Can you elaborate? MCMC is not really being used to predict states in our library but to create a Markov chain that resembles a posterior distribution.
Are you perhaps thinking about Markov Matrices that represent transition probabilities between states at different time steps?