Consistent and robust Generative Models for Volatility Surfaces

Implied volatility surfaces are a key concept in financial markets, central to the pricing, risk management, and regulation of financial instruments. Despite their importance, numerous mathematical questions about implied volatility surfaces still confront practitioners and regulators today, such as how to fill in missing values in a given surface.
Previous MITACS-sponsored work has shown that certain types of neural networks show promise for this “surface completion” problem and related tasks. We propose to build on these preliminary results by researching alternative formulations to better meet industry needs, such as guaranteed stable and realistic predictions.
Achieving these technical goals would allow Riskfuel to provide services which help financial institutions and regulators better manage financial risk.

Faculty Supervisor:

Ioannis Mitliagkas

Student:

Partner:

Riskfuel

Discipline:

Engineering

Sector:

Professional, scientific and technical services

University:

Université de Montréal

Program: