Classifying earnings conference calls

• Autor(a) principal: Salomão, Antônio Elias Xavier
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/143253
• topic: Machine learning; Event study; Natural language processing; Domínio/Área Científica::Ciências Sociais::Economia e Gestão

This study examines whether it is possible to classify the sentiment of earnings conference calls of U.S. publicly traded companies not by using standard metrics such as standardized unexpected earnings, but by but using the general sentiments, opinions and affective states present in the earnings calls. This classification task is attempted using the naïve Bayes classifier. Results show that due to the high signal to noise ratio present in the earnings calls, the classifier of choice was unable to adequately distinguish positive earnings calls from negative ones and vice-versa. Nonetheless, the classifier did shed light on the extent to which company CEOs tend to be overly optimistic when partaking in the conference calls.