear challenges. Some of these challenges includes the variability of ESG data and scoring methodologies. This study aims to develop a machine learning-driven framework, that combines ESG scores with financial metrics. With the aim of optimizing portfolio allocation, and addressing the complexities of ESG data while enhancing risk management. This research employs the use of machine learning techniques such as Random Forest, Logistic Regression, and Ensemble Learning to analyze ESG and financial data; from companies across different sectors. Portfolio weighting decisions and risk assessments were done using Volatility metrics, Sharpe Ratios, and Monte Carlo simulations. The results highlight that portfolio which incorporates ESG data, deliver improved risk-adjusted returns and lower volatility. This is particularly observed in the technology and financial services sectors. Dynamic ESG metrics, such as rolling averages, provide a more accurate reflection of long-term sustainability performance. Machine learning models outperform traditional methods in ESG risk classification and portfolio optimization. This study concludes that ESG integration, supported by machine learning; significantly enhances portfolio resilience and sustainability. Future research could explore real-time portfolio rebalancing and the application of ESG metrics to a wider range of asset classes.