Fast Calibration of Implied Volatility Model

ETUDE INTERNE
AUTEURS : ERNESTO LOPEZ FUNE, JONATHAN VONGDARATH

Calibrating financial models becomes increasingly challenging as they become more and more complex. This step is, nevertheless, crucial for guiding professionals in minimizing the risks of inaccurate pricing or hedging results, which can lead to significant financial losses. In the framework of the SVI model, its complexity can be time consuming due to the non-convex nature of the problem, which is why in this article we investigated the performance of several optimization algorithms, including machine learning regression ones, in terms of accuracy and computational efficiency. Furthermore, our off-line computations enable us to adjust the parameters in real time, without the need for time-consuming recalibrations. This provides greater flexibility and adaptability to changing market conditions, which is crucial for financial institutions seeking to stay ahead of the curve.