Publication Date

Spring 2020

Degree Type

Master's Project

Degree Name

Master of Science (MS)


Computer Science

First Advisor

Thomas Austin

Second Advisor

Mark Stamp

Third Advisor

Robert Chun


Bitcoin, Twitter, cryptocurrency, sentiment analysis, machine learning


‘‘Cryptocurrency trading was one of the most exciting jobs of 2017’’. ‘‘Bit- coin’’,‘‘Blockchain’’, ‘‘Bitcoin Trading’’ were the most searched words in Google during 2017. High return on investment has attracted many people towards this crypto market. Existing research has shown that the trading price is completely based on speculation, and its trading volume is highly impacted by news media. This paper discusses the existing work to evaluate the sentiment and price of the cryptocurrency, the issues with the current trading models. It builds possible solutions to understand better the semantic orientation of text by comparing different machine learning techniques and predicts Bitcoin trading price based on Twitter feed sentiment and additional Bitcoin metrics. We observe that the statistical machine learning model was able to better predict the sentiment of Twitter tweet feed compared to the advanced BERT model. Using Twitter feed sentiment and additional Bitcoin metrics, we were able to improve the prediction of bitcoin price compared to only using bitcoin’s previous day closing pricing.