Gaussian Processes for Financial Time Series, a C++ Implementation

Authors: Sebastian Ferrando and Massimo Pascazi, Ryerson Polytechnic University.

This technical report gives a brief overview of regression with Gaussian Processes (GP) and describes an implementation of GP models which can be of interest for empirical modelling of financial time series. Our implementation maximizes the likelihood of the GP by using a global optimization algorithm. The likelihood is evaluated by means of the innovations algorithm. There is also available a software package written in C++ which implements some of the models discussed in this report.