Three SEPS Research Projects Receive SNSF Funding

The Swiss National Science Foundation (SNSF) supports outstanding research across Swiss universities and institutions. We are proud to announce that three projects by SEPS faculty members have been selected for funding, underscoring the quality, impact, and innovation of research conducted at our School.

We warmly congratulate our researchers and their collaboration partners on this achievement.

Prof. Francesco Audrino – “Financial Social Networks”

Grant No. 240150 | CHF 601’810
Project details on SNSF website

Prof. Francesco Audrino’s project investigates how information shared by retail investors on social media affects financial markets. By measuring and analyzing Financial Social Networks, the research explores how online interactions influence asset volatility, market risk, and price dynamics. Using network-based forecasting models, the project aims to improve risk assessment tools and provide early-warning indicators for market instability—supporting both professional and individual investors.

Prof. Roy Gava (with Mariana Valente and Matias Dewey) – “Illegal Transactions in the Digital Age: Emerging Practices and Regulatory Intermediaries in Latin America”

Grant No. 10007134 | CHF 856’340
Project details on SNSF website

This interdisciplinary project examines how digital technologies are reshaping illegal markets in Latin America. By studying how social media platforms, messaging services, and payment providers are misused for illegal transactions, the research explores the evolving role of regulatory intermediaries. Combining perspectives from legal studies, political science, and sociology, the project offers insights into how digital regulation and enforcement influence illicit economies in Argentina, Brazil, and Mexico.

Prof. Tobias Sutter – “Convex Q-Learning: Optimal Policy Learning under Distribution Shifts”

Grant No. 10.006.581 | CHF 303’000
Project details on SNSF website

Prof. Tobias Sutter’s project develops a new computational framework that enhances how machines learn optimal decision-making strategies under uncertainty. By combining convex Q-learning with distributionally robust optimization, the research provides a statistically grounded approach to reinforcement learning that remains reliable even when underlying data distributions change. The project bridges theory and application across economics, energy systems, and control problems, while releasing open-source software to foster further research.