Nate Silver Bets the Farm: Obama 92.2% Odds of Winning

[DMX7]

http://fivethirtyeight.blogs.nytimes.com/

The forecast is now the now-cast.

[RandomGuy]

90.9% now.


It's all over except the whining about how unfair the media is/was by conservatives.

Because the only real explanation is that the "media" was biased, and it couldn't possibly be any fault of the conservative ideology or the fact it produces shitty candidates for national office. That is just crazy talk.

[Clipper Nation]

Don't confuse neocon ideology with conservative ideology, tbh.... it's the neocons who suck ass.... if the GOP allowed Ron Paul to be nominated, Obama would be packing his bags right now....

[101A]

Nate Silver is the guru? I took the conventional wisdom of the board and believed it.

I just Googled the guy.

His whole deal started on the Daily Kos?

His claim to fame is his correctly predicting "49 of 50" states in 2008?

Weren't 47 of them pretty much understood then?

I believe Obama will win today; mainly because of Ohio, but not because of anything this guy is writing. Those of you Blue Teamers thinking he is some kind of Svengali have really drunk the Kool-Aid.

[Clipper Nation]

He also correctly predicted everything in 2010.... he's a statistician first and a writer second, and his methods are pretty accurate.....

[101A]

He also correctly predicted everything in 2010.... he's a statistician first and a writer second, and his methods are pretty accurate.....

Yeah, I just read the rest of his Wiki; not perfect in '10, but pretty good (and sucks for Romney supporters - his tendency is apparently to over-predict for the Republican)

The hope Red teamers can hold on to is that the ntl polls he is aggregating are systematically making the same error; enthusiasm gap, etc....

[boutons_deux]

The media is gonna look really (MORE) stupid for "it's too close to call." "It's a dead heat" which we heard going into 2008 election.

[timvp]

Tbh, I'm not impressed by his final numbers. He's basically saying there's a 9.1% polls are worthless. That's a damn high percentage. I wonder if his model truly has reason to believe that the polls mean nothing or he's just giving himself an out if it turns out everything he does was actually a GIGO exercise.

Great blog to read but this final percentage makes me trust his entire methodology less :td

[George Gervin's Afro]

Tbh, I'm not impressed by his final numbers. He's basically saying there's a 9.1% polls are worthless. That's a damn high percentage. I wonder if his model truly has reason to believe that the polls mean nothing or he's just giving himself an out if it turns out everything he does was actually a GIGO exercise.

Great blog to read but this final percentage makes me trust his entire methodology less :td

timvp sighting...! It must be a huge day in politics for him to show up..

[boutons_deux]

"He's basically saying there's a 9.1% polls are worthless."

No, he's saying Bishop Gecko has a 9.1% chance of winning.

If he said 55% Barry / 45% BG, would you interpret that as him saying 45% of the polls are worthless?

[timvp]

timvp sighting...! It must be a huge day in politics for him to show up..

While I enjoy reading the political forum -- it's usually at the very least entertaining -- I'm too agnostic and cynical politically these days to actually add to any discussion in a constructive matter. I'd rather read than interrupt with my cynicism.

[timvp]

No, he's saying Bishop Gecko has a 9.1% chance of winning.

The numbers, the trends, the historical precedence and everything else his model uses points to a clear cut Obama victory. Maybe not a landslide victory but a relatively easy victory nonetheless. The only explanation of a Romney victory is if his entire model turns out to be worthless. A 9.1% chance that his model means nothing is pretty damn high.

[MannyIsGod]

Tbh, I'm not impressed by his final numbers. He's basically saying there's a 9.1% polls are worthless. That's a damn high percentage. I wonder if his model truly has reason to believe that the polls mean nothing or he's just giving himself an out if it turns out everything he does was actually a GIGO exercise.

Great blog to read but this final percentage makes me trust his entire methodology less :td

Nah you're interpreting the percentage incorrectly. He definitely thinks that a lot of the polls have very little weight but that has nothing to do with the final percentage.

Basically its like this. Imagine you're in a car speeding down an interstate toward a destination. When you leave your point of origin you have so many option on the path to take and where you'll end up. As you move further and further down that highway you loser the possibility of the locations you can end up and your possible outcomes narrow down.

As we've moved to election day the possible outcomes of this race have dramatically reduced. He's basically saying that at this point with polls the way they have been and the recent movement there are just very few scenarios left where Romney wins. Thats why you saw such large gains in one day for Obama relative to a single day a few weeks ago.

He specifically tweeted out today that the margin of the model does not reflect that they do believe it to be a close outcome. Its not going to be a blowout. But they simply think Obama wins that close outcome 9 out of 10 times.


Also, FWIW he does give Romney a ~10% chance of winning. A Romney win doesn't automatically mean the model is wrong as the model does have Romney winning in certain scenarios.

