This study aims to overcome some of the limits of the sustainability ratings of firms (known as ESG scores - Environment, Social and Governance) by using artificial intelligence techniques to better identify the components of these metrics that contribute most to identifying efficient portfolios.
The proposed methodology selects indicators which, both as regards sustainability factors as a whole and the environmental component alone, contribute significantly to identifying portfolios with better risk-return profiles (i.e. they provide greater returns for a given risk or present lower risks for a known return).
