So....I've been building my 2026 forecast system. It's not perfect so far, but I do have enough to discuss the house and senate somewhat.
Senate
So, where do things stand? Well, it looks like we're likely not gonna see much progress at all here. I dont have data for all states, but right now, it looks like we're gonna end up with another 47-53 Republican result, but with us trading Michigan for North Carolina.
Now, to discuss individual results though, this is where things get interesting.
Michigan is weird and the result depends on who wins the democratic primary. While for most races I just went with the most likely candidate to win, the race between Stevens and McMorrow is in a dead heat. If Stevens is the candidate, dems win with 1.3%, but if McMorrow wins, republicans win with 1.5%. And keep in mind these guys are pretty close to each other in the primary. So I averaged the two together and I basically got R+0.1.
However, we seem to have a soft lock on North Carolina to make that up.
Maine is another contentious one. If Mills wins, the republicans keep the seat. If Graham Platner wins, the democrats win. Mills is currently ahead in the primary, so the republicans are more likely to win. It's a lean R for me.
Ohio and Texas are surprisingly squishy here given how strongly they went for Trump and are de facto tossups. In Texas, there's uncertainty as there are primaries on both sides. For simplicity's sake, I went Paxton on the republican side, and that gives us R+1 for both Crockett and Talerico (both of them are tied in the dem primary currently).
Honestly, I think the polling for Ohio and Texas is too good to be true. I think its weird they're so squishy while Maine and Michigan are so uncertain themselves. Even if we did flip 8 points since 2024, we're still looking at R+5. Still, to keep my own opinion about what way the polling error is likely to go out of it, that puts all 4 seats in near tossup territory, where the GOP only has a 60% chance of winning the senate, the dems have a 31% chance, and there's a 9% chance of a tie.
House
So, before I post the house forecast, I want to explain what I did here, and what its methodology is. First, here's the full unredacted version of it. I wont lead with this in future updates because look how small it is, and how hard to read it is, but I at least wanted to post a screenshot of it so people know what I'm looking at when I make these things.
So, here's what I did. I realized using Cook PVI scores themselves are kinda misleading. I know in 2024 I had a system where I used the actual PVI score in lieu of polling data, but the PVI scores dont necessarily translate that way voting wise in practice. So instead, I took their ratings for individual districts, excluded the safe ones, looked up the 2024 election results in each of them, and put them in the above spreadsheet. This gave me 60 districts to look at. And then...I took the generic congressional vote from 2024, the ACTUAL results, btw, not the polling projection (it was R+2.7), I looked up the current generic congressional vote for 2026 currently (D+5.2), and then I shifted the above 60 districts the net difference (7.9 points) to deduce how I would expect people to vote if the public actually did shift the 7.9 points it appears that they did (and I believe this is actually accurate, given I saw something else recently suggesting the public shifted 8 points on biden vs trump).
Will this be accurate? To some degree. I mean, individual districts very, and I know gerrymandering attempts in some states are throwing a wrench into things. However, when I actually looked at what districts are likely to flip from gerrymandering, I find the GOP is stealing 5 districts which can be seen above. Dems are stealing 2 back. And then there's a bunch more that are accounted for among the safe districts where the dems are getting a net 2 more. So, the GOP is taking 5, the dems are taking 4 back, and the net difference is R+1.
Which....brings me to the results. Here's the short version of the above chart with only the important stuff I need for an overall forecast:
So....as I said, we're seeing a shift of 7.9 points from the 2024 result. And keep in mind, the GOP only won 220-215 under such conditions and I actually thought the house was a tossup in 2024. But with this shift to the left, we're seeing a 97% chance of democratic control. As for the margins, the model spits out 235-200 Dem, but again, keep in mind the net effect of gerrymandering seems to currently be 1 seat, so we're talking 234-201 Dem.
That's my current prediction, that's what the data says and I think it's fairly accurate. It's basically a repeat of the 2018 results. That sounds about right. And yeah. Expect the GOP to lose a lot of seats.
Conclusion
I may make a governor map at some point, but I'd like to see more data come in. I also might switch to a more conventional house election chart later on. However, for now, this seems to be what I can do this early and with limited polling data (to be fair congressional races are horrendously underpolled).
But yeah....2026 is shaping up to be a bloodbath for the republicans. Assuming free and fair elections they will almost certainly take the house assuming polling is accurate. Their margin will basically be a repeat of 2018's midterms.
The senate is more interesting. While I still give the edge to the GOP, it's shaping up that there is a real and growing chance the dems can just win big enough to flip states I previously saw as near untouchable like texas and Ohio. However, there's a lot of "ifs" at work here, and there's a lot of data that doesnt make sense. Michigan should be near solid blue if texas and ohio are in play, and if we applied the same trick I did to the 2024 senate results, i think the odds of the dems flipping those states is a lot lower. I'd expect more like R+5 in them honestly. Even though I think winning maine, north carolina, and michigan are plausible.
Still, keep in mind it is early. These are very early predictions with very limited data. Nothing about these projections are final, they're about as valid as this prediction from 2024. Although that one actually did get it right...
Hmm...
Well, we'll see. Just take it with a grain of salt.
No comments:
Post a Comment