Using expectile regression with latent variables for digital assets

In this paper we introduce a linear expectile hidden Markov model with the goal of modeling the entire conditional distribution of asset returns and, at the same time, to grasp unobserved serial heterogeneity and rapid volatility jumps typical of financial time series. The temporal evolution of asset returns is captured by introducing time-dependent coefficients evolving according to a latent discrete homogeneous Markov chain. To implement the procedure, we consider the Asymmetric Normal distribution as a working likelihood for the estimation of model parameters and the estimation procedure is carried out using an efficient EM algorithm. The empirical application investigates the relationship between daily Bitcoin returns and major world market indices.