Predicting Credit Ratings for FinTech Companies Using Ensemble Learning and Synthetic Data

Most FinTech firms remain unprofitable, raising concerns that they could trigger global financial instability. A random subspace ensemble classification and decision tree ensembles with an ADASYN algorithm are applied to predict credit ratings for FinTech firms. Due to the limited number of publicly rated FinTech firms with an issuer-level credit rating, our dataset is both small and highly imbalanced. Our approach improves the predictive capability of individual classifier algorithms, achieving an F1 Score between 90% and 97%. Our findings suggest that market capitalization is the most influential factor, followed by profitability, coverage, solvency, and growth metrics. These results contribute to the literature on FinTech credit rating prediction and provide practical insights for investors, regulators, and policymakers by generalizing across multiple countries and business models. JEL classification: C45, G17, G23, G24, G32