ABSTRACT
The motivation for this chapter is to introduce the use of hidden Markov models for financial modelling, specifically for trading applications. Hidden Markov models provide an analytically and computationally tractable method of modelling a wide range of complex financial phenomena. Hidden Markov models also have the advantage of undemanding modelling assumptions and are consistent with a range of fundamental theories in finance, such as the efficient market hypothesis.
In this chapter we introduce hidden Markov models, explaining their modelling properties and advantages, particularly with respect to investment and trading. We then discuss the computational implementation using empirical data. Specifically, we introduce the calibration method that is most frequently applied in economic and financial research literature (Hamilton filter) but also a less well-known method, the Baum–Welch method. Although the Baum–Welch method has many superior qualities to the Hamilton filter and is generally more frequently applied in the sciences and engineering, it has been less commonly used in finance. We then give a practical study of both methods applied to financial markets and a survey of current hidden Markov model applications. This chapter will be of interest to those wishing to utilize hidden Markov models and improve their models.
