Big Data in Financial Risk Management: Evidence, Advances, and Open Questions. A Systematic Review

LEONIDAS THEODORAKOPOULOS^*Alexandra TheodoropoulouAristeidis Bakalis

• Department of Management Science and Technology, Panepistemio Patron, Patras, Greece

The intersection of big data analytics and financial risk management has catalyzed a new era of methodological innovation and organizational transformation. Yet, despite a surge in research activity, the literature remains fragmented, with persistent blind spots in comparative effectiveness, cross-sectoral generalizability, and the operationalization of non-traditional data. This systematic review, grounded in the PRISMA 2020 protocol, synthesizes evidence from 21 studies published between 2016 and June 2025. The analysis maps the comparative strengths and trade-offs of neural networks, ensemble learning, fuzzy logic, and hybrid optimization for managing credit, fraud, systemic, and operational risk. Findings reveal that while advanced machine learning models routinely deliver strong predictive performance, real-world deployment remains uneven, concentrated in Chinese and European banking and fintech, and rarely extending to broader regulatory or sectoral contexts. Integration of alternative and unstructured data, such as IoT signals and behavioral analytics, remains experimental, with substantial technical and governance challenges. Methodological diversity is a hallmark of the field, but systematic benchmarking across risk types and organizational settings is rare, and the imperative for explainability is only beginning to be addressed. This review highlights an urgent need for comparative, cross-jurisdictional research, robust field validation, and open science practices, establishing a critical agenda for bridging the gap between technical advances and operational impact in big data–enabled financial risk management.