Dow Jones Index change prediction using text mining

• Ano de defesa: 2018
• Autor(a) principal: Vale, Marcos Neves do
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal do Rio de Janeiro; Brasil; Instituto Alberto Luiz Coimbra de Pós-Graduação e Pesquisa de Engenharia; Programa de Pós-Graduação em Engenharia Civil; UFRJ
• Palavras-chave em Português: Finanças; Reconhecimento de texto; Mineração de dados; CNPQ::ENGENHARIAS::ENGENHARIA CIVIL
• Link de acesso: http://hdl.handle.net/11422/13421

The recent advances in data and text mining techniques are enabling new research on financial market prediction (TMFP). The purpose of this work is to present a new prediction model for Dow Jones index trends throughout the day. The model was developed using RapidMiner along with SQL scripts. The process uses existing text mining processes and a new alignment technique that is briefly made by picking up the news published by YahooFinance and Google Finance corresponded to the 5 stocks with highest trading volume in each minute. The quality of the model is measured by Precision, Recall and F-Measure indices. The results obtained were excellent and surpass existing techniques today and also in the literature for this purpose. The model proved to be robust and efficient, demonstrating that the use of text mining techniques along with the correct strategy applied in the financial market is an alternative to be considered and contributes to the state of the art in this area of research.