538 prediction vs actual

[libertango]

538 prediction vs actual
Originally uploaded by halobrien

The map comparing Nate Silver's predicted Electoral College map (bottom), vs the actual results (top).

http://www.FiveThirtyEight.com/ has clearly been a big winner this cycle. Here's his only variances:

* Indiana went Democratic
* Missouri was a "toss up" -- it looks like it'll go Republican
* Nebraska's Electoral College votes are split by congressional district. They're still counting, and the possibility remains that the district with Omaha will go for Obama.

Given Nate was predicting 55 individual races -- 50 states, 3 districts in Nebraska, 2 in Maine -- that's a pretty good batting average.

Comments

[pecunium.livejournal.com]

I'd say, given the count in Missouri, I'll give him that one.

It's a rout. Better than I hoped for.

[ckd]

It's sort of 57 races, since NE and ME give the 2 "Senator" EVs to the statewide winner. (The odds of there being some sort of weird blowout in one district while the other one or two are narrow wins is pretty close to zero, though, so simply treating the state total for NE as the "whoever gets two" and only worrying about ME if it looks like it'll split means 55 is a more reasonable number.)

Right

[hal-obrien.livejournal.com]

"It's sort of 57 races, since NE and ME give the 2 "Senator" EVs to the statewide winner."

Which is how each of the 48 other states allocate all of their EVs.

So there are 50 statewide races, and 5 individual congressional district races.

It's not like Maine and Nebraska run two statewide races apiece.

Re: Right

[ckd]

Yeah, I should have gone to bed earlier instead of trying to catch up on LJ and my RSS feeds; even with a certain amount of "mark all in feed as read" I was still up way too late.

OTOH, we both forgot Poland DC.

[daveon.livejournal.com]

He was especially good at the actual percentages in the popular vote.