I am working on a hmm for financial time series price data using the hmmlearn package (http://hmmlearn.readthedocs.io/en/latest/). I would like to implement a 2 or three state model to fit my data on. However I would like to have different distributions depending on the hidden state. It seems you can implement a custom emission probability using the _BaseHMM override. However I am not completely sure how to do this form the documentation. Are there any available example or could somebody please provide how to set e.g. two different emission probabilities for each state in a hmm framework like this?
Many thanks in advance.
