KS · Senate 2026 · Updated Aug 27, 2026, 2:03 a.m. ET
Kansas
Rated Likely R
Roger Marshall wins this seat in 91 in 100 simulations. Call it clearly favored.
Roger MarshallRIncumbent91 in 100- Adam HamiltonD9 in 100
Over time
How the forecast has moved
The margin
How close it lands
Each simulation produces a final margin. The curve shows where those margins fall; the tinted areas on either side of even are the futures each party wins.
Under the hood
What's driving the number
The forecast blends a weighted poll average with a fundamentals model: the state's partisan lean, incumbency, the national environment, and fundraising. Positive points favor the Democratic side of the margin; negative favor the Republican side.
- Partisan leanR +15.8
- IncumbencyR +2.5
- National environmentD +5.1
- FundraisingR +0.3
- Fundamentals sayR +13.4
- The polls sayR +1.8
- Blended forecast marginR +11.9
The blend puts 13 percent of its weight on the polls and 87 percent on the fundamentals; polls take over as more arrive and election day nears.
The markets
What the betting markets say
Prediction markets put a price on this seat too: what people are paying for a contract that pays out if Roger Marshall wins, read as an implied probability. Shown for comparison only. This is not betting advice, and market prices never feed the model.
- Roger Marshall winsThis forecast91 in 100
- Roger Marshall winsKalshi82 in 100
- Roger Marshall winsPolymarket80 in 100
Market prices as of Aug 27, 2026, 2:03 a.m. ET
The evidence
The polls, as the model sees them
Every public poll of this race, with the weight the model gave it in the latest run. Weights reward recency, sample size, and pollster track record; the house adjustment corrects each pollster's habitual lean before averaging.
| Poll | Dates | Sample | D–R | Margin | House adj. | Weight |
|---|---|---|---|---|---|---|
| Public Policy Polling (D) | Aug 7–8 | 569 LV | 45–46 | R +1.0 | +0.4 | 89% |
| GBAO (D) | Jul 8–13 | 600 LV | 43–47 | R +4.0 | −0.3 | 11% |
| Tavern Research (D) | Jan 26–28 | 1,013 LV | 46–54 | R +8.0 | — | <1% |
How the weights, the blend, and the simulation work is spelled out on the methodology page.