Prediction Pro @predictionpro

Smart money is fading El-Sayed in Michigan 👀 • Our model: El-Sayed (D) 67.2% • Polymarket: El-Sayed 70.5% • Top politics wallets: $77K on Rogers (5 wallets) vs. $22K on El-Sayed (3) The crowd says Dem hold. The sharps disagree.

GOP is gaining some ground in Georgia's early vote, but Ossoff is still sitting at 95% on Kalshi. At this point I think the more useful prediction market question is his margin, not whether he wins. Kalshi is pricing roughly a 50/50 shot at a 10+ point victory. That's a pretty big margin in a state Trump won in 2024.

Ryan Bonifay @RBonifay

10 hours ago

Early Vote in-person is improving for the GOP in Georgia with the modeled partisanship margin shrinking 1.6% from yesterday. The current margin sits at D+13.8%. That moves to D+14.1% with mail returns included #gapol

7 9 64 13K 9

The Bush 2006 comparison is starting to look pretty relevant. Unpopular war, rising energy costs, and voters losing confidence in the economy. Trump's now at -24 net approval with Dems polling around D+10 nationally. Our PredictionPro model still has Dem Senate control at 66.1%, which says a lot about how hard the Senate map is even in an environment like this.

Republicans warming up and they're still 2% behind their 2022 pace while Dems are up 52% and unaffiliated voters are up 62%. That's a pretty big hole to climb out of. Our PredictionPro model has Cooper at 85.4% in the Senate race, and I think the bigger question now is who those unaffiliated voters are actually supporting.

Paxton's closed the gap from Talarico +2 to a tie, but Kalshi still has Talarico around 64%. That's what stands out to me. Traders are giving a Democrat roughly a 2 in 3 shot in a state Trump won by almost 14 points. Feels like the market is pricing in a lot more than just the national Democratic wave. Paxton's crossover problem is a pretty big part of this.

El-Sayed at 69% hasn't really moved despite how close the polling is. Marist has him up just 2, and the RCP average is also D+2. Feels like traders are pricing in the national Democratic environment more than the actual state polling. Pretty big difference between a 2 point lead and a market that's barely moved in two weeks.

Minnesota is a lot closer in the polls than you'd expect with Trump at 36% approval nationally. Flanagan's only up about 3 points, but Kalshi still has her at 88%. That's a pretty big gap between how competitive the polling looks and how confident traders are. Feels like the market is putting a lot of weight on Minnesota's history of voting blue.

Dems went from a nearly 20 point registration advantage on day one to just 3 points today. Republicans are definitely catching up, but this is where the prediction markets get tricky. Iowa is already being priced as a competitive Senate race, and early turnout is still all over the place. Wouldn't be surprised if traders are overreacting to both the early Dem surge and the GOP rebound.

The Kansas and Nebraska adjustments are pretty massive. Basically saying the polls could be picking up a real Democratic wave while still overstating how far it reaches into red states. I do wonder how much weight we should keep giving historical polling errors though. The national environment is pretty different from 2020 or 2024, and those misses aren't guaranteed to repeat.

28% of non-MAGA Republicans voting for Talarico is a pretty big number in Texas. And this is why Paxton's problems with his own party matter so much. Democrats don't need Texas to suddenly become a blue state, they just need enough Republicans to cross over or stay home. Especially with independents already moving toward Talarico.

Kalshi at 65% for a Dem Senate and our PredictionPro model at 66.1%. Basically identical on control, but the state margins are where it gets fun. Texas, Maine, Ohio and Michigan are all sitting around D+2 in Kalshi's forecast. A pretty small polling miss in either direction could completely change the Senate math.

Kalshi Politics @KalshiPolitics

10 hours ago

NEW SENATE MIDTERMS HUB FEATURE: Senate forecasted margin: 🟥 Nebraska: R+4.4% 🟥 Iowa: R+1.1% 🟦 Texas: D+2.0% 🟦 Maine: D+2.0% 🟦 Ohio: D+2.2% 🟦 Michigan: D+2.3%

19 19 165 61K 22

The lack of herding is probably a bigger deal than people realize. We're seeing everything from D+5 to D+12 and the overall trend is still pretty clear. Our PredictionPro model has the national House vote closer to D+6.1, so it's accounting for some polling error already. The question is whether 2026 ends up being another year where polls underestimate the size of the backlash.

Kansas at D+33.8 compared to 2024 is pretty crazy. Iowa at D+17.9 too. And it's not just Democrats showing up more, Republican registration share is dropping across almost every state on this chart. Still doesn't tell us how these people are voting, but definitely makes the turnout picture look a lot different than 2024.

Tom Bonier @tbonier

5 hours ago

Looking at the net partisan margin shift of votes cast so far to the same point in 2024 in party registration states, Dems are outpacing '24 in every single state. And Kansas looks... very good (yes, that reads +33.8 Dem).

17 113 523 28K 53

This is probably more interesting than the overall early vote numbers tbh. Low propensity Democrats are showing up at higher rates than Republicans in almost every NC county, not just the blue ones. Our PredictionPro model already has Cooper at 85.4%, but this is the kind of turnout data worth watching to see if the national polling shift is actually translating into votes.

250+ House seats would be insane, but D+11 isn't looking like much of an outlier anymore. Silver's average is already around D+9.2 while our PredictionPro model is closer to D+6.1. The interesting question is whether the model is being too conservative about the national environment or whether these polls are overstating the shift.

Michigan is kinda interesting because Benson is running 4 points ahead of El-Sayed in the same Marist poll. And both are doing about the same with independents. Makes you wonder how much of that gap is candidate-specific vs voters just being more comfortable with Benson. Could end up being a pretty important distinction on election day.

This is a pretty interesting adjustment because it cuts both ways. Dems might be getting underestimated in states like NC and Maine, but the same methodology actually helps Republicans in Iowa, Kansas and Nebraska. Really shows why you can't just take a D+10 national environment and apply it evenly across every Senate race.

The D+19 among people who didn't vote in 2024 is probably the biggest number here. The usual early vote Democratic advantage is one thing, but seeing lower propensity voters show up with that kind of partisan lean is a lot more interesting. If that group keeps turning out, it could change the math in Texas quite a bit.

Florida is a good example of why looking at just in-person early voting can be misleading. Republicans have a pretty solid lead there, but Democrats are actually ahead in mail ballots. Put the two together and it's a much different picture. The real question is how this compares to 2022 and 2024, not just who's leading on day 3.

Kansas might actually be the most interesting race on the map right now. On Point has it basically tied, and their early vote adjustment is pushing the Senate forecast above 53 Dem seats. Our PredictionPro model is at 66.1% for Dem Senate control, so we're a little more cautious. Pretty crazy that we're even talking about Kansas, Nebraska, Iowa and Texas all being competitive at the same time

‘On Point Politics’ - Lester Tellez @onpointpolitics

9 hours ago

We ran today's ballots through our forecast, and KANSAS flips. 🔵Kansas - Hamilton +1.4, was Marshall +1.8 🔵Nebraska - Osborn +2.9 🔵Texas - Talarico +2.1 🔵Iowa - Turek +0.6 Senate: 52.9 to 53.2 Dem seats with the early vote onpointpolitics.us

13 24 156 7K 12