In this talk we present data science approach to evaluate, predict and optimize financial regulation. Since the financial crisis in 2007/08 there has been a vibrant discussion on how to reduce systemic risk in the financial system and various legislations have been passed. Yet a decade later, experts still do not agree in their judgement on whether or not current regulation is sufficient in order to prevent a future economic meltdown. If expert judgement is inconclusive, how can data science help?

We introduce a model of financial systems that captures the impact of a financial regulation on all levels of detail - from a single trade to systemic effects - using weighted directed graphs. As virtually all trade data of the real financial system is proprietary, we show how to use random graph algorithms calibrated to realistic distributions to generate financial systems, which are open and completely accessible to scientific research. Finally, we simulate how these financial systems evolve over time under different regulations. This provides a data set on which the impact of financial regulation on systemic risk can be studied by data science techniques such as summarization, visualization and anomaly detection. We will illustrate this method on prominent financial regulation passed since the crisis.

The research heavily relies on open source technologies in Python (networkx, numpy, pandas, matplotlib) and C++ (boost, QuantLib, Open Source Risk Engine). The talk is based on a sequence of papers, see below, and ongoing research. No prior knowledge of finance or regulation is neccessary.

[1] O’Halloran, Sharyn; Nowaczyk, Nikolai; Gallagher, Donal A.. A Data Science Approach to Predict the Impact of Collateralization on Systemic Risk, Available at SSRN: https://ssrn.com/abstract=3090617

[2] O’Halloran, Sharyn; Nowaczyk, Nikolai; Gallagher, Donal A.. Big Data and Graph Theoretic Models: Simulating the Impact of Collateralization on a Financial System, Proceedings of the 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining 2017, ASONAM ’17, p. 1056–1064, https://doi.org/10.1145/3110025.3120989, Available at SSRN: https://ssrn.com/abstract=3090617

[3] Caspers, Peter; Giltinan, Paul; Lichters, Roland; Nowaczyk , Nikolai. Forecasting Initial Margin Requirements – A Model Evaluation, Journal of Risk Management in Financial Institutions, Vol. 10 (2017), No. 4, https://www.henrystewartpublications.com/jrm/v10, Available at SSRN: https://ssrn.com/abstract=2911167