Nate Cohn and pollsters tell me I'm wrong. That studying the numbers is irrelevant because Republicans and NPP voters won't vote the way they have in the past. They're voting Democratic. If they're right, these predictions will be awful and you can point a finger at me and laugh. I'd rather get my predictions wrong because I used an incorrect assumption than get them right because I'm trying to conform to what the pollsters tell me will happen.
We have only one day more of VBM returns. The last day before election day is usually pretty big and it’s possible I’ll want to adjust my predictions before then. But they’d only be small tweaks. The link for my past and current predictions will be at the bottom. Don't want to spoil it. if you read this blog in 2014 and 2016 you can find the posts with these predictions.
In 2014, I made predictions on 11 congressional races. I got within 2% of the Republican voting percentage on 8 of them. I really blew CA-52 when I called it big for Carl DeMaio, but CA-52 was changing fast and the data hadn’t caught up with it. I called tight races in CA-3, 9, 16, 24, and 31 when the media didn’t even bother to look at them. The races were tight.
In 2016 I expanded to 20 contests. I was within 2 on only 10 of them, although I was within 3.4 on 15.
While other people tentatively make predictions on which candidate will win, I boldly predict actual percentage of the vote. So even if I'm off by 5 points, you probably won't find anyone else who dared to even try.
I developed a formula based on adjusting the VBM for the environment. I decided to use that rather than play around with the percentages to get something I agreed with. Live by the math, die by the math.
I’m skipping a few races that I predict but don’t expect to be competitive, but they are on the spreadsheet. CA-7: This district has been competitive in each cycle. The VBM margin suggests it could be competitive again. So an upset is possible. But I adjusted in a Democratic friendly environment to Bera 52.7%-47.3%.
CA-10: I was pessimistic on Jeff Denham earlier, but as of now there is only 1 more Democratic ballot than Republican ballot. That’s a real positive for a Central Valley Republican. If the wave peters out anywhere it may be districts like this one that don’t have a lot of white collar suburbs. Denham 51.5%-49.5%.
CA-16: Jim Costa got a scare in 2014 and so far VBMs are more Republican than that year. This one will be more competitive than people think. Costsa 52.0%-48.0%.
CA-21: This district is also more Republican than 2014. David Valadao has had relatively easy elections and I don’t see this one as any different. Valadao 59.4%-40.6%.
CA-22: Devin Nunes has been on the Democratic wish list all cycle. I don’t see it as being close, although it’ll be Nunes’ smallest margin. Nunes 56.8%-43.2%.
CA-24: This district has been close, even in 2014 when Republicans spent nothing and supposedly had a candidate who was too extreme. I gave Democrats the benefit of environment here. Carbajal 53.3%-46.6%.
CA-25: I know the NYT/Siena poll was kind to Steve Knight. Full Disclosure: I did some volunteer work for the campaign in 2016. Fortunately, no one there knows me. So they won’t hate me for this. Hill 51.4%-48.6%
CA-39: The polls may be down on Young Kim but my numbers suggest otherwise. I think if she goes down, every vulnerable Republican does too. Kim 52.2%-47.8%
CA-45: When I ran the numbers I came out with a 50.0%-50.0% tie. Yes, seriously. Of course that’s just me inputting the numbers I think make sense. Since the VBMs have been trending away from Walters, Porter 50.1%-49.9%.
I know that the way people look at predictions they’ll judge I’m right if Porter wins and wrong if Walters wins. That’s not how you should look at it. If Walters wins with 50.2%, my prediction was brilliant. If Porter wins with 54% my prediction was awful.
CA-48: I’ve maintained that Rohrabacher’s past voters like him and I think the data shows it. Rohrabacher 52.7%-47.3%.
CA-49: For a while there Harkey looked like she had a shot, but I do think it’ll be much closer than people think. Levin 52.0%-48.0%.
CA-50: When an incumbent is in a scandal voting can get funky. Reliable voters defect not because they like the other guy but because they're disgusted. I don't know if that'll happen here, but I'm saying no. If they do defect I think you'll see down ballot candidates in SD-78 and AD-75 do as well as I'm showing here. Hunter should win 59.2%-40.8%.
Here's the spreadsheet