538 model & poll tracker thread

[emailking]

Are the orders different?

[emailking]

Georgia flipped back.

[emailking]

I think the NBC poll pushed it back to 87.

[emailking]

I don't know about anyone else, but the number I'm watching is the "Biden wins in a landslide" probability. It's been as high as 37, now back to 36.

If we keep getting polls like this from now until election day, it may approach 50.

Anything more than Biden + 7 and a D senate I'm happy with

Dems are more likely to lose seats than not lose seats in 2022 so you want as much cushion as possible so judges can still get appointed after the first 2 years.

[emailking]

Filibuster has to go to pack the courts too.

[emailking]

The forecast is actually moving all over the place all day if you look at the decimals. It rounds off to the whole number on the forecast page, maybe to hide this somewhat, but it gives the impression it's static. They update it basically within minutes of each poll release.

[emailking]

Thank you, professional statistician, for "everything is a 50/50 chance tbh" A+ take

He's saying those are approximately the odds of those scenarios as predicted by the model at the moment, not that diametric alternatives are always equally likely.

[emailking]

https://twitter.com/NateSilver538/status/1319787685758861313

So this means he has no chance to win if polls don’t move in his favor?

About 1/16 I think.

[emailking]

https://twitter.com/NateSilver538/status/1320843000575152131

[emailking]

He said all the double digit polls from a few weeks ago were falling off, but this was balanced by the election getting closer and uncertainty going down. That's why he's been stuck in a 2% range for a few weeks. The polls this week though will count heavily toward the final prediction.

[emailking]

So here's some weirdness from the negative state correlations side of things:

If Biden wins Georgia and North Carolina, he has a 16% chance of winning Mississippi.

If Biden wins Georgia, North Carolina, and Minnesota, he still has a 16% chance of winning Mississippi.

However, if Biden wins Georgia and North Carolina, but loses Minnesota, he suddenly has a 53% chance of winning Mississippi.

Could just be the finite number of simulations. Probably very few maps where he wins GA & NC but loses MN.

[emailking]

The dot matrix is rounded off. Favored vs. clearly favored isn't.

[emailking]

The final prediction is Biden 89.16% chance of victory, Trump 10.39% chance.

Biden takes FL, GA, NC, AZ, NV, PA, MI, WI. Trump takes TX, IA, OH.

[emailking]

I doubt he's ever cared in that way. It's nice to be monetized but that's not going away, heck he just nailed the WS. I think the model did just fine, especially given the raw data.