Date of Publication :29th June 2023

Abstract: The future performance of stock markets is an essential factor for portfolio creation. With the advancement of machine learning techniques, new possibilities have opened up for incorporating prediction concepts into portfolio selection. The paper proposes a hybrid approach, involving machine learning algorithms for stock return prediction and a mean-VaR model for portfolio selection, as a unique portfolio construction technique. Two machine learning regression models, XGBoost and linear regression, are used for stock prediction, and a novel optimizer, Grey Wolf optimizer, is employed for parameter optimization with both XGBoost and linear regression. The results show that the mean-VaR model with linear regression prediction produces better results than the mean-VaR model with XGBoost prediction.

Reference :