A Review of Research on Machine Learning-enabled Quantitative ESG Investing: A Perspective on Multi-factor Model Optimisation and Performance Evaluation

Wen Xueyan

 Guangzhou Software College, Financial Management

Abstract：

Against the backdrop of the deep integration of sustainable finance and quantitative investment, Environmental, Social and Governance (ESG) factors have become critical incremental information in asset pricing and portfolio allocation. Machine learning, with its powerful capabilities in data processing, feature extraction and non-linear fitting, has opened up entirely new avenues for the optimisation of multi-factor models in traditional ESG quantitative investment. Taking the two core perspectives of multi-factor model optimisation and performance evaluation as its starting point, this paper comprehensively reviews the research threads concerning the integration of machine learning, ESG quantitative investment and multi-factor models, summarising key advancements in ESG factor engineering, model structure optimisation, portfolio construction and performance evaluation systems. This paper is a narrative review that systematically integrates and sorts out previous studies rather than verifying a new empirical model. The aim of this study is to fill the gap in the literature regarding a review of the three-dimensional integration of “machine learning-multi-factor models-performance evaluation”, providing a theoretical framework for subsequent empirical research, whilst offering practical methodological support to institutional investors and market regulators.

Key Words：

machine learning; ESG quantitative investment; multi-factor models; model optimisation; performance evaluation