Election Update: 10/23/26

 Okay, so we're getting down to the point where the election is in less than 2 weeks, and next week's Friday update will be the last of the season. We've come a long way, and it would be a shame to choke now. Sadly, that's what I'm gonna be spending much of today discussing.

Before we get to that though, I want to discuss the rest of the election season. So, next week I'll post my last friday update. I'll then make a semi final update Monday, with me finalizing my predictions on Tuesday before polls close. I also have been toying with a mid week update next week, although doubt there will be significant need for it. With that said, let's get to it.

Senate

 

So...if people have been following this stuff lately, YIKES, this map is a marked shift toward the republicans. It's only about a one point shift, but seeing movement in the republican direction does concern me. If we want to understand why it's so concerning, let's look at this chart:

Yeah....we were going in a direction where the democrats were getting a more and more solidified chance over the senate, but now they're regressing to the mean. This is by no means fatal to them. But this movement can be problematic. Im saying can be because, I'm not sure about it. On the one hand, I was thinking, okay, maybe things are shifting republican because the last minuters genuinely like MAGA. It's a terrifying thought but sadly my opinion of the median voter is such where it wouldn't surprise me. 

However, there is another explanation, which is far more benign in a way, but also concerning in its own way. See, polling aggregators use the last few polls an organization puts out to judge their performance for the election season. This is putting pressure on the polling organizations to herd, rather than putting out risky polls that could sour their reputation (see: Ann Selzer's 2024 poll). After all, if someone comes out with a bold prediction and it's wrong, they're never gonna hear the end of it, and they won't be included in future election cycles due to being wrong, and kind of become less relevant. On the other hand, if these polling organizations play it safe, and post generic results indicating what they THINK will happen rather than what their data science suggests, well, that makes their final polls less risky, meaning they have a greater chance of being included in the average in future cycles. 

This is problematic because we should want pollsters to put out the truth no matter what. Accuracy is important, yes, but the integrity of the industry is also important, and herding kind of undermines the trust of the industry. people already say we shouldnt listen to polls as they're often wrong or fake, and if they're basically faking or weighting their results in a way to get a result that looks publicly palatable and noncontroversial but is wrong, that's problematic. We should want pollsters to put out the truth no matter what it is, and no matter how controversial it is. 

I admit, even I kind of think the results will be more conservative than polls have been indicating. I think dems are favored to win the senate, but recognize that it's very possible for republicans to overperform. last week, I pointed out the chaos a R+2.5 environment would cause to senate results. This week R+1.5 would deliver the same effect, because the tipping point shifted almost a point since then, and most states that were in the D+2-3 territory are now in D+1-2 territory. This is still "lean dem", but it's more 60-65% probability vs around 70%. 

Given the ambiguity as to why this is happening, it also gives us conflicting narratives of what's happening. It could indicate the electorate shifting more toward the republicans in the final weeks of the campaign. It could also indicate, if the herding explanation is accurate, that the final result will be far more DEMOCRATIC than anticipated. The polls could be trying to compensate for some perceived left wing bias, meaning the polling is more right leaning than the actual reality of things, and we could see democrats OVERPERFORM the polls. 

Which explanation is true? I dont know either way. I just wanted to talk about it.

As it stands, Ive given up trying to predict which way the politics winds are blowing. The data is the data, I'll stand with it, take it or leave it. The probabilities are the probabilities as well. Lean D, but decent chance of an R outcome. 

House

Regardless of what's going on with the senate races, one thing I can confidently say is the democrats still have a lock on the house. Once again, it's more a matter how much the democrats will get, not whether they take control. 

I also can say we're getting more polling this week. Not enough to justify making its own model, although I might sneak in a polling hybrid model by election day kinda like the 2024 model. We can compare which one ends up being more accurate. I won't say the polling model is necessarily better or worse for dems than the fundamentals driven one, but it does give different results. While sometimes it corroborates my model, other times it gives radically diverging takes on a handful of house seats. Sometimes it's more optimistic, sometimes less so. Again, not really better or worse, just different. 

Governors

 

Here I dont really have much to say and the results are similar to last week's.

Conclusion

With that said, once again, this is one of the last ones of the season. Next week is the last weekly one, and then after that it's gonna be my prediction prediction. 1.5 more weeks until this stuff is over. I kinda cant wait. By the time we get to this point, it's like...I just want the season to be over. I like following races, but you know, it's like...let's do this already. I wanna settle in on election night and watch results already. 

But yeah. House will likely go to the democrats. The senate still leans democrat but it's becoming a bit less certain. The governors are what they are. And yeah, that's this week's update. See you next week.