Founder of Silver Bulletin: https://nitter.cf/t.co/mYtb4rgUyT, and cohost of Still Counting podcast!
New York
Joined August 2008
- Tweets45.5K
- Following1.8K
- Followers3M
- Likes10.8K
Topline prediction market odds are catching up with FLIPR, too.
FLIPR update is in. Alaska goes from "Lean D" to "toss-up" with the Peltola scandal flag, though it's on the D side of "toss-up". Otherwise, another very good day of polls for Democrats.
natesilver.net/p/nate-silver…
As we do the final rounds before Election Day, we'll add roughly 10 scandals to our database today that were disclosed since the start of the campaign or that we missed originally. A fairly even mix of Ds and Rs. There is a penalty for these in FLIPR's "fundamentals" estimates.
FYI, we're making a patch to FLIPR in MT-Sen to reflect the weird nature of the race in Montana that we don't think FLIPR was handling quite right. I guess I'll just put this in one long tweet.
Montana features a Republican (Kurt Alme), a Democrat (Alani Bankhead) and a Democratic-leaning independent (Seth Bodnar) who would presumably caucus with the Democrats. De facto, it should be handled more like Alaska, where there can be more than one candidate from the same party on the ballot and FLIPR adjusts for this in various ways.
This has two major implications:
1. Bodnar and Bankhead's vote should tend to be inversely correlated. In other words, as he rises she'll fall, and vice versa.
2. The various adjustments FLIPR Classic and Deluxe make, i.e. for fundamentals and expert ratings, need to be shared between Bankhead and Bodnar as "co-Democrats" whereas before they were only being applied to Bankhead (and Alme).
While #1 has somewhat ambiguous effects, change #2 will hurt Bodnar. FLIPR was confused why Bankhead was *so* far behind Alme, and therefore FLIPR Classic was adjusting her vote upward and Alme's downward as it applied the fundamentals while leaving Bodnar's vote unchanged. But the reason is that the D- and D-leaning vote is basically split between Bankhead and Bodnar, deflating her vote.
With the patch, Bodnar's projected vote won't change much, but because Alme was too low (and Bankhead too high), he now has a higher target to aim at.
We expect the race to shift from "Lean R" to "Likely R" as a result of the patch. It's definitely an interesting race and Bodnar has some chances and has been gaining in the polls, bringing him within striking distance, but the tightening polls mean Montana needed more scrutiny than back when Bodnar was way behind. We think this handling is more internally consistent with the precedent we'd already built for those multi-way Alaska races. We'll do a new run of FLIPR later tonight that reflects the update.
Nate Silver retweeted
Friends! Today, we'll be releasing our final poll @TheArgumentMag AND previewing the midterms with @NateSilver538, @galendruke, and @mattyglesias.
Please join us at 7:00 PM Eastern for a great conversation and a giant nerd-out over the midterms!
luma.com/xt3uy8gq
Very strong polls for Democrats in individual Senate races over the past couple of days. Polls can be wrong, that's what models/probabilities/markets are for, but I think we're past the point where you can honestly say "polls show the Senate is too close to call".
And speaking of models, here's today's FLIPR update.
natesilver.net/p/nate-silver…
As compared to when I wrote about it this weekend, the gap between FLIPR and prediction markets has really tightened in individual Senate races to the point where there's almost none now. Markets have moved in FLIPR's direction in races where there were bigger disparities.
Here's Sunday's article where I talked about the differences ... but again, several of them have been erased now. natesilver.net/p/lets-talk-s…
This has turned into kind of a weird election. And this is just based on social media, not any reporting. But based on vibes from Republicans, you'd expect Democrats to end up with 55 Senate seats. But based on the vibes from Democrats, you'd expect 50.500001.
Still Counting episode is out. youtube.com/watch?v=kATFvj53…
"I'll knock out a quick post on South Carolina" turned into a 3500-word story, but here is today's slightly belated Model Talk.
natesilver.net/p/some-though…
That South Carolina NYT poll, plus a Graham +3 from Citadel that also entered overnight, didn't actually move our South Carolina polling average at all! The race has been polling this way for a while.
Still, it's good to see it confirmed by high-quality polls. And FLIPR's chances for Andrews are up a *bit* as it gains more confidence in the polling. FLIPR update here, and I'll see if we can't get a newsletter out on South Carolina later today/tonight. natesilver.net/p/nate-silver…
Nate Silver retweeted
We had @NateSilver538 on Central Air to preview the election, talk Cheesecake Factory and assess what kind of gay Trump is. Plus exciting plans: A listener meetup in Manhattan on November 2, and a livestream with The Fifth Column on election night! Info: centralairpodcast.com/p/the-…
