Sunday, July 31, 2016

Clinton vs. Trump.
31 July 2016,
100 days from the election

Hello, friends, family and strangers. Welcome to the first of the weekly reports on the 2016 presidential electoral college race between Hillary R. Clinton and Donald J. Trump. I include strangers because I am making an effort on social media - notably Twitter and Facebook - to get this blog out to as many people as possible because the most respected name in poll aggregation, Nate Silver, is producing something called a "now-cast" that says the race for the presidency is close.

My numbers say it isn't. My numbers do not claim to be predictor of the election, only a snapshot of what the polls say today. Even so, what is being reported by Nate Silver amounts to mathematical malpractice and people need to know not all aggregators agree with him.

If you are a stranger, your first question should be: Who the hell is this guy?

I'm the guy who beat Nate Silver in 2008 and beat him again in 2012

If I had some small disagreement with Silver's numbers, I might send him snide messages on Twitter, which he would likely ignore. The thing is, my numbers don't say a small difference. He said Trump was leading early this week and after the convention he now says it's 51%-49% for Clinton.

To repeat, I don't call my numbers a prediction, just a snapshot of what polls say now.

The snapshot says 99% to 1% for Clinton.

Let me quote Glengarry Glen Ross.

"You think I am fucking with you? I am not fucking with you."

We are 100 days away from the election and it feels like 100 years in purgatory. Having someone you trust telling you to worry makes it worse.

I'm here to say I'm as good as Nate Silver is (okay, just a little better) and right now he is peddling crap just as stinky as Ron Fournier, Howard Kurtz or Mark Halperin.

Let's go to the numbers, shall we?

The current count: 
Clinton 313
Trump 207
Toss-up 18 

In my system, only 50%-50% counts as a toss-up.

You might notice that Trump's red line is trending upward and Clinton's blue line is trending downward.

It is worrying, but not distressing.

Here is how we get to this state of affairs. These numbers will explain my lack of distress.

Solid Trump (more than 95% Confidence of Victory [CoV]): ND WY OK ID WV KY AL NE KS TN LA SD MO AR NH TX AK MT MS SC IN 
Total: 162
Leaning Trump (between 50% and 95% CoV for Trump): AZ GA IA NV UT
Total: 45
Toss-up (exactly 50% CoV): OH 
Total: 18
Leaning Clinton (between 50% and 95% for Clinton): FL OR WI
Total: 46
Solid Clinton (more than 95% CoV for Clinton): MA DC HI CA VT IL RI MD NY NJ WA DE CO NM PA CT VA NC MN ME MI
Total: 267

The number that explains my current relative lack of distress: 267 Solid Clinton electoral votes. Yes, that's 267 out of a necessary 270. If she holds onto the stuff where she looks very good right now and wins Oregon or Wisconsin, she's the next president. Obama won both those states comfortably, and there are many other possible paths with nice noticeable odds, paths with over a 1% chance of happening. Trump's best single path to victory is currently at 0.05%.

For people who like to gamble, that path is now at 1 chance in 2,000.


The numbers that still give me distress: I wish this was a blow-out that would repudiate Trumpism for several elections to come. I don't think Clinton can get to 400 electoral votes, but over 350 is in the realm of possibility and that would be good. Under 300 would be a bad thing.

A blowout may not be needed to show all the cracks in the GOP coalition. We could see some very messed up stuff for them even before November.

 
And now the numbers that might make anyone to the left of Mitch McConnell sleep better this week.

Current probability of victory if the election were held today:

Clinton 98.9%
Trump 1.1%

Feel any better, my left leaning friends? If so I'm glad. I live to serve.

To get an idea historically, Trump is in much worse shape right now than McCain or Romney were after the conventions were over. One advantage both the last two Republican nominees had was when it came to battleground states, they weren't playing defense. States didn't lean Republican, they were almost all Solid Republican, even the most contentious having a 95% Confidence of Victory for the GOP or better, most well over 99% CoV. They just had to steal several states from Obama, much easier said than done. Trump not only has to steal, he has to worry about states that were solidly Republican in the last few electoral contests. Playing defense in Utah and Georgia really surprises me, and I wouldn't be surprised to see them go back to Solidly Trump, but right now the deal has not been closed. 

