Co-movement forecasting between consumer sentiment and stock price in e-commerce platforms using complex network and entropy optimization

Stock price and consumer sentiment consistently serve as pivotal economic indicators for the performance and growth of e-commerce enterprises. It is essential to comprehend and forecast the co-movement between the two to inform financing and investment decision-making effectively. Prior research has focused on predicting individual indicators, but not much of them attempt to forecast their co-movement. We propose a novel Rule Combination based on Bivariate Co-movement Network (RC-BCN) approach for bivariate co-movement forecasting. Bivariate co-movement features extracted utilizing the BCN’s topological nature instruct the entropy optimization in order to enhance the RC-BCN’s predictions. We conduct four sets of experiments on 1,135 data sets from JD.com between 2018 and 2022, where consumer sentiment is measured using text sentiment analysis of online reviews. The results indicate that RC-BCN’s prediction accuracy reaches at most 91% under distortion preference and is improved by 18% compared without entropy optimization. This study highlights the value of complex network and entropy theory in forecasting bivariate co-movement for e-commerce enterprises.