Look, polls didn't forecast with perfect accuracy what would happen in the most recent American presidential election. But you know who was saying that, with numbers and words, this whole time? Nate Silver.
Look at their distribution of possible outcomes. They had hundreds of electoral college votes across the bulk of their distribution. And there are only 538! Those are crazy error bars. They weren't just using one poll, either, they were using all of them, combining them and trusting them in the best way they new how. And how much information did they have, at the end? Not an insane amount. And they were extremely honest about it, and made no attempt to project confidence they didn't have.
That being said: their odds of a Biden win smashed the prediction markets. The prediction markets had Trump at 1/3 chance for months, swung way up to predicting a very likely Trump win, then swung way down, all on election night. They settled, oddly enough, almost exactly on the numbers 538 produced months before. This was after votes were already being recorded, states already being called, prediction markets were, at that moment, endorsing the odds Nate Silver gave from the beginning.
Look, pollsters should have an insane amount of humility about this. But Nate Silver isn't a pollster. And people shouldn't give him any grief. I think it's intellectually lazy for people to say they predicted better than 538, as some otherwise reasonable people have claimed.