A dynamic analysis of stock markets through multivariate latent Markov models

A correct classification of financial products represents the essential step to achieve optimal investment decisions. The first goal in portfolio analysis should be the allocation of each asset into a class which groups investments with an homogenous risk-return profile. We address this objective by means of latent Markov models, which also provide the probabilities of switching between the different phases of financial markets. Our results allow both to discriminate the stock’s performance and to assess the stock’s future dynamics.