Thijs van Pelt @Thijs_Van_Pelt

I am wanting to use the works of Article Bayesian Portfolio Selection in a Markov Switching Gaussian ...

to attempt to construct portfolios that can better capture stylized facts of returns (non-normality and fat tails) and react better to varying economic conditions. To try and understand the sampling process I am trying to derive the conditional posteriors for the Markov Switching Gaussian Mixture model to help with intuition. I am struggling with how to deal with the different states and regimes though. I assume one would use the data augmentation technique where a joint posterior is constructed with the data, states and regimes? Knowing the form of the joint posterior would already simplify matters greatly. Additionally, what other models could be used to model returns more realistically? Thankful for any help!

Kind regards.