Markov-Switching Dynamical Systems on Stock Volatility

 

The Markov-switching phenomenon for stock volatility is widely studied, however, little attention has been paid to switching between different model classes, such as a Heston model and a 3/2 model. In this article, we propose a novel heteroskedasticity-based E-M algorithm for inferring such a model-class switching model, and a testing procedure to demonstrate the existence of such a switching phenomenon. We verify our proposed method through both simulation and real data analysis of both the implied and realized volatility of the stock market.