T-minus 16 days
[ Show me the math ]
Every rule,
written down.
A forecast built from every public poll, the measured history of how wrong polls have been, and tens of thousands of simulated elections run each morning. This page is the machine with the cover off: the rules, the numbers they produce, and every poll behind them. As of October 18, 16 days to the election, forecast dated October 18.
▸ 01 // Polling average
One average per race. Each poll is weighted by recency (half life 10 days), sample size, its Silver Bulletin pollster rating, and whether a party or campaign paid for it: half weight, shaded 3 points against the sponsor in Senate races and 4.5 in House races. Repeat polls from one pollster share weight. Each pollster's lean is measured against the race averages and subtracted. The leans are listed at the bottom of this page.
▸ 02 // Fundamentals
What the race would look like with no polls at all. In the Senate: the state's 2024 presidential lean, today's generic ballot, an incumbency bonus of 3 points, blended with the seat's last result shifted for how the national mood has moved since. In the House: margin = national + 1.74 × PVI + 4.75 × incumbency, fitted on the 2022 and 2024 results, where national is the generic ballot plus the 1.7 point offset the fit showed. Where polls exist the two are blended by how much polling there is.
▸ 03 // Simulation
40,000 Senate runs and 20,000 House runs each morning. Every run draws one national polling error shared by every race (sd 3.6 points), then a race level error that never falls below 2.8 points however many polls there are. House districts in one state share a 2 point error on top. Both sizes come from replaying the average on 412 Senate races since 1998, the same races the track record below is scored on, so that score is a fit, not a test. The rules were locked before looking at any 2026 output, and every change since is logged.
▸ Governors // Same rules, third chamber
Thirty six states elect a governor this year, and every one of those races runs through the Senate rules with nothing changed. The same polling average: recency half life 10 days, sample size, the Silver Bulletin pollster rating, half weight and a 3 point shade for a party sponsor, repeat polls from one pollster sharing weight, each pollster's lean measured against the governor race averages and subtracted. The same fundamentals: the state's 2024 presidential lean, today's generic ballot, the 3 point incumbency bonus, blended with the seat's last governor result (2022 in most states, 2024 in New Hampshire and Vermont, which elect every two years) shifted for how the national mood has moved since. The same blend by poll count, and the same error sizes in the simulation: a shared national error of 3.6 points and a race floor of 2.8. A majority of governorships is 26 of 50.
The known limitation, written before any number was looked at
Governor results track a state's presidential lean less tightly than Senate results do. Vermont, Kentucky and Kansas have all elected governors from the party that loses the state for president, usually a popular incumbent who crosses party lines. The fundamentals will be wrong in places like that, and the last result term carries the crossover only when the same person is running, since it gets half the weight for an incumbent and a quarter for an open seat. These rules were set before the model produced a single 2026 governor number. No governor backtest exists yet; the constants come from the Senate backtest alone, and nothing here will be tuned after seeing the output.
What is deliberately left out: presidential approval. The front page shows it beside the generic ballot because people ask, but it does not enter any race. The generic ballot already carries most of what approval would say about the national mood, and approval has never been through the backtest, so adding it seventeen days out would be tuning, not modelling. Candidate favorability is not in the public poll feed at all.
[ Senate // The numbers ]
The Senate
Republicans hold 53 seats. Democrats need a net gain of four, and the Vice President breaks ties for the other side.
▸ Chance Democrats control the Senate
66%
65% if the independents in Nebraska, Montana, South Dakota and Idaho are not counted with Democrats.
▸ Expected Democratic caucus
51.8
Seats. 80% of runs land between 48 and 56. Republicans hold 53 today.
▸ National environment
D +8.2
Generic ballot, 62 polls in the last 45 days.
▸ How many seats // Share of simulations ending with each Democratic caucus size. 51 is control.
▸ The board // Closest races first. Margin is Democrat minus Republican in points. Open a race for its polls.
