Deep Calibration of Financial Models: Turning Theory Into Practice

Abstract

The calibration of financial models is a laborious, time-consuming and expensive task, which needs to be performed frequently by financial institutions. Recently, the application of artificial neural networks (ANNs) for calibration has gained interest. This paper provides the first comprehensive empirical study on the application of ANNs for calibration based on observed market data. We benchmark the performance against a real-life calibration framework. We show that the results of an ANN based calibration framework are very competitive and derive guidelines for its practical implementation to enhance and accelerate managerial decisions. Furthermore, we show that our calibrated parameters are more stable over time, enabling more reliable risk reports and business decisions.

Keywords: Decision Support, Global Optimization, Deep Learning, OR in Banking, Model Calibration

JEL Classification: C23, G21, G33, E43