🔮 Electile Dysfunction
How One “Dissident Right” Election Analyst Is Calling the Race
With an audience of over 300,000 across Substack, Telegram, Truth Social, MAGA-aligned podcasts, and elsewhere, you might call Captain Seth Keshel the Nate Silver of the right.
Captain K knows all the numbers. Unlike polling-focused analysts, Keshel looks primarily at demographics, voter registration, and other “open source” data. In one breath he is citing voter rolls in Michigan and party affiliation data in Erie County, PA in the next. He tracks the schedules of universities in hurricane-hit areas of North Carolina to estimate how many students will be on campus at voting time.
The former Army captain has always had an eye for data. “I was able to turn over the back of a baseball card and recognize numerical trends with ease,“ he says. Keshel got his start in forecasting by running a moneyball-style operation for his college baseball team and went on to serve in Afghanistan as an intel officer “connecting the dots” on the insurgency.
Keshel rose in MAGA media circles after the 2020 election, where his writing was cited by Trump in his efforts to overturn the results. He is a frequent guest on Steve Bannon’s War Room podcast and other media on what Keshel calls the “dissident right.”
Whether you agree with him or not, Keshel’s takes have a way of finding their way into the hands of the big guy himself and his views are certainly represented in the Polymarket order book, for example those predicting a Trump blowout win (🔮11% odds).
The following interview with Keshel is based on multiple conversations and edited for length. All answers are his own.
The Oracle: What’s the state of the race today?
If this were 2016, and nobody had ever heard of the 2020 election, I would bet everything I own and everything my children own, that Trump will win because there's literally nothing positive in the data for Harris at all, except for some polls. Look at the polling of Pennsylvania. Trump actually leads the RCP average in Pennsylvania now for the first time. But if you take out the Bloomberg plus five for Harris in the last two weeks, it's Trump plus two, where Morning Consult is the pollster that's probably getting paid to drop the polls that skew the average right now. And I don't like to sound conspiratorial in everything I say, but I would bet everything I own on a Trump win based on the measurements that I track, and I'll show them to you.
The Oracle: Can you explain the methodology that goes into your calls?
My models rely on traditional indicators like voter registration by party. I consider population. I look at the county trends, I have four different county trends, which dictate the behavior of a county, and then, of course, the potential for ballot stuffing. And you find that in places like Nevada, California, and Washington. So I'm an all source analyst. I use mostly open source data, and I understand the trends at the local, state, and county levels.
The Oracle: My Twitter feed is telling me the Amish are going to tip Pennsylvania for Trump. Is that right?
The Amish like Trump for sure, but there's only 80,000 of them in Pennsylvania. What I look at in Pennsylvania is the last 20 years of voter data and how it compares with the final margin that aligns well to the right of the Democratic registration advantage. Take a look at this chart:
As a state begins to realign, the severity of the difference of margin minus registration lead decreases. I see it in other areas, including strong Republican ones. But you'd have to go all the way back to 2000 to find a differential in which Harris would still win, and it would be less than 50k votes. Trump on average plays 783k to the right of the registration number in his two races. The entire six cycle average is 621k votes to the right - and it is always to the right.
Every 70k votes or so is about 1%. So, any polls showing Harris by 4-5% are alleging a 280-350k margin of victory, worse than Romney-Obama before the working class realignment began.
Trump at his 605k differential from 2020 would win Pennsylvania by 280k votes or 4%. Trump at the Trump average, which is probably too much on the high end for now, would win by 6.5%, or about 457k votes. A 2% win is around 140 margin, which is very plausible and very conservative.
Harris is easily the underdog given these dynamics, and the fact that mail-in voting requests right now are way down in volume, and substantially less Democratic than in 2020, when it was a Biden squeaker. Plus, she's not an Irish Catholic from Scranton and she has video haunting her in which she wants to ban fracking.
The Oracle: How do you know that party registration data is a good way to predict the result? Do other analysts do this?
All the academians and professors don't like my analysis. But I try to let the data speak for itself. I put something out about North Carolina the other day. I had no opinion associated with it other than what I think it says. It is what it is. It's just data.
One of the big critiques of my work is that, well, just because somebody's registered a Republican doesn't mean they have to vote that way, or vice versa. But when you line these numbers up, they're very consistent.
