I’m not defending modern Nate Silver as a person — he seems to have become a bit of a gambling addict — but in 2016, 538’s model had Trump’s chances at like 33% and the competing models had his chances at 1-2%. It wasn’t a bad model so much as a “when polls are off, they tend to be off in the same direction” situation. The 2016 538 model at least took that into account.