[FuzzyLumpkins]

Tbh, I'm not impressed by his final numbers. He's basically saying there's a 9.1% polls are worthless. That's a damn high percentage. I wonder if his model truly has reason to believe that the polls mean nothing or he's just giving himself an out if it turns out everything he does was actually a GIGO exercise.

Great blog to read but this final percentage makes me trust his entire methodology less :td

He considers those polls but it's an aggregate number and compounding probabilities based on electoral votes. A whole bunch of the polls indicate that Obama is up in the safe starting points like NY, California and the usual democratic suspects. That starts them out at around 190 to 130 or somesuch. Then you have the other states and their aggregate numbers for polls.

They have their margins of error and using those margins he determines the likelihood that for example Romney will win Florida. So these are made up numbers but lets say his aggregate is Romney with a 2 point lead in Ohio and the error is +/- 5%. Over that 10% range there is 70% ie >-2% chance that Romney wins. Lets say he has the same for Obama winning Iowa up by 4% with the same margin of error then that gives Romney only a 10% chance to win. Assuming their equally rated, they are not, then Romeny only has a 7% chance to take both, a 67% chance to take FL and not IA etc. If Romney needs both because he is behind in other states then he has only a 7% chance to win the election.

You then factor in how many electoral votes each guy needs and total up ALL these various possible outcomes and he comes up with the final numbers. And the reason why he has Romney up by so much is because Romney is polling poorly in the swing states:

http://graphics8.nytimes.com/images/2012/11/04/us/politics/fivethirtyeight-1003_1/fivethirtyeight-1003_1-blog480.png

Combine the two notions and Romney's outlook does not look good.

[LnGrrrR]

Nate Silver is the guru? I took the conventional wisdom of the board and believed it.

I just Googled the guy.

His whole deal started on the Daily Kos?

His claim to fame is his correctly predicting "49 of 50" states in 2008?

Weren't 47 of them pretty much understood then?

I believe Obama will win today; mainly because of Ohio, but not because of anything this guy is writing. Those of you Blue Teamers thinking he is some kind of Svengali have really drunk the Kool-Aid.

He was a baseball statician before he was a political statician. He did pretty well in both jobs.

[boutons_deux]

"he is some kind of Svengali have really drunk the Kool-Aid."

We'll see by Wed morning if he's a one-trick election pony, or whether repeats today, then all the "nate silver truthers" can go suck the Kock Bros.

[MannyIsGod]

Yeah, I just read the rest of his Wiki; not perfect in '10, but pretty good (and sucks for Romney supporters - his tendency is apparently to over-predict for the Republican)

The hope Red teamers can hold on to is that the ntl polls he is aggregating are systematically making the same error; enthusiasm gap, etc....

Poll tendencies, not his actually. I talked a lot about how the polls undersample democrat demographics in 2008 and its still continued. I heard the same thing then about how the polls were D+6 an obviously skewed. Same shit every time I guess.

[timvp]

Nah you're interpreting the percentage incorrectly. He definitely thinks that a lot of the polls have very little weight but that has nothing to do with the final percentage.

Basically its like this...Yeah, I understand how time was factored into the model. Makes perfect sense and is standard for predictive modelling.


A Romney win doesn't automatically mean the model is wrong as the model does have Romney winning in certain scenarios.

My issue is that his "oops the polls must mean less than I thought they meant" percentage should be lower than 10%. On paper, I just don't see how his model isn't >95% Obama. Everything he accounts for is clearly in Obama's favor. The only explanation is the model factors in a significant GIGO padding, which in itself is a little disappointing.





(Then again, he's a statistician with a lucrative career in front of him. Silver probably realizes that it's much safer for his livelihood to explain away being wrong about 10% than being wrong about 1-2%. If he had this thing at 98-99% and Obama loses, his career would effectively be over. He can survive being wrong about 10%.)

[scott]

Tbh, I'm not impressed by his final numbers. He's basically saying there's a 9.1% polls are worthless. That's a damn high percentage. I wonder if his model truly has reason to believe that the polls mean nothing or he's just giving himself an out if it turns out everything he does was actually a GIGO exercise.

Great blog to read but this final percentage makes me trust his entire methodology less :td

Your post is basically saying there is a 100% chance you don't understand how statistics works.

[timvp]

Your post is basically saying there is a 100% chance you don't understand how statistics works.

Get it, because he used a percentage it's funny.

[ChumpDumper]

90.9% now.


It's all over except the whining about how unfair the media is/was by conservatives.

Because the only real explanation is that the "media" was biased, and it couldn't possibly be any fault of the conservative ideology or the fact it produces shitty candidates for national office. That is just crazy talk.You forgot....

