I thought they were basically pretty close as to the popular vote. I don't think it would be easy for them to be accurate on the electoral vote. You'd need several timely state-wide polls (by reliable pollsters) to get a real electoral vote guesstimate. 538 got the popular vote very close, but only gave Trump a 30% of winning the electoral vote. Even still, 30% is more likely to happen than hitting red twice in a row on a roulette table.
The pollsters also seemed pretty close on the 2018 mid-terms. Except Rasmussen, of course.