But here we are looking at polls where just four years ago, Wisconsin was six points off on the average. And [the pollsters] continue to roll those out as if they're credible. Oh, they say they fixed their methodology. Like, if you're trying to tell me that Nevada is going to go for Trump, but Arizona is going to stay blue. Well, there's a first time for everything, but it's never happened, and Arizona's been a state since 1912. I would say that it's most likely that Arizona is red and Nevada is blue, or they're both red, or they're both blue, not Nevada's red, with Arizona blue. Those are the kind of things that I pick out.
The Oracle: How does the hurricane that hit North Carolina impact the race?
If we wake up after the election and Harris won, it’s most likely because she managed to get her hands on North Carolina. A lot of people are talking about the possible impact of the hurricane by hurting Republican turnout. But you have to remember that there are only about four counties there who are on the balls of their asses. Of those, Buncombe County is the big Democratic county, so I think that’s going to balance it out and the Republicans are going to bend heaven and earth to get their people to turn out. Also Charlotte and Raleigh are not going to see the same margin gains that they had in 2020. You have to remember that in North Carolina, 97 out of 100 counties have had a Republican shift since 2020.
The Oracle: Now for the fun part. When you look at the election odds, do any scenarios stand out as off?
Michigan and Pennsylvania. My gut is that Trump has both states easily (🔮56% - 45% Trump in PA, 51% - 49% Trump in MI). I do not think that Josh Shapiro or Gretchen Whitmer are going to go balls to the wall to try to win their states for Harris, because I think that they want to run in 2028 and I think they want to run as the middle America working class Democrat. And they could say ‘hey, Kamala is from California and lost our state.’
New Hampshire. I think Trump is at least 1 in 3 to take New Hampshire (🔮88% Harris in NH). If Trump sticks to his average performance off the voter registration index in two cycles in each of the counties, particularly Rockingham and Hillsborough, he has an edge of just under 5,000 votes there, which is subject to change with any updates or exploitation of election loopholes there. I don't think either candidate has a ceiling in New Hampshire of much more than 10,000 votes, and if Trump manages to take a large victory, four states come into view - Virginia, New Hampshire, New Mexico, and Minnesota, and likely in that order. The other six states I group into the decisive states - Arizona, Nevada, Georgia, Michigan, Wisconsin, and Pennsylvania - would have already gone for Trump, along with North Carolina by a lot.
Trump Blowout Victory. Polymarket has odds for a Trump blowout victory (100+ electoral vote margin) at 11% and I think it’s around 25%. This is defined as something like Trump at 319+ electoral votes, which would mean he takes North Carolina and the six decisive states I just mentioned (312) plus Minnesota (322) or Virginia (325), or a combination of New Hampshire and New Mexico (321). Of all these, Minnesota is the least likely to go for Trump thanks to the actions of the Democrat trifecta legislature and governor, plus the "fortification" of the Twin Cities metro, which dominates the vote.
Florida Margin of Victory. There is a market on whether Trump will win Florida by 8+ points (🔮38% odds). I’d buy that with Trump projected to be 9-12 points up. Based on my registration modeling I’d give him a 75% chance to exceed 8 points. You have all 67 counties in the state with a rightward shift, and Trump doing much better with Latino voters. I would love to see a market on if Florida will vote to the right of Texas, I think it will. It is also a bullish sign in favor of a Trump landslide in Florida that Milton could have been much worse and likely will not depress pro-Trump turnout to great extent.
The Oracle: If you get a call to advise the Trump campaign today, what do you think they should do?
I would tell them to make sure that the get out the vote effort for the hurricane impacted counties is good to go. Do a love tap on North Carolina one or two more times. And then I would be all over Pennsylvania, Georgia, Michigan, and Wisconsin in the final weeks.
Disclaimer
Nothing in The Oracle is financial, investment, legal or any other type of professional advice. Anything provided in any newsletter is for informational purposes only and is not meant to be an endorsement of any type of activity or any particular market or product. Terms of Service on polymarket.com prohibit US persons and persons from certain other jurisdictions from using Polymarket to trade, although data and information is viewable globally.
Captain Keshel knows his numbers - bookmark this for November 6. Is there a way to bet on which pollster is most accurate? I'd take him over Silver.
Uhh, does nobody else see the oxymoron of a partisan statistician?