Once again, this is a snapshot and not a prediction. If you look around and think Trump is running a fucked-up campaign, here's some confirmation. The polls are saying most Americans agree with you.

To reiterate, if I pretty much agreed with Nate Silver's numbers, I wouldn't be trying to get this out to a larger audience. But we don't agree and I'm better than he is, based on the evidence from 2008 and 2012.

I'm a nerd and I'm a geezer, so let me quote The Guns of Will Sonnett.

"No brag, just fact."

     
(Note: Thanks to Ellis Weiner for his unpaid copy editing.)
  

31 comments:

  1. I want every Democrat to be so scared of Trump that we take back the Senate and the House tho. :D

    ReplyDelete
    Replies
    1. I want to see Democratic loyalty to the other races on the ticket as well. A good turnout could take back the Senate. The House is sadly more problematic, but the GOP infighting could shatter their fragile coalition.

      Delete
    2. Thanks for this...I remember you in 12 and am very happy to see this. I question Nate's results too.

      Delete
    3. Thanks for this...I remember you in 12 and am very happy to see this. I question Nate's results too.

      Delete
    4. Hi, Terry. Nice to hear from you. It's going to be a long grind this year, but I'm glad people are paying attention.

      Delete
  2. Nate Silver acknowledged three days ago that his model is more aggressive than others based on recent data, and he was basically hinting that his model isn't the best at certain times during the race.
    http://fivethirtyeight.com/features/election-update-why-our-model-is-bullish-on-trump-for-now/

    ReplyDelete
    Replies
    1. "Aggressive" is a strange word to use. In some ways, I'd say my snapshot that says we don't have a horse race is more "aggressive" than the ones that say we do.

      Again, I am very skeptical that anything can really be a forecast of what the political landscape will look like 100 days from now. All I know from experience in 2004, 2008 and 2012 is that in terms of a chance to win, Trump's odds are much worse than the lowest points for Kerry, McCain or Romney. He's running a chaotic campaign and he isn't expanding his base.

      His strategy right now is denigrating Clinton. The problem with that is his negatives are worse than hers and he is so vain he fails to recognize it.

      Thanks for stopping by, I hope you will be here again.

      Delete
  3. Real Clear Politics "no toss ups" map shows Clinton at 322 and Trump at 216, but I believe that they are strictly polls only.

    ReplyDelete
    Replies
    1. RCP is one of my two sources for polls, along with the HuffPo website Pollster. Each source presents polls the other source does not.

      RCP and I use different methods for freshness dating and I'm keen on the median instead of the average, a method also used by Sam Wang of Princeton Election Consortium.

      I'll publish a page discussing my methods later this week.

      Thanks for stopping by.

      Delete
  4. Replies
    1. I do want to say it's early and things can change.

      There's a waggish saying that statisticians have the same problem artists do. We fall in love with our models. My model is much simpler than Silver's and is not as skittish. It's also served me well in the past two presidential elections. 2004 was my first year and the model went through a lot of trial and error that year. Sadly, the website where the data was stored is gone.

      2004 was a much closer electoral contest and in the last weekend, the numbers actually wobbled back and forth between Kerry and Bush. The Wall Street Journal miraculously found the blog and I made the prediction that Kerry would win on the Sunday before. New polls came in on Monday and tipped the balance slightly to Bush's favor, but it was past WSJ's deadline. That year, it was all about Ohio and Florida and Bush had to win both, which he did. Right now, Trump has a much tougher path to victory.

      If you have questions, feel free to ask.

      Delete
  5. What do you think of Wang and Rothenberg? Are there any others to follow for comparison. I've heard Wang say median not mean. Also, I am more than curious about the reasons why you seem to distrust Silver so much. All very interesting.

