IDLab Workshop (original) (raw)

Viktor Krakovich presented the results of the study "News Sentiment as a Predictor of Bank Failure"

The topic of bankruptcy of financial organizations is always relevant, however, in this study, the classical model is supplemented by an indicator of the sentiment of news texts posted on the Internet. The paper reveals that the tone of the news is a significant determinant of the predictive model for assessing the probability of bankruptcy of credit institutions. It has been established that a cluster containing news about punishments from the Central Bank of Russia reduces the likelihood of a bank going bankrupt, while clusters that combine more positive topics lead to an increase in the likelihood of a bank closing ceteris paribus.