The role of Twitter in the U.S. financial market

• Ano de defesa: 2018
• Autor(a) principal: Ramacciotti, Fernando Martinelli
• Orientador(a): Giovannetti, Bruno Cara
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Palavras-chave em Português: Redes sociais on-line; Mídia social; Mercado financeiro - Estados Unidos; Comportamento - Avaliação
• Palavras-chave em Inglês: On-line social network; Social media; Financial market - United States; Behavior - Assessment
• Link de acesso: http://hdl.handle.net/10438/24591

Significant market events in the financial market will only occur if there is synchrony among large groups of people and the media is the main vehicle to it. Previous works could find some relationship between newspapers and financial market indicators. This work revisits, in some sense, the findings of the traditional literature on financial market investors’ behavior and its relationship to the news, but now in the Modern Era context of social media. The main goal of this work is to identify if the overall sentiment of Twitter users has some relationship with the financial market. We created a database that joins non-structured data from tweets and structured time series data related to the S&P500 index, its trading volume and implied volatility (measured by the VIX). The non-structured data was processed in order to categorize each word from each tweet into several semantic categories defined by the Harvard-IV psychological dictionary. Then, two sentiment indexes were created via Principal Component Analysis (PCA): Engagement Factor (EF) and Optimism Factor (OF). Using Vector Autoregressive (VAR) framework, we simultaneously estimated the effects of sentiment on financial market variables and vice versa. Our results indicate that Twitter users seem to respond to financial market events, i.e. their sentiment is a consequence of financial events and do not have predictive power.