Are you Smarter than Nate Silver-- US Election Prediction Thread

[Lanny_McDonald]

[TheScorpion]

I have McMullin winning Utah, for what it's worth.

[flylock shox]

I reckon Clinton wins with 317.

[Acey]

Originally Posted by TheScorpion

I have McMullin winning Utah, for what it's worth.

He's 10-15 back of Trump. Any insight as to why you're predicting this?

[octothorp]

I've got it as a relative landslide, Clinton wins E.C. 334 - 204, Dems get control of the senate, Clinton finishes within rounding-up distance of a plurality.

Swing States
------------------
Florida - D
Ohio - R
Penn - D
NC - D
Mich - D
Wisconsin - D
Virginia - D
Colorado - D
Nevada - D
Minn - D
NH - D
Iowa - R
Maine - D
Maine Districts - D
Nebraska 2nd - R
ARIZONA - D

Senate Seats
-----------------
North Carolina - D
Missouri - D
NH - D
Indiana - D
Nevada - D
Penn - D
Florida - R
Wisconsin - D
Arizona - R

Popular Vote
-----------------
Clinton - 50 %
Trump - 44 %
Other - 6 %

[Iowa_Flames_Fan]

Here are my predictions at the top of the ticket:



For the Senate, my predictions are:

North Carolina: R
Missouri: R (though this one will be really close)
New Hampshire: D
Indiana: R
Nevada: D
Pennsylvania: D
Florida: R
Wisconsin: D (this one will be close too I suspect, though my guess is Feingold gets there)
Arizona: R

Popular Vote:
Clinton: 51%
Trump: 46%
Other: 3%

Since most people who pay attention to polls are predicting a Clinton win, I'll add one more prediction to spice things up: my guess is that the election will be called for Clinton at 9:25 Calgary time tomorrow night.

[bob-loblaw]

What's the number of seats needed in order to be called a 'rout'?

[ben voyonsdonc]

I'd say 320 + would be a rout.

[GGG]

Originally Posted by bob-loblaw

What's the number of seats needed in order to be called a 'rout'?

Winning any one of Arizona, Alaska, or Georgia would be my definition of a rout or about 8-10 points nationally, and taking the senate.

Winning all the swing states listed above is a solid victory but not quite a good enough one to destroy the Republican Party. If you flip one of "their" states it will severely limit the power the white nationalists have going forward

[Senator Clay Davis]

So I'm going with a bit of a twist to my map: Trump wins Michigan, but loses Arizona. Hispanic factor will also be too much to overcome in Florida. I also think Alaska will be very tight and give Hillary a 45% chance of flipping it.

[troutman]

What’s Wrong With 538?

http://www.huffingtonpost.com/entry/...b0334571e09e74

With the same information, 538 is currently predicting a 65 percent chance of a Clinton victory, while HuffPost’s Natalie Jackson and Adam Hooper are projecting a 98 percent chance,[1] and Sam Wang at Princeton Electoral Consortium is predicting a >99 percent chance.[2] What gives?

Underlying this is an overly complex and opaque set of assumptions that are probably too smart for their own good.

The mantra for model building has always been “garbage in, garbage out.” You can have the most intricate set of assumptions, but if your model is spitting out unrealistic results, there’s something wrong with your model.

538’s probability of a Clinton or Trump victory relies on the results of at least 20,000 model runs each time a new poll (or set of polls) comes in.[5] To get their probabilities of victory for each candidate, they sum up the number of times that candidate got 270 or more electoral votes on each run, much like Jackson and Hooper at HuffPost or Sam Wang at PEC.

It’s great to build a complex model and load it with empirical data like polls, economic reports and presidential approval ratings. But if the output of that model is implausible, it’s time to go back to the drawing board.

[CorsiHockeyLeague]

^That article is fallacious. I don't have any difficulty believing that there is indeed a 3 percent chance of a historic polling error in Clinton's favour, for example. All Silver's model is doing is not outright dismissing the possibility that the data is highly misleading, while at the same time assigning a very low probability to that scenario.

