Thanks! I hadn't thought of doing Markov Chains since I was focusing on RNN/LSTMs, but I think I'll try that for the next article, see how that goes.