Aggregate Investor Attention and Bitcoin Return: The Long Short-term Memory Networks Perspective

Wang, Chen; Shen, Dehua; Li, Youwei

Abstract

Investor attention is a scarce cognitive resource which affects investment decisions, and recent studies suggest that investor attention also have impacts on asset prices. Although Bitcoin is found to be one of the most unpredictable cryptocurrencies with excessive volatilities, researchers are still looking for determinants of Bitcoin prices. In this study, we firstly adopt the Long Short-Term Memory Networks (LSTM) approach to evaluate the effect of investor attention on Bitcoin returns by constructing an aggregate investor attention proxy. We combine both direct and indirect proxies for investor attention, in addition to the Bitcoin trading variables as the LSTM inputs. Our empirical results suggest that the including of attention variables could effectively improve the LSTM's prediction accuracy of Bitcoin returns, whereas direct proxies, i.e., Google Trends and Tweets, contain more valuable information to further improve the LSTM's forecasting capacity.

Citation

Wang, C., Shen, D., & Li, Y. (2022). Aggregate Investor Attention and Bitcoin Return: The Long Short-term Memory Networks Perspective. Finance research letters, 49, Article 103143. https://doi.org/10.1016/j.frl.2022.103143

• Journal Article Type: Article
• Acceptance Date: Jul 8, 2022
• Online Publication Date: Jul 9, 2022
• Publication Date: 2022-10
• Deposit Date: Jul 10, 2022
• Publicly Available Date: Jul 10, 2023
• Journal: Finance Research Letters
• Print ISSN: 1544-6123
• Publisher: Elsevier
• Peer Reviewed: Peer Reviewed
• Volume: 49
• Article Number: 103143
• DOI: https://doi.org/10.1016/j.frl.2022.103143
• Keywords: Investor attention; Bitcoin; Machine learning; LSTM; Social media; Google Trends; Tweets
• Public URL: https://hull-repository.worktribe.com/output/4028194
• Publisher URL: https://www.sciencedirect.com/science/article/pii/S154461232200366X