Identifying Corporate Credit Risk Sentiments from Financial News

Anonymous

Abstract: Credit risk management is one major practice for financial institutions, that helps them measure and understand the inherent risk within their portfolios. Historically, they relied on the assessment of default probabilities (via structural or default intensity models) and used the press as one tool to gather insights on the latest credit event developments of an entity. However, because the current news volume and coverage for companies is generally heavy, analyzing news manually by financial experts is considered a highly laborious task. To this end, we propose a novel deep learning-powered approach to automate news analysis and credit adverse events detection, with the aim of scoring the credit sentiment associated with a company in order to assist credit risk management efficiently. The result is a complete system leveraging news extraction and data enrichment (with targeted sentiment entity recognition to detect companies and text classification to identify credit events), as well as a custom scoring mechanism designed to provide the company's credit sentiment, called Credit Sentiment Score™ (CSS). Additionally, studies are shown to illustrate how CSS helps to gain knowledge about the company's credit profile but also discriminates between defaulters and non-defaulters.