Okay, so, new election update. This isn't a very good one (given my obvious biases), but I'm gonna report the facts as I see them.
Presidential
I'm not going to mince words. We are LOSING. New polls this week have eroded Harris's advantages in the swing states she was ahead in, although to be fair, Trump's advantages eroded in the states where he was ahead. All 7 swing states are now effective tossups, with 6 of them being in the republican column. While I must emphasize that this IS still a tossup, and Harris still has a healthy 45% chance of winning, it is a marked decline from where things were. Last week, we were a point in the other direction, and while the race was STILL in tossup territory, Harris had the advantage. She is losing is and the election is shifting to Trump. I don't like this shift, and I don't like where the energy is going. Remember what happened in 2016. Clinton was ahead, by A LOT (looking at my old predictions from this point in October I had her at like 90%), but then by election day she eroded to 56% and lost. I always said that because polls are a lagging indicator, we didn't really measure the shift fast enough for it to happen. We could be seeing that happen here, and we really don't want that. Even if the race is a tossup, just the energy shifting like this could mean that people intend to vote for Trump an ARE voting for Trump. Keep in mind, mail in ballots are being received and I just got word that my vote has been received by the state and I am locked in. So this is happening now. People are voting now. And that map is getting a little pink in the swing states.
Even worse, look at the trends.
I normally only include the first chart here. It's the election odds, but this is the strongest trump has been in two months. The second map, which measures projected electoral votes, is really breaking for Trump. Again, this is because 6/7 of the swing states are now tilting Trump. The only one we're holding onto is Wisconsin, and we're only holding onto it by 0.3%. The others we're losing by 0.2-0.9%. Again, not lethal in and of itself. A singular point shift in either direction could toss the race either to a 226-312 Trump win scenario, or a 319-219 Harris one. It's THAT close where any error at all can still shift the race either way. I just don't like this because again, when you're on the wrong side of a trend line, that can be dangerous. We still have about 3.5 weeks to turn it around, but I'm starting to get nervous.
Now, in all fairness. Let me present the counter argument to this narrative, and that is that the republicans are poll bombing the averages. They like to do this sometimes. There's a reason I tried excluding trafalgar and making my own averages in 2020. But that also ended up throwing my results way off, where I was way too bullish on Biden, and that's why he only barely won when my projection had him winning in a relative landslide. Yes. The averages were off in the trump direction either way, but excluding those polls made them WAY off. So you know what? I have the same mentality and motto as much of the election prediction community has this time: "throw it onto the pile" (or throw it into the average). Stop trying to read individual polls like they mean anything, just give me the average. Of course, if the republicans have so many polling agencies clogging up the averages it's gonna be off anyway. Still, not wanting to make the same error as I did in 2020, I'll just let the averages speak for themselves. I also don't think the unbiased polls are that much more optimistic than the nonpartisan ones anyway, and I ain't under the impression that the averages shift much. I normally use RCP but I did try looking at 538, race to the white house, and a few others, and while they still have harris leading, they are still seemingly recognizing the same shifts, as they had harris up more like 2-3 anyway, whereas I had her up <1. Now RCP has trump winning by <1 and the others have harris winning by <1. Honestly, we're splitting hairs over tenths of a percent here. I'm not gonna act like it's a huge real. I always used RCP going back to my original prediction in 2008, I feel like they do good work, I have no reason to switch at this time.
So...that's the presidency. What about the Senate?
Senate
So, adding the Osborn race has changed the calculus about the senate, giving the dems a better chance of hitting 50-50. That's now my most likely outcome at 52%, with the GOP having a 33% chance and the dems a 16% chance.
A big point of worry though is the fact that the rust belt has hollowed out in the past two weeks. Many of those races were "likely dem", now they're only "lean dem" showing that the race is tightening up as we go into the final month here. To be fair, the Texas and Florida races are softening up for dems, but I honestly don't expect them to flip. Florida is technically now "lean rep" but when you're at like the 1, 4, and 8 mark, you're kinda in that in between zone where i could theoretically classify you in either, and I probably would still call Florida as likely rep given its history. If it were going from 2->4, sure, lean. But going from 6->4 or whatever, yeah, likely. Either way I'm not too optimistic there, and it is still pretty obvious the rust belt states have a better chance of flipping at a 16-31% chance, whereas Florida is at a 16% chance and Texas is at 12%. Now I did do some simulations, but Ill save those for their own section. Which I'm going to transition to...now.
Simulations
Now, again, with simulations, take things with a grain of salt. My simulator treats each individual state as an independent trial, so no trends are involved at all. Meaning in the real world, if PA goes blue, I'd also expect MI and WI to. However, in this simulation, anything can happen. You can see a simulation with blue Texas and red Minnesota. It kind just brute forces possibilities and some of them are kind of nonsensical like the one I mentioned above. So, with all of that said, what results do I get?
Well, with the presidential, I get 51 Harris outcomes, 48 Trump outcomes, and 1 tie. It's definitely closing up on the simulator. For some reason the thing inherently favors Harris. Idk if this is due to the distribution of random numbers or if something about this map just leads to more harris outcomes, but yeah, all things being equal, harris tends to win the simulations slightly more often. And even with a 0.5% handicap, it's still slightly winning. Take it for what it is.
The senate is more interesting in a sense. The senate still heavily favors a republican map, with 68 outcomes being republican, 24 being ties, and 8 being democratic leaning. Again, if we follow the normal linear model of elections, I'd expect about 52% ties, 33% republican outcomes, and 16% democratic ones. Take it as you will. This part of my forecast is mostly experimental. I would say my official forecast are the charts above.
Conclusion
As such, there's a lot to get nervous about with this week's prediction. Perhaps it is early to be sounding alarm bells, but at the same time, we're seeing a decent amount of movement in the republican direction over the past 2 weeks or so, and given how we only got 3.5 weeks left and voting is already happening, there's reason to be concerned. This kind of movement was witnessed by myself in the week leading up to the 2016 election. And even though i still favored Clinton, Trump won, because polling is a lagging indicator. When polling shows a trend, it often does so behind the actual trend happening in the real world. So seeing movement toward trump at this late date concerns me, especially as early voting is a thing post covid now. We could be seeing the first signs that trump is gonna win. At the same time, it's also possible republican polling firms are just poll bombing the averages. Either way, I say throw them on the pile, and my prediction is based on said pile.
I will say that this isnt necessarily lethal for Harris. I know people mention in 2022 polls went red toward the end too and then the democrats overperformed by 3 points. But at the same time, republicans could ALSO overperform by 3 points. Honestly, a single point or two can make or break things, especially for the presidential. For the senate things are still a bit more robust, but a 3 point shift would more likely hurt democrats than republicans here.
Still, the above is what the median outcome is projected to be. Over 80% of outcomes will likely either feature a 319-219 Harris map or better, or a 226-312 Trump map or better. All I know is with this movement, I'd probably bet on the trump outcome over the harris one as things stand right now. Still, we got a few weeks, things can shift back the other way.
No comments:
Post a Comment