Author: Noemie Claret, Research Analyst Intern
On the weekend of the 26th of February, FNA participated in the LSE (London School of Economics and Political Science) ‘s Data Science Weekend 2022. This two-day ‘datathon’ organized in partnership with the LSE Data Science Institute was open to students from all UK universities, who were invited to solve analytics and modelling challenges provided by the society’s industry partners.
The application of Data Science and Network Analytics is prevalent in disciplines studied at the LSE, from econometrics modules to philosophy courses, leveraging network science to evaluate G.Tononi’s Integrated Information Theory of Consciousness. The LSE Data Science Society (DSS) is a student-led organization providing a platform to express this enthusiasm, and Data Science Weekend is the society’s annual flagship event.
The challenge FNA posed to students followed the theme of Kimmo Soramäki and Amanah Ramadiah’s December masterclass on the financial applications of network science to explore FNA’s BIS Cross Border Banking Exposures dashboard on the G20 Monitor.
The dashboard monitors the evolution of consolidated foreign claims by banks organized by their primary nationality. It aims to understand banks’ foreign exposures and the ramifications on global financial stability. Identifying the exposures of countries like Germany and France to Greece in 2008 can help authorities understand how shocks might propagate globally and inform the appropriate ex-ante response to future shocks.
The challenge, led by FNA’ers Will Towning, Giuseppe Matera, Linda Esmeralda Rojas Parra and Lubos Pernis, offered students a choice of two research options to explore:
- Develop a model for anomaly detection in cross border banking exposures that could help authorities anticipate potential financial stability implications.
- Conduct exploratory analysis of the network encapsulating global banking exposures data and develop insights of relevance to central banks and financial authorities. This involves the study of clusters/ communities, systemic risk and evolution of network properties over time.
Winners of the challenge, Mukun Liu and David Chen, both undertook the first research option, proposing a proof-of-concept measure for global systemic risk. Their work demonstrated a thorough understanding of network science and appropriate methods, including SinkRank or Markov chains. Their metric, when back-tested, succeeded in flagging the EU Sovereign Debt crisis in historical data.
All winners will be invited to work with FNA on our internship programme.
During this ‘datathon’, students also had the opportunity to attend workshops tailored to the theoretical demands of the datathon challenges – notably, AWS Senior Developer advocate Sean Tracey delivered a workshop on leveraging deep neural networks for time series data anomaly detection. FNA also provided support and guidance to teams in the development stages of their project during mentor office hours.
FNA’s Academic Partnerships Programme is a crucible for collaboration with universities. It aims to develop talent in advanced network analytics and foster the next generation of innovative solutions to challenges financial institutions and policymakers face.
For more information on the LSE Data Science Weekend, go to: https://lnkd.in/dtcmjzGW.
To read more about FNA’s G20 Monitor, go to: https://lnkd.in/dm6N9Bb6
To learn more about FNA’s Academic Collaboration Programme, go to: https://lnkd.in/duMM2VRu