New insights on hidden Markov models for time series data analysis

The goal of this thesis is to develop novel methods for the analysis of financial data by using hidden Markov models based approaches. The analysis focuses on univariate and multivariate financial time series, modeling interrelationships between financial returns throughout different statistical methods, such as graphical models, quantile and expectile regressions. The dissertation is divided into three chapters, each of them examining different classes of assets returns for a comprehensive risk analysis. The methodologies we propose are illustrated using real-world data and simulation studies.