In the analysis and prediction of many real-world time series, the assumption of stationarity is not valid. A special form of non-stationarity, where the underlying generator switches between (approximately) stationary regimes, seems particularly appropriate for financial markets. We introduce a new model which combines a dynamic switching (controlled by a hidden Markov model) and a non-linear dynamical system. We show how to train this hybrid model in a maximum likelihood approach and evaluate its performance on both synthetic and financial data.
Bibliographical noteThis is a preprint of an article submitted for consideration in the European Journal of Finance © 2001 copyright Taylor & Francis; European Journal of Finance is available online at: http://www.informaworld.com/openurl?genre=article&issn=1351-847X&volume=7&issue=4&spage=289
Azzouzi, M., & Nabney, I. T. (2001). Dynamical local models for segmentation and prediction of financial time series. European Journal of Finance, 7(4), 289-311. https://doi.org/10.1080/13518470110071155