Wait, hold on--remember that the correlation factor sets the weighted Coinflip probability!

So it's in fact still perfectly likely for you to get 100% STV for Trump with this model, since I'm drawing from a Binomial distribution with a probability 90% of a vote for Trump in the extreme case, very Republican case.

Does that make sense?

Even with a p=0.1, it's still possible to achieve 100% STV by sheer chance.