Improving the Interpretability of Asset Pricing Models by Explainable AI: A Machine Learning-based Approach

Abstract

The study examines the integration of machine learning (ML) techniques and explainable artificial intelligence (XAI) for stock price prediction and the interpretation of predictive models. The aim is to improve the accuracy of short-term price forecasts using advanced models like Random Forest and XGBoost, and to utilise XAI tools such as SHAP and LIME to better understand the contribution of each variable in the predictions. This approach can be particularly useful for investors interested in sustainability-related securities, such as those with high ESG ratings, as it provides a deeper understanding of market dynamics and allows for more informed and transparent investment decisions. The integration of XAI not only enhances prediction accuracy, but also helps mitigate the risks associated with understanding and trusting machine learning algorithms, ensuring that these can be used with greater awareness and control, especially in a complex and regulated context like finance.