[GirlySports]

huffington post is the absolute worst, nobody should read that

[mrkajz44]

Originally Posted by troutman

It’s great to build a complex model and load it with empirical data like polls, economic reports and presidential approval ratings. But if the output of that model is implausible, it’s time to go back to the drawing board.

How exactly do you know the output of the model is implausible though? Even when we know the result of the election, that doesn't mean any one model was right or wrong.

Honestly, I think this election is going to be much closer than a lot of people are giving it credit. I still think Hillary wins, but not in a landslide fashion. The problem is Hillary is going to win and then all the groups that gave her a high chance of winning will point to their model as the best, even though they were being way overconfident.

[automaton 3]

I have it as 275-263 for Clinton. I think it is going to be pretty close.

I had NH, FL and NC as going Trump.

[Parallex]

Originally Posted by GGG

Winning all the swing states listed above is a solid victory but not quite a good enough one to destroy the Republican Party.

If your definition of a rout is one that fundamentally alters a political party then you'd need more then just flipping a red state. Bill Clinton took multiple red states in his two campaigns, Won the electoral college by more then 200 EV, didn't destroy the Republican Party.

Reason it didn't destroy the Republican Party was because they were voting for Bill Clinton... not the Democratic Party. If H. Clinton is re-elected under any margin that might do it (16 years is a long time to be locked out of the executive). Texas becoming a true swing state would likewise do it (removing 38EV from their anchor states would make it almost impossible to win without major internal changes).

I don't think it's possible to truly destroy the Republican Party... but it is possible to inflict so many losses that they stare into the abyss and are forced to realign themselves.

[JBR]

I have 275-262 Trump. Against logic, there is a path to 270 for him.

Looking forward to watching the series finally of "Murica" on CNN tonight.

[GGG]

Originally Posted by Parallex

If your definition of a rout is one that fundamentally alters a political party then you'd need more then just flipping a red state. Bill Clinton took multiple red states in his two campaigns, Won the electoral college by more then 200 EV, didn't destroy the Republican Party.

Reason it didn't destroy the Republican Party was because they were voting for Bill Clinton... not the Democratic Party. If H. Clinton is re-elected under any margin that might do it (16 years is a long time to be locked out of the executive). Texas becoming a true swing state would likewise do it (removing 38EV from their anchor states would make it almost impossible to win without major internal changes).

I don't think it's possible to truly destroy the Republican Party... but it is possible to inflict so many losses that they stare into the abyss and are forced to realign themselves.

Bill Clinton's win was expected though. 3rd term of republican rule in a recession is an unlikely election to win.

You had the moral majority coming into power in the republican party and after the loss took control of it as the largest coalition. You also didn't have a clear schism between the establishment republicans and the base that you do now. So I think that election was much more they were due to win one. Now they are facing having lost 3 elections decisively with a variety of candidates and demographics getting worse for them each year.

And when I say destroy the republican party I mean a democratic style reorganization and modernizing of social and immigration issues. Or they could go further into white nationalism with anti immigrant anti trade rhetoric but tone down the overt racisim and try to go after the Midwest.

[OBCT]

Clinton 337
Trump 201

Popular Vote:
Clinton: 53%
Trump: 42%
Other: 5%

[GGG]

So looking through no one had Pensylvainia going trump. A few had Michigan or Wisconsin but not both. Of the famous models Silvers was the closest, exit polls, Huff Post, and PEC took a beating.

It's interesting that things like swing counties and who gets the most votes in primaries which have to be correlations not causation are still alive and well.

Dilbert was wrong as his last post called for a Clinton rigged victory. I assume he is scrubbing that post from existence and claiming he was right all along.

I think a bunch of stuff from Silvers model was bang on.

- if trump he would lose pop vote
- swing states are more sensitive to changes in national popular vote.
- a two - three point polling error is very normal in election campaigns
- given a high number of undecideds and third party their was high uncertainty in the race.
- that if a model predicts something is 30% possible that is pretty good odds of hit happening.

I think future models will have more error added to them as sampling become more challenging

One of the most interesting use of stats will be to check if voter suppression efforts in NC and elsewhere were effective in surpressing AA turnout or if was lack of enthusiasm relative to Obama