Lots of people doubted him. On here and around the nets.
The baseball geek was absolutely bang on. Fifty for fifty. Once again, Silver is the best poll analyst in American politics. It’s time for everyone who doubts his abilities to suck it up; Silver knows his shit. If you disagree with Silver, you’re wrong.
I’ve been touting Silver for years. I am in awe of his genius.
I think it’s a bit premature to call it 50 for 50. Let’s at least wait until all the results can be tabulated. I’m also interested to see how the popular vote and the Senate races shake out.
Yep. Nate just became God. He can now write his own ticket to pretty much anything he wants.
Congrats Nate.
Has Montana been called yet for Senate? I actually think it is a better test of his model than the Presidential election.
Looks like he was a bit short on the electoral college. Florida is close to getting called for obama.
Charles Krauthammer already jabbering on Fox about how Romney lost because he was too liberal.
The elector count is an average, not a projection of the most likely count. I think there was some discussion (there, here, or both) about what sort of map it would take to hit it exactly, and the scenarios were a bit odd.
Well, Florida alone remains uncalled.The very fact that Florida is the last race left going down to the wire, Obama is leading ( and will very likely win based on where the outstanding vote total will come from ) and that Silver’s final model spat out a 49.797% chance Obama vs. 49.775% chance Romney inclines me to go ahead and give him #50. I think RickJay has the right of it.
To be fair, I think he knew that one was going to be off and I’m sure he thought it was going to be short. His model wasn’t predicting what was most likely to happen with the EC. Just what the simulations told him would be an average result based upon the data inputted.
The thing that’s been proven is that Nate Silver wasn’t actually making any predictions. He was just reporting the facts (poll numbers) and applying them.
Silver deserves credit for being the first to do this in a public way but it’s not like he’s designed some unique formula. Now that he’s proven what should have been obvious all along - that objective facts are more accurate than some pundit’s gut feelings - you can bet that every major media source will have a 538 clone set up for the 2016 election.
Indeed he was. Props to him sticking to sticking to his model.
Silver’s most likely outcome was 332 -206. Looks like that is going to be dead on as well ;).
Technically, he was predicting an outcome, but it wasn’t that number. The ‘most likely’ EC outcome can be seen on the chart further down the main page, marked “Electoral College Distribution”. The highest peak looks to be near 330.
The number posted at the top (expected outcome) can be calculated from that chart. Take the probabilty (as a fraction) for any of the values on the chart and multiply it by that value. Do it for the whole chart, add them all up, and that’s your mean (aka expectation).
The last update showed the most likely elector count at a little more than 330 (the graph isn’t very big). There were just more possible outcomes below it than above it, so the average was lower.
I’m not so sure. If Nate assigned 55% odds for 10 states to Obama and Obama won all 10 states, then his model is miscalibrated. I’d like to see how his percentages score against other modeling efforts.
Hijack! What happens when that statheads say that a Presidential candidate has a 4% chance of victory in, say, July of an election year? A roaring economy plus incumbancy might do that. What’s the best strategy for the underdog? I asked a similar question a couple of months back, but I gave >10% odds. The veil has been lowered; things will never be the same.
Curse you Nate Silver!!! 
Yes, guys, I know. RNATB appeared to be referring to the prominently displayed average, so that’s what I was addressing.
No. What Balance said. And if you look at Nate’s chart of EV probabilities, the one he had with the highest probability of occuring? ** 332**.
ETA: What everyone else said too.
I was agreeing with you, not correcting you :).
Well, that’s all right, then. 
Those 10 state outcomes weren’t independent, though. Whichever way they diverged from the polls, they were likely to all diverge in the same direction. Get one right, and you’re likely to get them all. Senate races are closer to being independent, which is why looking at his record there is more illuminating than counting states called correctly in the presidential race.