Smith Students Turn Cutting-Edge Analytics into Real-World Financial Insight

With the spring semester past its midpoint at the University of Maryland, students at the Robert H. Smith School of Business are completing projects that expand beyond classroom assignments to professional engagements.

Through Experiential Learning Projects (ELP) and capstone courses supported by UMD's Smith Enterprise Risk Consortium, undergraduate and graduate students are tackling real-world questions facing today’s financial markets—using machine learning, artificial intelligence and hands-on analysis of large-scale data in collaboration with leaders from global institutions. The result: work that students can clearly point to as evidence of their skills, preparation, and readiness for careers in finance, analytics and risk management.

From Classroom to Capital Markets

In a recent ELP, a team of MS in Finance and Quantitative Finance (MQF) students partnered with senior executives from the World Bank and T. Rowe Price on an empirical analysis examining how homeowners insurance data can serve as a signal of natural hazard risk—and how those signals affect mortgage default, prepayment behavior, and valuation for mortgage-backed securities (MBS) and credit investors.

Over seven weeks, students progressed from having no prior background in mortgage markets to producing a sophisticated analytical framework grounded in real market data. The team analyzed a dataset of more than 500,000 mortgage loans, merging loan-level information with ZIP-code-level homeowners' insurance data. They then developed and compared a suite of models—including logit, Lasso, Random Forest, XGBoost, and neural networks—to isolate the impact of insurance-related hazard signals on mortgage performance.

Their findings and interpretations reflected a deep understanding of borrower behavior, regional risk exposure, and investment relevance. The final presentation, delivered to Ramon de Castro of T. Rowe Price, translated advanced analytics into clear insights for credit and MBS investors—demonstrating the type of strategic thinking expected in professional financial services roles.

Building on Prior Google-Sponsored AI Research

This work follows a 2025 Google-sponsored ELP in which a team of Smith MQF students used Google’s suite of AI tools—including Gemini, AI Studio and Notebook LM—to analyze more than 25,000 pages of public financial documents. The team developed comparative ratings across banks of varying sizes and fintech companies, demonstrating how generative AI can accelerate large-scale document analysis and support more transparent, data-driven assessments of financial institutions.

Undergraduates Take on Private Credit and Banking Risk

At the undergraduate level, students in the Computational Finance minor are also conducting timely and substantive research. In a capstone project sponsored by Google, students are examining the growing intersection between U.S. banks and the rapidly expanding private credit market—an area drawing increased attention amid heightened fund redemptions and market uncertainty.

Using actual data reported by the top 50 U.S. banks by consolidated assets, students are analyzing bank call reports to assess trends in private credit lending. Early findings indicate:

• A 38% year-over-year increase in private credit lending by major banks
• A 100% increase in loans extended to private equity funds
• Approximately $1.35 trillion in total private credit exposure as of Q4 2025

Rather than relying on aggregate figures alone, students are evaluating exposure on a bank-by-bank basis—examining asset composition, risk concentrations and public disclosures to understand where potential vulnerabilities may or may not exist. For example, at the largest institutions, private credit lending may represent a relatively small share of overall assets, while at others the concentration may be more meaningful.

The project extends beyond balance sheets. Leveraging Google’s generative AI tools, students are also exploring banks’ exposures to AI-related investments and digital assets, analyzing how emerging technologies are reshaping risk profiles across the financial system.

By combining traditional financial analysis with AI-driven research techniques, students are gaining experience with the same tools and questions increasingly used by analysts, regulators, and investors.

A full report will be delivered in May to Google sponsors Pedro Morales, CFA, and Rikesh Patel, CAMS, marking the culmination of the semester and another applied research milestone for Smith students.