Loading...

How People's Sentiment and Attention Affect the Return of Bitcoin?

Dolatzadeh, Hirad | 2022

97 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 55736 (44)
  4. University: Sharif University of Technology
  5. Department: Management and Economics
  6. Advisor(s): Aslani, Shirin; Talebian, Masoud
  7. Abstract:
  8. With the huge growth of cryptocurrencies in recent years, the attention of investors has been drawn to predict and invest in it. Because of the high volatility of this market, which includes a large share of Bitcoin, there is a need for good forecasting in it. Although past studies have been able to accurately predict the price of Bitcoin using fundamental variables and variables related to the blockchain network, less attention has been paid to the use of variables related to investor sentiments in this market. In this research, variables widely used in the literature that show the emotions and attention of investors, such as sentiment analysis of Twitter texts, Google search index, Wikipedia page views, and market fear and greed index, have been collected and the next day's return is predicted using them. For this, we have used different machine learning models that can have better out-of-sample performance than econometric models. To compensate for the interpretability limitation of these methods, the SHAP model has been used, which interprets the effect of the used variables using the Shapley value method. There is no use of these variables together and their interpretation in the literature. We were able to predict the return of Bitcoin in the test period with 62% accuracy. Also, the best predictor variable is the change in fear and greed index. We found that high values of this variable and the market moving from fear to greed cause negative returns in Bitcoin.
  9. Keywords:
  10. Bitcoin ; Machine Learning ; Prediction ; Sentiment Analysis ; Analytical Interpretation ; Econometrics

 Digital Object List

 Bookmark