Predicting Corporate Solvency using Sentiment Analysis of Risk Analyst Textual Assessments

Credit scoring has long been crucial to financial institutions’ decision-making processes, providing a means to evaluate the risk of extending credit to consumers and businesses. With recent advances in artificial intelligence (AI), decision-making algorithms have become more precise and efficient, leading to a wide diversity of applications across various industries. The financial sector has increasingly adopted AI-powered algorithms and advanced data analytics for tasks such as credit scoring and pricing. Traditional credit risk analysts, however, have relied on qualitative data, such as news articles and press releases, to assess a company’s situation before granting credit. This paper introduces a novel framework that combines credit risk text assessments made by credit analysts with a pre-trained financial sentiment analysis model. By comparing the sentiment analysis of the text with the actual ratings given by risk analysts, we demonstrate the effectiveness of our proposed framework.