Capital Markets Tomorrow: Modeling and Computational Issues

Scientific project

Training

• 1 postdoctoral researcher + 3 CIFRE PhD positions within the Chair’s ecosystem

Scientific Output

• Around ten publications on the program’s main topics

Main Objective

• Applying supervised learning on simulated financial data
• The Chair is situated at the intersection of investment banks’ growing computational needs, driven by increased regulation, and machine learning techniques that can serve this purpose.

Recent Applications

• Sensitivity calculations for hedging and risk analysis of CVA
• Quantification and management of model risk in an XVA hedging valuation adjustment framework
• A fast regression and neural quantile regression algorithm for FVA and KVA calculations
• Statistical learning of conditional value-at-risk and expected shortfall: a mathematical, algorithmic, and numerical study
• Quantitative analysis of the convergence of statistical approximation algorithms optimized for value-at-risk and expected shortfall
• Static hedging of multi-underlying derivative products using vanilla baskets: a mathematical study and neural network–based numerical approaches
• (Ongoing) Creation of a reference dataset for training purposes, intended for practitioners and academics in the field and beyond

Teaching

• Contributions to the XVA analysis course in M2MO and the derivatives course in M2ISIFAR (Université Paris Cité)