So, we've made it. Election day is tomorrow. So I'm gonna START making my final prediction. Now, in other years, I would use the old fashion charts where I would fill everything in myself and do all the calculations, and THAT would take me hours. But as we know, I now have excel charts and I use them for my predictions. With that said, I will post charts now, but I may edit and update them tonight and tomorrow as needed to reflect the final count.
Tomorrow I may either just update this article or post final charts to make my predictions finalized. I really do like to wait until last minute to get ALL of my predictions in.
Presidential
Okay, so, I've put a lot of thought into how I'm going to do this, and I decided I'm going to do TWO predictions. My formal prediction is going to be my model's prediction, based on polling. But then I'm gonna make a second, informal prediction because well, polling is crap this year, and let's face it, error happens either way anyway.
So, formal prediction: my model has Donald Trump winning 257-281 with a 54% chance of victory. It's EXTREMELY narrow, and it is basically tied. PA is the deciding state, and it's a true tossup, with us having no clue what's gonna happen there. Still, the polling gives Trump a slight edge there.
Now, I'm going to be honest, at a 54% chance, with Harris still having a 46% chance, this race is a tossup. Anything can happen. We can have movement up to 8 points in either direction and it can still be accurate. If Trump overperforms by 1 point, he wins Wisconsin and Michigan. He overperforms by 4 and he gets New Hampshire and possibly Minnesota. He overperforms by 8 and he can theoretically get Oregon, Virginia, and New Mexico. I don't think he will overperform more than that, that's outside of my 95% confidence interval, but it can theoretically happen...about 2.5% of the time.
In the other direction, one point and Harris wins Pennsylvania and Nevada, giving us a 276-262 win. 4 points and she wins North Carolina, Georgia, and Arizona. 8 and she can take Iowa, Kansas, Maine CD2, and Florida (Ohio and Texas WERE within this range, but RCP changes the averages last minute so now they're out of the margin of error). I don't think she will outperform that. By the way. That selzer poll? The way people were framing it, it had her overperforming by 10. That's why that was so bonkers.
Informal prediction
With all of that said, what about my informal prediction?
Well, here's the thing. I think that the energy is on Harris's side. I think that the early indicators trending toward a Harris overperformance. The selzer poll seems to indicate that she will overperform in the rust belt, although I highly doubt it's by the truly bonkers amount predicted. I think that those comedian's comments about puerto ricans and other races ended up tanking trump's chances in PA, given the large latino vote here. I think that polling this year has significant issues with herding, and while I feel obligated to stick to my model for my official predictions, unofficially? I highly suspect that they're herding toward 50-50, where we're getting massive discrepancies in the swing state and popular vote polling (harris should be about 2-3 above the tipping point state). I also suspect that they overtuned their models to overrepresent trump voters, trying to avoid making the same mistakes as 2020.
In addition, the campaigns' internal polling seems to indicate a pro harris direction. Trump is already crying it's rigged in PA, and as some have pointed out, you don't do that if you're winning. I also heard that Trump is possibly struggling in NC, which is why he's trying to shore up down there last minute rather than spending ALL of his time in PA. You don't do that unless you're afraid of losing it. On the flip side, I've heard rumors out of the harris campaign that they're feeling confident and their polling has her winning over voters by double digit margins last minute. The reason? Again, that fricking Madison Square Garden rally. That unforced error really seems to be the October surprise. Well, that and she finally improved her messaging somewhere to focus on improving peoples' lives rather than doing the Liz Cheney crap.
So that said, how do I think the swing states are gonna go?
Michigan & Wisconsin- I think polling is persistently underestimating Harris in Michigan and Wisconsin. My model shows them within a point, but I think they're probably up more like 1.5-2 points. Maybe even more, but I'm not gonna push it. I am relatively conservative in my predictions in the sense that I'm not gonna get super ballsy about them. But I did notice that when I tried my anti herding measures last week MI and WI should've been a good point left of where they were, and the selzer poll really kinda sealed it for me.
Arizona & Georgia- On the flip side, I think polling is about right on Arizona and Georgia. I think that they're solidly lean republican, and probably going R by a margin of 2-3. Maybe even more. The polling seems to range from a dead heat to being up by a lot. The point is, harris isn't favored down here at all, and I would give these to Trump.
North Carolina- I would say this one is like Georgia in a sense, but a little further left. I think the R+1-2 is accurate. I think that it's going further left than it otherwise would because of the Mark Robinson thing. I think he's really tanking trump down there, and he's the reason it's not up for Trump by 3-4. But still, I'm not gonna get overly confident. I think it will still go lean R, but I'm not sure.
Pennsylvania- I think that given its combination of rust belt and northeast nature, it's gonna go Harris. On the west side of the state you got the more rust belt midwestern vibes where the selzer poll predicts an overperformance. On the east side, you got a lot of latinos and puerto ricans who are PISSED over the MSG thing. And it is making up their minds, and they are trending toward Harris. So my opinion? It's gonna go to Harris. Idk the margin, I'd be inclined to say maybe by 1, but idk, it could be anywhere from 0.1 to 2.
Nevada- This is the one I'm having the hardest time predicting. It's not a rust belt state so it doesnt get a power boost for harris. And while the democrats won it in 2016 and 2020....it does seem like it's not going super smoothly for harris. Nevada has been all over the place. Sometimes it polls like a D leaning state, other times an R leaning one. For most of this election cycle, it's been performing closer to an R leaning state I'd say, polling most similarly to NC, AZ, and GA, so I'd be inclined to put it in that category as a tilt R.
EDIT: Nevada shifted further toward the democrats, I'm now shifting my prediction to tilt D. This shifts my informal prediction from 270-268 Harris to 276-262 Harris.
