Tuesday, October 22, 2024

Election Update 10/22/24

 So, I've decided that from here on out, I'm gonna start doing these twice a week given how close to election day we are. We're officially 2 weeks away, and I figure updating every Tuesday and Friday will work. Now, on the Monday before, I'll do a semi final election update then, and then finalize it the next day. 

With that said, lets get to it.

Presidential

As I said, it's still 42-58 in Trump's favor. Trump has an edge in all 7 swing states. Other sites are a little more optimistic, but to be fair even many of them are starting to admit Trump is taking the lead.

It's just what it is, folks. Let's not "cope", let's not go "well ackshully the polls are wrong". They CAN be off, but that's literally why i give probabilities here. In any 7 random election outcomes with these probabilities, I'd expect Harris to win 3, and Trump to win 4. So you're saying there's a chance? Yes, there's a chance. Trump won in 2016 with similar odds. But at the same time, kerry lost in 2004. Because we are talking coin flip territory still. 

Still, as I said, I ultimately believe it comes down to trend, and who does the trend favor? Trump. 

Again. It is what it is. Cope or doom all you want but that's where it stands.

Senate


Similar outcome in the senate, not much changed. Quite honestly, you may be wondering why I'm even bothering with a forecast if it's basically the same as Friday's. 

I'll get to that next.

House


 

Holy crap, are you doing a house forecast after all?! Yes, I am. it's a pain in the neck, but I spent 2 hours building a chart for the house polling data. It was a pain but I don't want to not at least take a shot at it. So...this is something I'm gonna have to walk people through.

I started with RCP's house map to look at what districts were in play. They start out with 32 districts with the dems having 196 safe seats and the republicans 207 (since leans/likelies on RCP often are beyond the margin of error). Still, because they're not always beyond the margin of error, if I could find polling data for other districts and it was within 12 points, I added it to this map.

Now, you might notice, a lot of districts don't have data for them. So how did I come up with a forecast for them? Easy. I took the generic congressional vote and combined it with cook PVI. Cook PVI looks at how a district generally votes relative to the country as a whole. a D+2 district votes 2 points more democrat than average. A R+2 district votes 2 points more republicans on average. And then I look at the generic congressional vote, which looks at how the country intend to vote nationally. It's currently D+1. So I assume a D+2 district will actually vote D+3. And a R+2 district will vote R+1, because the GCV indicates that the environment nationally is D+1. So yeah. That's my attempt to use fundamentals to estimate how a district will vote without any polling data. Is this a perfect solution? No, but it's an attempt. And it's giving me, I think, reasonable results. 

Anyway, from here, I use my chart to tally up the results, and as one can see, there's no winner here. i get 214 dem seart, 216 rep seats, and there are literally 5 tossups, 3 of which are R+1 districts whose partisan lean is cancelled out by the D+1 GCV. So as such, the house is anyone's game. I give it a 50-50 shot at either party. Similations don't quite give me the same result, but before I get to simulations, a word about that.

I'm NOT building a full simulator for the house. It's too much work, the chart is, as you can see, twice as big, and I expect to keep adding to it as time goes on. I've had to add 3 extra rows today for races that just got polling data in today. I expect this to continue, and given it takes hours just to build a simulator, and tons of time to maintain it, in light of my discussion on "efficiency" the other day, I will NOT be building a full on house simulator. I do, however, have the smaller, lower maintenance simulator to give me 100 outcomes in about 5 minutes, and I did do simulations on it.

That said, simulation time.

Simulations

I won't post my charts, but TLDR:

Presidency- 576 Trump, 404 Harris, 20 ties.

Senate- 801 Rep, 22 Dem, 177 Ties

House- 62 Rep, 38 Dem

So basically, the simulator now gives results similar to the numerical odds for the presidency, and Trump wins more than Harris, although Harris isn't out of it.

The senate leans overwhelmingly republican, with them winning over 80% of simulations, and the best the dems can realistically hope for is a tie.

For the house, the simulation favors republicans simply because there are more dem seats on the table open to being overturned. Dems start with 191 seats, republicans with 203. The democrats need 27 to win the house. The republicans need 15. The republicans simply have a shorter path to victory and that's reflected in the simulator. Otherwise, yeah I would say it's gonna be close.

Conclusion

And yeah, that's my prediction as of today. I'll do another one on Friday like normal, but this is where things currently stand. Take it or leave it.

No comments:

Post a Comment