Examination of Calendar Effect and Sentiment Analysis on Bitcoin
Open Access
- Author:
- Liu, Chen Han
- Area of Honors:
- Finance
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Brian Spangler Davis, Thesis Supervisor
Brian Spangler Davis, Thesis Honors Advisor
Quanwei Cao, Faculty Reader - Keywords:
- Sentiment effect
Bitcoin
Efficient market
Regression
Power ratio
Python
Calendar Effect
Sentiment Analysis
Bitcoin
Efficient Market Hypothesis
Regression
Power Ratio
VADER
Python - Abstract:
- This research investigate if Bitcoin exhibits the calendar effect and how sentiment affects Bitcoin’s pricing. I examine the day-of-the-week effect and intraday effect using regression analysis with dummy variables and power ratio analysis. Bitcoin’s abnormal returns on Fridays and early mornings, as seen from regression analysis, prove the existence of these two effects, but power ratio analysis shows no anomalies between weekdays and intraday. Sentiment analysis is studied based on data that include sentiment scores and affiliated information from over 4 million posts on Twitter. The findings offer evidence that the number of tweets posted, unweighted sentiment compound score, and sentiment compound score weighted by retweets correlate more with Bitcoin’s returns. The investigation also presents that the correlation of sentiment parameters such as weighted and unweighted compound scores on Bitcoin’s price is not consistent over the time horizon, which forms a positive leading index, but turns negative as a simultaneous or lagged index. The degree of influence of Twitter information over Bitcoin’s price is less significant 20 minutes before and after a tweet. The analysis results explain why recent research about calendar anomalies and sentiment analysis in Bitcoin’s returns contradict each other. Lastly, I also note that a tweet’s information correlates more with volatility than with Bitcoin returns.