Análise de sentimento dos tweets e sua influência sobre o preço do bitcoin

Abstract

Social media has the power to influence the feelings and decisions of individuals, and in the financial market, the situation is no different. In recent years, the cryptocurrency Bitcoin has experienced significant price fluctuations due to various reasons. Based on these observations, a study was conducted on how investors' sentiments, as expressed in their Twitter posts, influence the closing price of the Bitcoin cryptocurrency. Firstly, a database containing multiple variables and a substantial number of tweets was found on the Kraggle website, covering the period from 10/02/2021 to 06/01/2023. From this database, the NLTK library was used for data preprocessing, which includes both semantic and statistical analysis. The VADER library was employed to measure the polarity of sentiments expressed in the posts. Additionally, a database containing information about Bitcoin during the same period was utilized. By evaluating the sentiments of tweets and the data related to the closing price of Bitcoin, an Autoregressive Distributed Lag (ARDL) regression model was created to assess whether the selected variables influence the price of Bitcoin. The results revealed that the selected variables do have an influence on the price of Bitcoin. Consequently, the hypothesis that tweets influence the price of Bitcoin was not rejected.