| Race | Model margin | D wins |
|---|---|---|
| open, R held | -0.2 | 49% |
| open, R held | +1.1 | 59% |
| R incumbent | +1.4 | 60% |
| R incumbent | -2.0 | 36% |
| special, R appointed | +2.7 | 68% |
| R incumbent | +3.3 | 72% |
| open, D held | +4.2 | 80% |
| R incumbent | -4.5 | 24% |
| R incumbent | -4.8 | 24% |
| special, R appointed | -5.4 | 17% |
| open, D held | +6.8 | 88% |
| R incumbent | -7.6 | 11% |
| D incumbent | +8.8 | 92% |
| open, R held | +9.3 | 95% |
| R incumbent | -10.4 | 7% |
| open, D held | +10.7 | 95% |
| open, R held | -11.3 | 4% |
| open, R held | -12.2 | 4% |
| open, R held | -15.0 | 1% |
| R incumbent | -15.6 | 1% |
| D incumbent | +15.7 | 99% |
| D incumbent | +17.8 | 99% |
| open, D held | +18.2 | >99% |
| open, R held | -18.6 | <1% |
| D incumbent | +19.1 | 99% |
| D incumbent | +20.0 | 99% |
| R incumbent | -22.7 | <1% |
| R incumbent | -23.7 | <1% |
| open, R held | -24.3 | <1% |
| D incumbent | +24.9 | >99% |
| D incumbent | +27.0 | >99% |
| D incumbent | +32.2 | >99% |
| D incumbent | +33.8 | >99% |
| R incumbent | -36.7 | <1% |
| open, R held | -37.3 | <1% |
[ Governors // The numbers ]
The governors
Thirty six states vote. 24 governors are Democrats and 26 Republicans today; 26 of 50 is a majority. The Senate rules, nothing changed.
▸ Chance Democrats hold 26 or more
80%
A majority of the 50 governorships. 6 Democratic and 8 Republican governors are not on the ballot this year and are counted as they stand.
▸ Expected Democratic governors
28.2
Of 50. 80% of runs land between 24 and 32. Democrats hold 24 today.
▸ Where the number comes from
32
Races polled. The other 4 run on fundamentals alone: presidential lean, incumbency, the last governor result, and a generic ballot of D +8.2.
▸ How many governors // Share of simulations ending with each number of Democratic governors. 26 is a majority.
▸ The board // Closest races first. Margin is Democrat minus Republican in points. Open a race for its polls.
| Race | Model margin | D wins |
|---|---|---|
| R incumbent | -1.6 | 40% |
| open, R held | +1.8 | 62% |
| open, R held | +2.1 | 63% |
| R incumbent | -2.2 | 36% |
| R incumbent | -2.3 | 32% |
| open, D held | -2.5 | 35% |
| open, R held | +3.8 | 74% |
| open, R held | -3.9 | 25% |
| R incumbent | +3.9 | 73% |
| open, D held | +6.7 | 85% |
| R incumbent | -8.2 | 9% |
| open, R held | -8.3 | 10% |
| open, R held | +8.7 | 91% |
| open, D held | +8.9 | 94% |
| D incumbent | +9.0 | 92% |
| open, D held | +10.9 | 95% |
| open, D held | +11.3 | 96% |
| D incumbent | +11.5 | 97% |
| R incumbent | -13.1 | 2% |
| open, D held | +13.5 | 99% |
| R incumbent | -14.8 | 1% |
| open, R held | -15.6 | 1% |
| open, R held | -16.9 | <1% |
| D incumbent | +18.2 | >99% |
| open, R held | -19.1 | 2% |
| D incumbent | +20.4 | >99% |
| D incumbent | +20.9 | >99% |
| D incumbent | +22.7 | >99% |
| open, D held | +23.1 | >99% |
| open, D held | +24.3 | >99% |
| R incumbent | -26.7 | <1% |
| open, D held | +30.5 | >99% |
| D incumbent | +30.6 | >99% |
| D incumbent | +35.1 | >99% |
| D incumbent | +36.6 | >99% |
| open, R held | -38.9 | <1% |
Same columns as the Senate board. The prior is the Senate fundamentals with the seat's last governor election in place of the 2020 Senate result. Alaska's polls carry its ranked choice field, so the fullest ballot rule reads the four way question there.
[ House // The numbers ]
The House
218 seats is a majority. Republicans hold 220 today.
▸ Chance Democrats win the House
91%
218 seats is a majority. Republicans hold 220 today.
▸ Expected Democratic seats
241
Seats. 80% of runs land between 218 and 267.
▸ Where the number comes from
106
Districts polled. The other 329 run on fundamentals: district lean, incumbency, and a generic ballot of D +8.2.
▸ How many seats // Share of simulations ending with each number of Democratic seats.