MASSIVE VOTE FRAUD

[timvp]

To give me reason to follow along today, I propose MannyIsGod the following bet:

Romney wins, I PayPal you $100

Obama wins, you PayPal boutons_deux $10


This should be a no-brainer for you since you apparently believe Silver's 90% number :)

[timvp]

Alright, MannyIsGod, your silence is a good negotiating tactic. $125 to you if Romney wins, you pay boutons_deux $10 if Obama wins plus $1 to DarrinS.

Deal?

[LnGrrrR]

Shoot, I'll take that deal if Manny won't. :lol I"m pretty sure the amounts are supposed to be reversed though... (ie. 10 to Manny, 100 to Timvp)

[RandomGuy]

While I enjoy reading the political forum -- it's usually at the very least entertaining -- I'm too agnostic and cynical politically these days to actually add to any discussion in a constructive matter. I'd rather read than interrupt with my cynicism.

We could always use a new flavor of cynicism... :D

The old ones get, well, old.

[RandomGuy]

You forgot....

MASSIVE VOTE FRAUD

DUN DUN DUN!!!!

[DarrinS]

Alright, MannyIsGod, your silence is a good negotiating tactic. $125 to you if Romney wins, you pay boutons_deux $10 if Obama wins plus $1 to DarrinS.

Deal?


One dollar? I guess I could buy a quart of gasoline.

[George Gervin's Afro]

You forgot....

MASSIVE VOTE FRAUD

http://www.foxnews.com/politics/2012/11/06/new-black-panthers-back-at-philly-voting-site/

Black Panther sighting! Run whitey run!

lol

[DarrinS]

By the way, just voted for a friend. All they do is scan the bar code on the registration card. Easy peasy.

[Clipper Nation]

By the way, just voted for a friend. All they do is scan the bar code on the registration card. Easy peasy.
But what about that :cry MASSIVE LIB'RUL VOTE FRAUD :cry the neocons whine about? Just goes to show neocons are the real cheats...

[ChumpDumper]

By the way, just voted for a friend. All they do is scan the bar code on the registration card. Easy peasy.No, you didn't. You got an email saying someone else did. Easy peasy.

[MannyIsGod]

:lmao I'll take that bet

[timvp]

:lmao I'll take that bet

Nice. :lol

[DarrinS]

But what about that :cry MASSIVE LIB'RUL VOTE FRAUD :cry the neocons whine about? Just goes to show neocons are the real cheats...

I didn't, but there's absolutely NOTHING that would prevent me from doing it. They only make sure the bar code hasn't been used already. They were pretty adamant about me NOT showing ID. Weird.

[MannyIsGod]

:lmao

Best bet ever. Good job LJ.

[ChumpDumper]

I didn't:lol

[Clipper Nation]

By the way, just voted for a friend. All they do is scan the bar code on the registration card. Easy peasy.


I didn't, but there's absolutely NOTHING that would prevent me from doing it. They only make sure the bar code hasn't been used already. They were pretty adamant about me NOT showing ID. Weird.

:lmao

Son, no wonder you're wasting your vote on Willard, you flip-flop just like he does! :lol

[DarrinS]

:lmao

Son, no wonder you're wasting your vote on Willard, you flip-flop just like he does! :lol


I'm not a felon.

[ChumpDumper]

I'm not a felon.But you could be.

[MannyIsGod]

:lmao

[timvp]

I'm too agnostic and cynical politically

In reply to the PM I got, this doesn't mean I don't believe in god, tbh . . .

[Th'Pusher]

By the way, just voted for a friend. All they do is scan the bar code on the registration card. Easy peasy.
We're you wearing the Groucho Marx fake nose/mustache/glasses?

[Bartleby]

He was a baseball statician before he was a political statician. He did pretty well in both jobs.

And a professional poker player, IIRC.

[Drachen]

But you could be.

Easy peasy

[scott]

http://isnatesilverawitch.com/

[Drachen]

http://isnatesilverawitch.com/

You know who would say that they aren't a witch????

A FUCKING WITCH!

[scott]

You know who would say that they aren't a witch????

A FUCKING WITCH!

Valid point, tbh

[Drachen]

Good, so we are all agreed here through rational thought and deductive reasoning that Nate Silver is a witch.

I will start working on the website now.

The URL is

www.natesilverisawitch.com

[timvp]

The numbers, the trends, the historical precedence and everything else his model uses points to a clear cut Obama victory. Maybe not a landslide victory but a relatively easy victory nonetheless. The only explanation of a Romney victory is if his entire model turns out to be worthless. A 9.1% chance that his model means nothing is pretty damn high.

:lol @ 10% chance Romney wins.


Anyways, boutons get yo money.

[boutons_deux]

:lol @ 10% chance Romney wins.


Anyways, boutons get yo money.

send it to RedCRoss for Sandy victims. Boutons bless

[AntiChrist]

I like this Nate Silver fellow.

[MannyIsGod]

send it to RedCRoss for Sandy victims. Boutons bless

http://i.imgur.com/9uTnR.jpg

Done!