Title: Use of social media big data for predicting credit rating changes of companies
Authors: Leonie Tabea Goldmann - University of Edinburgh (United Kingdom) [presenting]
Raffaella Calabrese - University of Edinburgh (United Kingdom)
Jonathan Crook - University of Edinburgh (United Kingdom)
Abstract: Predicting the financial performance of companies using various data sources has been the focus of a large number of studies. However, current models can still be improved to achieve even better predictions. Several contributions are made. First, a new method is introduced to more accurately predict the probability that a companys credit rating changes. More specifically, the vast increase in social media data put out by companies is exploited which enables to develop more predictive models than those developed in the past. By analysing the tweets put out by different companies, we show that there is a correlation between specific words in the tweets and the credit rating. This is done by using differential language analysis, a method which is usually used in psychology and health studies and has not been used in a financial context before. Second, we will show that using these words in a predictive model with additional twitter variables, such as the tweet frequency, sentiment score and length leads to an increase in predictive power when comparing with a model not containing these predictors.