Option pricing PDE solution and calibration with neural networks
We will outline the calibration of a financial asset price model in the context of financial option pricing.
Particularly, to provide an efficient calibration framework, a data-driven approach, by means of an Artificial Neural Network (ANN), is proposed to learn the solutions of financial option pricing models and to reduce the corresponding computation time significantly.
This ANN-based method is extended to calibrate financial models. Specifically, fitting model parameters is formulated as training hidden neurons within a machine-learning framework.
The rapid on-line computation of ANNs combined with a flexible global optimization method (i.e. Differential Evolution) provides us fast calibration without getting stuck in local minima.
Prof. dr. ir. Kees Oosterlee is part-time full professor at Delft University of Technology in Applied Numerical Mathematics. He is also senior scientist at CWI – Centrum Wiskunde & Informatica — and member of CWI’s management team. He was the editor-in-chief of the Journal of Computational Finance, between 2013-2018, and the Chair of the Dutch-Flemish Society for Computational Sciences (SCS), which has 400 members, from 2014-2019. He has also been coordinator of various EU Marie Curie projects on financial risk management. His research focus is on developing and analysing novel, robust and efficient algorithms, with a particular interest in computational finance. He has written two textbooks, the most recent one is “Mathematical Modeling and Computation in Finance” with Lech A. Grzelak, and approximately 150 research articles.