As such, I'm going to say this, even though my model has Trump winning, informally, I think Harris is gonna win 276-262.
Here's the map for that.
So yeah. This is effectively D+1 relative to my formal prediction. I don't wanna overdo it and get overconfident, although I do think that Harris can overperform even this. She overperform by 2, 3, or even 4 in my estimation. This could flip North Carolina, and even Arizona, and Georgia.
But I don't see it winning Iowa or anything like that, unless the polling is so massively off Ann Selzer was somehow right about everything, and I ain't basing my entire prediction off of a single poll. No. I do think Harris will overperform in the rust belt, but I'm gonna say by 1, and maybe more, but we don't know exactly. I don't think this will trend to the south though.
If I'm wrong on any states here, I'm gonna guess Nevada or maybe North Carolina. Nevada since it's a tossup that can go either way IMO, and north carolina, because I've seen some subtle signs that it could overperform for democrats. Still, I'm pretty comfortable with guessing 276-262 at this time. I think that's the map we're most likely to see.
Simulations (Presidential)
Doing 100 simulations, I get:
35 Harris
61 Trump
4 Ties
So, originally, I had it 48-49 in favor of Trump, but then RCP cleaned up their averages, a lot of non swing states swung right by 1-2 points, and now Trump is winning most outcomes. Oh well, this simulator is experimental and not really indicative of my final forecast, it's just for fun.Senate
Yikes. So....the senate I'm far less rosy about the democrats' chances. I've basically been predicting 49-51R for most of this cycle, outside of when I was huffing 538's copium about Dan osborn winning nebraska. It can still happen, it's just a long shot.
And now we are losing Brown's seat in Ohio. This keeps getting better and better (for the republicans).
So, formal prediction, republicans control the senate with 52 seats, and they have an 86% chance of pulling this off. Yeah. Wish I could be more optimistic there, but we're losing WV due to Manchin retiring, we're losing Tester's seat. And now we might lose Brown's. To be fair, that's still 50-50, but the energy has been going that way for a while.
If we overperform slightly we could keep brown's seat, but the energy has been going toward brown for a while now, and I'm not about to go against my own model on that one. It can happen, but I'm going to say I agree with my model, as that is what the evidence is telling me. I do think we stand a solid chance of defending the rest of the rust belt seats though.
Still, if we weigh possibilities. D+1, nothing changes because Ohio swung so far right. D+4, we finally flip Ohio but fall short of flipping anything else. D+8 we flip Texas, Florida, Nebraska goes to Osborn, and we even keep Tester's seat in Montana. I don't see that as particularly likely.
On the flip side. R+1 does nothing but solidify Moreno's win in Ohio. R+4 wipes out the rust belt trio and arizona. R+8 gives them Nevada. If I had to guess, I'm guessing the odds as are. I mean, it's fricking Ohio, man. It's a red states these days. Even if it is rust belt, it's very hostile toward democrats, and we're lucky brown is holding on the best we can. When he's out, we're never getting that seat back (this alignment).
EDIT: And since more polls now have Ohio going R+1, I'm just gonna call it for Moreno/republicans here. No informal prediction.
EDIT2: Now it's up to 1.7, yeah, I'm gonna give it to the GOP, it's lean R now.
Simulations (Senate)
Doing 100 simulations, I get:
Democrats-0
Republicans-98
Ties-2
I originally had a much more optimistic batch with only 86 republican outcomes, but then RCP cleaned up their averages, thing swung right, and now republicans won 98 outcomes.
House
So, I'm gonna say it up front, I'm gonna say that my house forecast is experimental. I have no idea what I'm doing with that. I built as much of a forecast as I could, but if I don't have polling data, I can't predict well. I did try to compensate for using the generic congressional vote and the cook PVI to estimate the lean of the districts and where things will go, but that's an imperfect measure.
So, here's my prediction:
So, i apologize for my house forecast being hard to read, the chart is really huge and hard to fit on my screen, but basically, I have the house in a dead heat, with the democrats winning 218-217, with a 57% chance of doing so. Much like the presidency, the house is a tossup, but I do have it slightly leaning toward democrats. Still, anything can happen, and you should take it with a grain of salt. Treat it like a 50-50 tossup.
Here's a map, although margins might be wonky as I made for a contest.
.png)
Simulations (house)
So 100 simulations for the house:
Democrats-49
Republicans- 51
The simulator seems to favor republicans slightly more, although in the final forecast with more republican seats up for grabs, it's kind of evening out. As I said, it's 50-50 just about. Treat it as a tossup.
Governors
So I havent been paying attention to the governors races, as most of them don't involve my state and hardly any are competitive. But for the aforementioned contest, I made this map.
.png)
The only two I really have anything to say anything about are the North Carolina and New Hampshire ones. New Hampshire seems to be lean R in their race. North Carolina is pretty much safe D (regardless of margins shown here, keep in mind i normally operate on a 1/4/8 system, this map is designed for a 1/5/15 system), because Mark Robinson is a perverted black Nazi and he's actually tanking NC so bad Trump can actually lose it.
Conclusion
With all of that said, that's my (semi) final forecast for 2024. I might touch up this stuff tomorrow if I get new data tonight or tomorrow, but suffice to say, the presidency is up for grabs. My model's official forecast is 257-281 Trump, but I'm informally going with 276-262 Harris since I think PA and NV will go blue.
The senate is a lot more in stone, with the GOP expected to win by 52 seats.
The house is a tossup, but I'm tentatively giving it a 218-217 win for the democrats, but yeah, treat it as 50-50.
Governors, who cares, look at the map.
No comments:
Post a Comment