Sunday, September 25, 2016

Clinton vs Trump.
25 September 2016,
44 days from the election

Another good week for Trump, but important states are moving both towards his column and away from them. More on that after the numbers. RED indicates a state moving from one category to another in Trump's direction, while BLUE means the state moved in Clinton's direction.

Solid Trump (more than 95% Confidence of Victory [CoV]): ND WY AL ID TN UT MT SD KY LA WV OK MS MO AK SC ME-2 KS AR IN TX  
Total: 160

Leaning Trump (between 50% and 95% CoV for Trump): IA NE GA OH NV AZ NC

Total: 77

Toss-up: FL
Total: 29
 
Leaning Clinton (between 50% and 95% for Clinton): CO MI PA WI ME  RI NH NJ

Total: 79

Solid Clinton (more than 95% CoV for Clinton): VA WA MN DE NM OR IL ME-1 CT DC NY VT MA HI CA MD
Total:193
 The current count 

CLINTON 272
TRUMP 237

TOSS-UP 29


The big changes this week were Nevada and North Carolina moving into the Leaning Trump camp and Florida leaving the Trump total and going back to toss-up.  With Clinton at 272, she cannot afford to lose a single state that now supports her, a daunting proposition. My model is a snapshot and it gives Clinton much better odds than many others do, but right now these are very small slivers of hope. I will explain below why those slivers are a little larger than you might think.

Here are the odds of victory if the election was held today, using the 15 battleground states according to my algorithm.

Trump's non-battleground count of electors: 122

The battleground states ranked by pqn
(Clinton%, Trump%, electors)

1. FL: 50%, 50%, 29
2. NC: 45%, 55%, 15
3. PA: 82%, 18%, 20
4
. OH: 18%, 82%, 18

5. MI: 79%, 21%, 16
6
. AZ: 37%, 63%, 11

7. CO: 71%, 29%, 9
8. GA: 10%, 90%, 16
9. NV: 34%, 67%, 6
10. TX: 3%, 97%, 38 
11. WI: 87%, 13%, 10
12. NJ: 94%, 6%, 14
13. NE: 10%, 90%, 5
14. IA: 7%, 93%, 6

15. RI 91%, 9%, 4


Current probability of victory if the election were held today:

Clinton 84.8%
Trump 15.2%
 

It should be expected that Clinton is still favored, but it might seem counter-intuitive that my model still has her as a solid favorite. It's all about playing defense. Yes, Trump only has to steal a state and has several to choose from. Colorado is easiest right now, but Michigan, Pennsylvania and Wisconsin are in the realm of possibility. His odds are made longer because he also has to play defense, most especially in North Carolina and Florida, and to a lesser extent in Nevada. At the moment, Ohio doesn't look much like a battleground, Trump having the lead in eight of ten polls conducted in September. That said, Clinton has those kinds of advantages in the four states where she must play defense.

State: # of polls with Clinton leading in September of total # of September polls
CO: 6 of 8
MI: 5 of 5
PA: 6 of 7
WI: 5 of 5

There are two analogies I can point to right now, and they do not point in the same direction. Obama in 2012 had a bad first debate and his numbers fell for awhile, but after about a week and a half the numbers climbed back to their previous levels. The other is 2004, where it really came down to Ohio and Florida, and Bush only won re-election because he swept them. 2012 is an analogy in Clinton's favor if her numbers bounce back and it should be noted Obama's worst day had him still winning 278 to 260, then slowly climbed back over 300 electoral votes. The 2004 analogy should favor Trump, but Ohio and Florida in 2004 were both closely contested while none of the states he needs to steal currently are close.

Though my numbers are more confident than Nate Silver's, I'm still anxious seeing how close the Electoral College numbers are right now.  I'll be back next week after the first debate, which has every chance to be the strangest presidential debate in living memory.

It will also be October when next I write. There were times this year I thought it would never come.
  
  

4 comments:

  1. Amazing how close it now is considering her opponent is a game show host.

    I think OH is now a lost cause but NC and FL are very winnable for Clinton. Hard to figure out what's going in NV as the state is heavily latino. If she wins two of those three then Trump will be hitting a wall in the 206-229 vote range.

    Last week I was worried it would be heading to a 2004-like situation but seems more like 2012. Do you think FL, NC, and NV will be the prime states to watch?

    ReplyDelete
    Replies
    1. It is amazing (or terrifying) how close this is.

      I expect numbers to improve for Clinton, though I doubt they'll go back to the August numbers. Right now, it's all about Colorado. I'd prefer a race where she ends up with more than 300.

      Thanks for stopping by.

      Delete
  2. Do you have an opinion on the PollyVote Model?

    ReplyDelete
    Replies
    1. Hi, Peter, sorry to take so long to answer. There's a lot to like about it. I especially like this sentence on their front page.

      Keep it simple: There is no evidence that complexity improves accuracy.

      Our methods vary. I like to show the predicted EV count from week to week and I don't touch the popular vote count. Also, Pollyvote doesn't deal with the Senate, which is a vital component to this election that is largely ignored in even normal election cycles and is almost invisible with the news being about Trump 24/7.

      I'm a little surprised they have the popular vote count getting closer after the debate. The state level polls are consistently good for Clinton, making it look like her lead is widening.

      Delete