▸ The board // Every district inside 10 points, closest first. PVI is the district's lean. Open a district for its polls.
| District | Model margin | D wins |
|---|---|---|
| R incumbent Bryan Steil | 0.0 | 50% |
| open, R held | -0.1 | 49% |
| open, R held | -0.1 | 49% |
| open, new or vacant | -0.2 | 48% |
| R incumbent Derrick Van Orden | -0.2 | 49% |
| D incumbent Shomari Figures | -0.3 | 48% |
| open, new or vacant | +0.4 | 52% |
| open, R held | +0.8 | 55% |
| R incumbent Mariannette Miller-Meeks | +1.0 | 56% |
| R incumbent Rob Wittman | +1.0 | 56% |
| R incumbent Bill Huizenga | -1.1 | 43% |
| R incumbent Scott Perry | +1.2 | 58% |
| open, R held | -1.3 | 42% |
| D incumbent Darren Soto | -1.4 | 41% |
| open, R held | +1.5 | 59% |
| R incumbent Ann Wagner | -1.5 | 41% |
| R incumbent Rob Bresnahan | -1.5 | 41% |
| open, D held | +1.7 | 60% |
| R incumbent Zach Nunn | +2.1 | 63% |
| R incumbent Brad Finstad | -2.3 | 36% |
| R incumbent Juan Ciscomani | +2.4 | 65% |
| R incumbent Tom Barrett | +2.4 | 64% |
| R incumbent Chuck Edwards | -2.4 | 35% |
| D incumbent Marcy Kaptur | +2.4 | 65% |
| R incumbent Brian Fitzpatrick | +2.6 | 66% |
| R incumbent Monica De La Cruz | -2.6 | 34% |
| R incumbent Max Miller | -2.8 | 33% |
| open, R held | -3.0 | 31% |
| R incumbent Mike Carey | -3.2 | 31% |
| open, R held | -3.3 | 30% |
| R incumbent French Hill | -3.5 | 29% |
| R incumbent Aaron Bean | -3.5 | 32% |
| open, R held | -3.5 | 29% |
| R incumbent Jeff Crank | -3.8 | 28% |
| open, new or vacant | -3.8 | 27% |
| R incumbent David Valadao | +3.9 | 73% |
| R incumbent Nick LaLota | -3.9 | 27% |
| open, R held | -3.9 | 27% |
| open, R held | -4.0 | 29% |
| R incumbent Mike Flood | -4.0 | 26% |
| open, new or vacant | -4.0 | 29% |
| R incumbent Ryan Mackenzie | +4.1 | 74% |
| R incumbent Jen Kiggans | +4.3 | 75% |
| R incumbent Jeff Hurd | -4.4 | 24% |
| R incumbent Anna Paulina Luna | -4.4 | 24% |
| R incumbent Mike Turner | -4.7 | 23% |
| open, R held | -4.7 | 23% |
| R incumbent Mike Lawler | +4.9 | 78% |
| R incumbent Thomas Kean Jr. | +5.1 | 79% |
| R incumbent Gabe Evans | +5.2 | 76% |
| R incumbent María Elvira Salazar | -5.2 | 20% |
| R incumbent Andrew Garbarino | -5.3 | 24% |
| R incumbent John McGuire | -5.5 | 19% |
| R incumbent Michael Baumgartner | -5.8 | 18% |
| open, R held | +6.0 | 83% |
| R incumbent Greg Murphy | -6.1 | 17% |
| open, new or vacant | -6.1 | 17% |
| D incumbent Marie Gluesenkamp Perez | +6.1 | 83% |
| R incumbent Eli Crane | -6.4 | 15% |
| D incumbent Don Davis | +6.4 | 84% |
| open, D held | -6.4 | 15% |
| D incumbent Vicente Gonzalez | +6.8 | 86% |
| R incumbent Gus Bilirakis | -7.0 | 17% |
| R incumbent Brian Mast | -7.0 | 18% |
| R incumbent Mario Díaz-Balart | -7.0 | 17% |
| R incumbent Rick Allen | -7.0 | 18% |
| R incumbent Pete Stauber | -7.0 | 17% |
| R incumbent Joe Wilson | -7.0 | 17% |
| R incumbent Victoria Spartz | -7.3 | 13% |
| R incumbent Tim Moore | -7.3 | 12% |
| open, R held | -7.5 | 15% |
| open, R held | -7.5 | 16% |
| open, R held | -7.5 | 16% |
| open, R held | -7.5 | 16% |
| R incumbent Lauren Boebert | -7.7 | 11% |
| D incumbent Kathy Castor | +7.7 | 89% |
| R incumbent Jeff Van Drew | -7.8 | 11% |
| R incumbent Richard Hudson | -7.9 | 11% |
| open, R held | -7.9 | 10% |
| R incumbent David Rouzer | -8.2 | 10% |
| D incumbent Jared Moskowitz Redistricted from the 23rd district | +8.6 | 91% |
| R incumbent Abraham Hamadeh | -8.8 | 12% |
| R incumbent Scott Franklin | -8.8 | 12% |
| R incumbent Brad Knott | -8.8 | 12% |
| R incumbent Warren Davidson | -8.8 | 12% |
| R incumbent Beth Van Duyne | -8.8 | 12% |
| R incumbent Tony Wied | -8.8 | 12% |
| open, D held | +9.1 | 93% |
| open, R held | -9.2 | 11% |
| open, R held | -9.2 | 11% |
| open, R held | -9.2 | 11% |
| open, R held | -9.2 | 11% |
| open, R held | -9.2 | 11% |
| R incumbent Rich McCormick | -9.3 | 7% |
| open, R held | -9.5 | 6% |
| R incumbent Stephanie Bice | -9.7 | 6% |
| open, R held | -9.7 | 6% |
97 districts shown. The other 338 sit outside 10 points: 205 on the Democratic side, 133 on the Republican side (including 17 and 1 with no opponent).
