Comment by Sohcahtoa82

The trick with Markov chains is that you don't need to.

Markov Chains are dead simple. There's not really a "training" as much as it's simply reading data and collecting statistics.

They're so simple that you can probably build one nearly as fast as you can read the training data.