538 Model Megathread

[KingSweden]

LOL, all the double digit polls changed nothing but a single IBT poll dropped Clinton by 1%.

Law of diminishing returns, I guess...

[Erich Maria Remarque]

So all of the landslide national polls this morning actually caused Clinton's chance of winning to tick DOWN slightly

Especially weird because Bloomberg polls has caused huge shifts in the past (because Nate absurdly overrates Selzer outside of Iowa).

538 is RIGGED!!!

[‼realJohnEwards‼]

I can't wait to see how the model reacts to these Emerson polls... so excited!

[dspNY]

I can't wait to see how the model reacts to these Emerson polls... so excited!

With the inclusion of the Utah poll, Evan McMullin's chances of winning UT have gone up to 24.7% in the Nowcast. However he is still only ay 10.9% in the polls-only

[‼realJohnEwards‼]

I can't wait to see how the model reacts to these Emerson polls... so excited!

With the inclusion of the Utah poll, Evan McMullin's chances of winning UT have gone up to 24.7% in the Nowcast. However he is still only ay 10.9% in the polls-only

Aaaaah! We need more Utah polling! Pollsters must not rest until McMullin is leading in now-cast, polls-only, and polls-plus! 10 EV freiwal!

[Sorenroy]

Utah has been added to the list of "States to watch".

[100% pro-life no matter what]

There is a cool little chart in an article of what a hypothetical Trump victory map would look like of the chances of Trump winning certain states, given that he won another.

For example, using Ohio, Pennsylvania, and Michigan:

If Trump wins Ohio:
-A 56% chance to win Pennsylvania
-A 48% chance to win Michigan

If Trump wins Michigan:
-A >99% chance to win Ohio
-A 61% chance to win Pennsylvania

If Trump wins Pennsylvania:
-A 97% chance to win Ohio
-A 51% chance to win Michigan

Another one: If Trump wins Minnesota or Wisconsin, there is a 72% chance that he is also carrying the other one.

Some aren't that coorelated.  For example, winning Nevada does nothing for him winning Virginia (only a 40% chance), apart from doing better nationally.

http://fivethirtyeight.com/features/if-trump-wins-heres-how-the-map-might-look/

[/]

FiveThirtyEight's presidential swing map as of 10/19:

[‼realJohnEwards‼]

Every datum's sacred, every datum's great

When a datum's wasted, Nate gets quite irate...

[Angel of Death]

Is it possible for Utah to turn purple in any of the forecasts in the remaining time or is the inertia that would need to be overcome just too great?

[Erich Maria Remarque]

Is it possible for Utah to turn purple in any of the forecasts in the remaining time or is the inertia that would need to be overcome just too great?

Ehm, he had lead in just one poll. But 3-4 polls showing him leading and he might take a lead in Now-cast.

[Speed of Sound]

Is it possible for Utah to turn purple in any of the forecasts in the remaining time or is the inertia that would need to be overcome just too great?

Ehm, he had lead in just one poll. But 3-4 polls showing him leading and he might take a lead in Now-cast.

Yeah, there's no reason he shouldn't be able to take over in Nowcast. As you say, though, he certainly shouldn't be yet.

[Beefalow and the Consumer]

IA is now exactly tied in the now-cast. 

[Interlocutor is just not there yet]

ME-2, IA, and AZ are within 1% margin of victory in all models. NE-2 is closely behind

[Speed of Sound]

Wasn't really sure where I wanted to put this, but the Princeton Election Consortium now has their odds of a Clinton victory at 97% in their random drift projection and at 99% in their Bayesian projection, which is the first time I've seen 99% pop up in their numbers. They have the EVs at 336-202.

EDIT: In fact, while I'm here, here's everyone's projections as collected by NYT

[Mr. Morden]

Wasn't really sure where I wanted to put this, but the Princeton Election Consortium now has their odds of a Clinton victory at 97% in their random drift projection and at 99% in their Bayesian projection, which is the first time I've seen 99% pop up in their numbers. They have the EVs at 336-202.

EDIT: In fact, while I'm here, here's everyone's projections as collected by NYT

Am I correct in my recollection that one thing that sets the 538 model apart is the comparatively high degree of correlation between the states in their model?  That is of course going to make a huge difference.

[Antonio the Sixth]

Wasn't really sure where I wanted to put this, but the Princeton Election Consortium now has their odds of a Clinton victory at 97% in their random drift projection and at 99% in their Bayesian projection, which is the first time I've seen 99% pop up in their numbers. They have the EVs at 336-202.

EDIT: In fact, while I'm here, here's everyone's projections as collected by NYT

Am I correct in my recollection that one thing that sets the 538 model apart is the comparatively high degree of correlation between the states in their model?  That is of course going to make a huge difference.

Yeah, and that's probably why their model is not as bullish on Hillary as almost all the others. They take into account the possibility of a relatively uniform swing across the board.

[Blackacre]

Trump's been gaining ground every day since the debate. Am I right to be a little nervous?

[Speed of Sound]

Trump's been gaining ground every day since the debate. Am I right to be a little nervous?

no

[Dr. Arch]

Trump's been gaining ground every day since the debate. Am I right to be a little nervous?

No, because he hasn't been. Open the model and look at the trend lines before saying such a thing, please.

[Gass3268]

Trump's been gaining ground every day since the debate. Am I right to be a little nervous?

If you look at the trend line, since the first debate you can see when there was lots of polling, Hillary's chances go up bigly and when there is a lull in polling her chances start to decline slowly.

[calvinhobbesliker]

lol, the Oklahoma poll hurt Hillary more than the ABC poll helped.

[Blackacre]

Trump's been gaining ground every day since the debate. Am I right to be a little nervous?

No, because he hasn't been. Open the model and look at the trend lines before saying such a thing, please.

I meant the third debate. I looked at the trend line.

Trump's been gaining ground every day since the debate. Am I right to be a little nervous?

If you look at the trend line, since the first debate you can see when there was lots of polling, Hillary's chances go up bigly and when there is a lull in polling her chances start to decline slowly.

So it's a lack of new information that's causing her numbers to go down? Okay, so its just noise. Not sure why it does that, though

[Lief 🗽]

Rasmussen and the IBD poll are weighing her numbers down, and there aren't enough high quality polls in the past week to drown them out.

[Ebsy]

Trump's been gaining ground every day since the debate. Am I right to be a little nervous?

The junk trackers have too much influence.