I am training markov regime model on On a time series data using code below .
model = MarkovRegression(returns, k_regimes=2, switching_variance=True).fit()
display(model.summary())
I am tuning model with parameter trying all possible combinations and different values of k_regimes.
But I want to know what is evolution metric to select best model. As I read from a blog that All p-values of the regime self-transition coefficients and the regime transition probability matrix's coefficient is smaller than 0.05, indicating the model should be able to classify the data into 2 different volatility regimes. Here number of regimes (k) 1<k<N . There could be many P values for different values of K . So How to select best model . Is there ay other model evaluation metric
Thanks