    ReplyDelete
    Replies
    1. Big fan of Wang. Both of us were located by the WSJ reporter back in 2004. I'm not in touch often, but he always responds to e-mail. You are right that we agree on the median being the more reliable method. It doesn't get pulled around so hard by outliers.

      I wasn't aware of Rothenberg until you mentioned him. I'm adding him to my bookmarks. Thanks.

      And then there's Silver. My feeling is he believes his press clippings too much and makes too many excuses when his models fall apart. Working for ESPN, he has tried to bring his methods to sports data, often with very embarrassing results. Sports are so much harder to predict, but I get this "I'M NATE SILVER!" vibe from him even when he crashes and burns on predictions.

      I know when I predict sports, I'm guessing. I used to gamble, mainly poker and backgammon, but I've given it up. I'm 22 years older than he is and I see him going through some of the same phases I did. This gives me a strange mixture of compassion and contempt.

      Thanks for stopping by.

      Delete
    2. Thank you. I'll check in again. (It's Mitch McConnell, not McDonnell. Unless there IS a Mitch McDonnell.)

      Delete
    3. Thank you for your unpaid copy editing. I have put a note at the bottom of the post giving you credit, but no pay.

      Did I say this was unpaid? I think I did, but remember, you'll get a lot of exposure!!

      Snark mode off: Really, thanks.

      Delete
  6. Silver has NC leaning Trump with 60% confidence, and wobbling back and forth the last few weeks. RCP has it a toss-up. How are you arriving at solid Clinton? I'm asking this recognizing solid Clinton makes more sense to me from the nos. than what Silver says.

    ReplyDelete
    Replies
    1. Polls that count in my estimation are the ones from June 20th through July 11. In that time on RCP and HuffPollster, I have six polls and Hilary leads in five. The median lies between a small poll with a huge lead (n=300, 10 point lead) and a fair size polls with a median lead (n=600, 6 point lead).

      North Carolina is very close to changing as soon there is some new polling. This early in the race I have to stretch to have more than one poll being used, even in battleground states. Come September and October, I'll be drowning in data.

      Thanks for stopping by.

      Delete
    2. Thanks, I'll be stopping by here often.

      Delete
  7. Thanks for bringing some sanity to this!

    I'm here to say there's no fucking way Oregon goes for Trump. Yes, the eastern two-thirds of the state is Trump territory, but there's no population there. The rest of the state (where the people are) is ultra-blue.

    The only reason I can imagine as to why we aren't in the solid Clinton camp yet is that there are a metric-shit-ton of Berniebro hangers-on here that haven't gotten the memo yet. They'll come around soon enough.

    ReplyDelete
    Replies
    1. Currently, Oregon's number comes from a single poll by an off-brand Republican pollster called Clout. I hate having a single poll representing a state that isn't a clear stone cold lock, but the poll in July was the first since May, and two months is way beyond my freshness limit.

      I'm so happy to have people coming by and commenting. I'm not sure I'll have time to answer everything in a timely manner when the data starts coming in hot and heavy.

      Delete
    2. Thanks for the reply. Hopefully newer polls will show a closer reality of what Oregon's like (we haven't voted GOP since... '84, I think).

      BTW, I got here from Tengrain's place. He speaks highly of you. I'm sure I'll be back.

      Delete
  8. Another Q. How are internal polls the campaigns use different from, if at all, the polls we see? Do you have access to them?

    ReplyDelete
    Replies
    1. I have no inside information. The polls I use are from Real Clear Politics and Pollster, the Huffington Post website.

      Delete
  9. This comment has been removed by a blog administrator.

    ReplyDelete
  10. This comment has been removed by a blog administrator.

    ReplyDelete
  11. This comment has been removed by a blog administrator.

    ReplyDelete
  12. This comment has been removed by a blog administrator.

    ReplyDelete
  13. This comment has been removed by a blog administrator.

    ReplyDelete
  14. This comment has been removed by a blog administrator.

    ReplyDelete
  15. This comment has been removed by a blog administrator.

    ReplyDelete
  16. This comment has been removed by a blog administrator.

    ReplyDelete