A Class of Gaussian Hybrid Processes for Modeling Financial Markets

Abstract

This paper proposes a one-factor model of financial markets using a class of Gaussian process that can be decomposed into a Brownian motion and an Ornstein–Uhlenbeck process. It is shown that this “hybrid” process is obtained as a continuous-time scaling limit of the differenced first-order autoregressive integrated moving average (ARIMA(1,1,1)) process. Parameter estimations using an ARIMA(1,1,1) framework and its variance ratio test show the accuracy of the proposed model. Construction of the one-factor commodity futures price model is presented as an application. A multidimensional extension of the hybrid process is also presented in the Appendix.