Machine learning application in financial return predictions

This project aims to predict the international stock returns by the help of artificial intelligence (AI) algorithms and the growing financial data sets. The main challenge of the AI application is a lack of interpretability, also known as a black box, in popular AI algorithms. This project will employ the newly developed and explainable AI technology to address the interpretability concern. Second, we will compare the explainable AI model with the widely used black-box model especially in the dimension of their out-of-sample predictability and profitability. Third, we will validate the AI models following a statistical analysis in order to find the most robust prediction model for predicting stock returns, as there was some anecdote evidence indicating the failure of return predictions when the financial markets have high return volatilities. We expect to find that the explainable AI technology can generate predictions that are robust to boom and recession periods. This finding will largely reinforce the confidence of economists and policymakers in AI methods so as to improve their decision-making procedures.

Faculty Supervisor:

Hsuan Fu

Student:

Partner:

Swansea University

Discipline:

Business

Sector:

Education

University:

Université Laval

Program: