Calibration of an agent based model for financial markets

Abstract

Agent based model are very widely used in different discipline. In financial markets they can explain very well known features called stylized facts and fit statistical properties of data. For such a reason in predicting future price movements they could perform better than standard models using gaussianity. At this scope calibration and validation in order to choose the model and the model parameters are very essential issues. However calibrating such models is a hard issue to tackle and not yet very well considered in literature. The present paper presents the attempt to calibrate the Farmer Joshi model by a Nelder Mead algorithm with threshold. Different objective function are considered in order to identify the best choice. © 2011 Springer-Verlag Berlin Heidelberg.