Impacto das opiniões dos grandes influenciadores e o efeito Twitter sobre os investidores individuais

• Autor(a) principal: Oliveira, Bruno Alves de
• Data de Publicação: 2022
• Tipo de documento: Trabalho de conclusão de curso
• Idioma: por
• Título da fonte: Repositório Institucional da Universidade Federal do Ceará (UFC)
• Texto Completo: http://www.repositorio.ufc.br/handle/riufc/70409
• topic: Análise de Sentimento; Twitter; Mercado Financeiro; Modelo de Previsão

The motivation for this research is the possibility of understanding interpersonal interactions in an increasingly instantaneous way. In turn, the academy has contributed to using Twitter as a sophisticated tool to capture real-time information from Internet users (Fan and Gordo, 2014, p. 76). The objective of this study was to identify, through the messages that are posted on Twitter, how the information that is disseminated online is associated with movements that occur in the Brazilian market, specifically with regard to returns and trading volume. The collection of data related to Twitter was done through the Tweepy library. The financial data were obtained through the Yahoo Finance database. The sentiment index was used after scraping the major financial market influencers on the social network and their textual interactions (tweets) with their audience. Accounts with the potential to influence individual investors were extracted from the profile of journalist Sérgio Charlab, who has an algorithm responsible for carrying out this ranking. The attribution of the feeling was done through machine learning, through the Textblob library. To meet the general objective of the research, two models were estimated, the first was a Simple Linear regression model and the second a VARMAX(8, 0) time series model. The analysis and interpretation of the data made it possible to perceive that, in general, a simple linear modeling would not benefit the individual investor in creating an investment strategy. The study starts from the premise that the individual investor is the one with little ability to use robust modeling to find an investment strategy through sentiment analysis. It was also found that there is the possibility of analyzing the financial market through non-linear modeling of time series, given the existing potential in the variables created from sentiment analysis, namely subjectivity and polarity. Furthermore, it was seen that the subjectivity of the messages has a predictive potential for the traded volume. Thus, the results are useful to show that there is a relationship between the information that is disclosed on the social network Twitter and the movements of the Brazilian stock market, bringing contributions to the literature. The study also brings practical contributions, since the activities that occur online on Twitter can be used as variables in investment strategies, since these are associated with market movements when the modeling is non-linear.