Re: the applicability of transition-based methods, I think it's the other way around - TGM or history-based randomisers are the best candidates...
Separate names with a comma.
