Stand up is hard. Too many, males and females, want to be the next Kinison, makes it hard to find anybody decent.

Your second paragraph is actually somewhat assuring, knowing that his model essentially puts a ceiling on the chances of the candidates because of the uncertainty and volitility factor. I do wish a time lord would visit by my spot today just to let me know to R-E-L-A-X. As a Chicagoan, this might have been the worst headline I've ever read. It made my soul cry.

http://fivethirtyeight.com/features/...an-trump-does/

this election is ridiculous

yeah that's the abc/wash post tracking i mentioned above. it concerns me. wasn't it like 8 a few days ago. but it's a tracking poll. like the monkey poll is too, but that one is called monkey so you know it has to be good since they are so confident they call it that.

i bet its just a short time now before comey resigns. both reps and dems are after him now for good reason and he was already disliked by many. so i think hes gonna be too controversial at this point to continue.

NBC/SM poll of 40k voters, half of which is post-Comey, has Clinton +7. The poll is released every Tuesday. Prior Tuesdays: C+6, C+8, C+7, C+6, C+5, C+7, and C+5.

I'm not going to pretend to be a stats expert, but if Trump's chances of winning 2 states are in fact higher than 1/10x1/10 aren't Hillary's chances of winning those same states also then higher than 9/10x9/10 making it a wash?

This. My model does this, PEC does this, I'm sure Upshot does this although minimal research on my part hasn't revealed a ton about their methodology. No model that treats state outcomes as independent should be taken seriously.

It's not a wash because Trump winning them is more likely to change the outcome of the election.

If the state results are strongly correlated then the expected # of electoral votes won stays the same compared to the independent case, but the variance of electoral vote results is higher. Higher variance increases the chances for the underdog.

Now how the hell did Hillary get emails onto this server??? LOCK HER UP!


Re: Nate's model - it only includes polls, not early voting results, which would appear to strengthen NV and NC somewhat if not considerably. Goofball you still feel confident with Trump <1%?

i feel like Clinton is taking the highroad a bit too much. Like if the independants can be swayed by just mentionning unread emails she could just come out and talk about the FBI having russian proof on trump and ****.

Even if its not true it will be too late and votes will be cashed in

Yes and no.

For example.

2 states, Trump has a 10% chance of winning State A & 10% chance of winning State B.

What are the chances Trump wins both states?

Correlation = 0 - 1%
Correlation = 1 - 10%

What are the chances Clinton wins both states?

Correlation = 0 - 81%
Correlation = 1 - 90%

Clinton is more likely to sweep if both states are correlated perfectly but so is Trump. The questions is who *needs* to sweep. If you assume Trump needs to sweep, he's helped by positive correlation, same for Clinton. The candidate who is trailing is more likely to need a sweep.

It could be that I don't math well, or I don't follow your logic.

Hypothetical, if Clinton is 90% to win the presidency, that means Trump is 10% to win.

He basically wrote, 'if trump wins one state, he's more likely to win the neighbouring state' so his chances of winning the presidency go from 10% to 20%.

Why doesn't that math then work in the reverse.

If Clinton wins a state, she's also more likely to win the neighbouring state, making her chances at a crushing defeat all that much higher.

Nate's model indirectly includes early voting results in as much as they show up in the polls. i.e some polls ask people if they've already voted, and weight those people higher than registered or likely voters.

The math works in reverse exactly the same way. It's just less material to the outcome (i.e. winner) of the election for exactly the reason you write. If you assume that the tails in the election are, on the left a Clinton landslide and on the right a small Trump win, interstate correlation makes both the tails more likely, which means it makes Trump more likely to win AND to get whomped. But 270 votes is all you need, so the variance doesn't help Clinton in the same way it helps trump.

Ok, gotcha.

Correct, but it doesn't increase her chances of winning, just of how she wins. Nate has mentioned that the chances of a Hil blowout are about equal to the chances of a small Trump win. As he sees it right now, there's a 25% chance of Trump winning, with pretty much all of that being a close win and virtually 0% of that being a Trump blowout. Then of the 75% chance of Hil winning, probably 50% is close/substantial win, and 25% blowout.

Also, to your point about it working the other way, yes, you're correct, but with her chances for each state already being at 90%, her scooping both states is much closer to what the odds predict so it really wouldn't be an outlier event. I guess what I'm saying is that him winning both states at 10% each shouldn't be looked at the 1/100 chance, but actually closer to just the 10% chance of winning one state. And because of that, it gives him mroe win equity for the whole election.

You said it better. TY.

Seriously, though, you still confident about the 99%??

FBI is completely out of control at this point. Not sure how to go about it, but any agent caught leaking this stuff should be ****canned immediately. Of course going HAM vs anyone caught talking about investigations close to the election would be easier if the ****ing head of the FBI hadn't done the same thing. What a cluster****.

There's a poll coming in iirc 8 days. It'll be fairly accurate. Methodology will still be questioned, that's for sure.

McMullin for President: He probably won't be in jail for inaguration!

Sorry, but if my boss is trying to help a fascist overthrow democracy in America, I'm going to the press with any and everything I know.

#SorryNotSorry

It will be nice when that one comes in.

Heard it's going to be a large sample size.

The next director should make agents watch ethical training videos A Clockwork Orange style for 5 hours a day, the next 3 years. A lot of crime will go uninvestigated, but, you gotta do what you gotta do.