Abstract
We design, implement, and evaluate a decision support system that combines fine-grained signals derived from news and social media with bitcoin prices using a Long Short-Term Memory (LSTM) neural network. Through this artifact, we construct a portfolio to trade bitcoin in three stages. In the first stage, signals and prices are used as inputs for the LSTM model to predict bitcoin prices. In the second stage, we identify the signals that are the most effective predictors of bitcoin prices and then combine these signals to generate predictions that outperform market benchmarks. In the final stage, we assess portfolios based on their financial performance relative to these benchmarks. This design artifact introduces a method to harness the inherent heterogeneity of online news and social media for bitcoin price prediction.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