[ Pollster leans // House effects ]
Who leans which way
How far each pollster has run from the race averages this cycle, in points. The lean is subtracted from every poll before it enters the average, and it is capped at 3 points either way.
▸ Senate polls
- The Bullfinch GroupD+2.4
- Marist UniversityD+1.9
- TIPP InsightsD+1.7
- Slingshot StrategiesD+1.2
- Trafalgar GroupD+0.9
- YouGovD+0.6
- ReconMR/Siena Research InstituteD+0.5
- Beacon Research/Shaw & Company ResearchD+0.3
- Wedgewood PollsD+0.3
- Abacus DataD+0.3
- Big Data PollD+0.2
- Tavern ResearchD+0.2
- Guidant Polling and StrategyD+0.2
- Fabrizio Ward/Impact ResearchD+0.1
- SoCal StrategiesD+0.1
- CNN/SSRSeven
- SurveyUSAeven
- Suffolk UniversityR+0.1
- Cygnal PoliticalR+0.3
- Mason-Dixon Polling & Research, Inc.R+0.4
- Barbara Jordan Public Policy Research and Survey CenterR+0.5
- Global Strategy GroupR+0.6
- St. AnselmR+0.6
- Harper PollingR+0.7
- Opinion DiagnosticsR+0.8
- Tulchin ResearchR+0.8
- Alaska Survey ResearchR+0.8
- InsiderAdvantageR+0.8
- Mitchell Research & CommunicationsR+1.0
- GBAOR+1.0
- Impact ResearchR+1.1
- The New York Times/Siena UniversityR+1.4
- High Point UniversityR+1.4
- Hart Research AssociatesR+1.4
- Change ResearchR+1.5
- Echelon InsightsR+1.5
- Data for ProgressR+1.6
- Stetson University Center for Public Opinion ResearchR+2.0
- Bowling Green State University Democracy and Public Policy Research Network/YouGovR+2.3
- GrayHouseR+2.3
- Emerson CollegeR+2.5
- Quantus InsightsR+2.9
- Glengariff Group, Inc.R+3.0
- Peak InsightsR+3.0
- Public Policy PollingR+3.0
- University of New HampshireR+3.0
- co/efficientR+3.0
▸ Governor polls
- TIPP InsightsD+3.0
- Siena UniversityD+2.7
- Change ResearchD+2.6
- Bravo GroupD+2.2
- Suffolk UniversityD+1.6
- Marist UniversityD+0.9
- CNN/SSRSD+0.9
- SurveyUSAD+0.8
- St. AnselmD+0.6
- University of California Berkeley Institute of Governmental StudiesD+0.5
- ReconMR/Siena Research InstituteD+0.4
- InsiderAdvantageD+0.2
- YouGovD+0.2
- Franklin & Marshall CollegeD+0.1
- Fabrizio Ward/Impact ResearchR+0.1
- Global Strategy GroupR+0.1
- Beacon Research/Shaw & Company ResearchR+0.2
- Kreate StrategiesR+0.2
- Echelon InsightsR+0.3
- DHM ResearchR+0.3
- Susquehanna Polling & Research, Inc.R+0.7
- Quinnipiac UniversityR+0.7
- University of New HampshireR+0.7
- The New York Times/Siena UniversityR+0.8
- Mason-Dixon Polling & Research, Inc.R+0.8
- GBAOR+0.9
- Big Data PollR+0.9
- Z to A ResearchR+1.0
- Hart Research AssociatesR+1.2
- EPIC-MRAR+1.2
- Cygnal PoliticalR+1.3
- Barbara Jordan Public Policy Research and Survey CenterR+1.3
- Mitchell Research & CommunicationsR+1.3
- Slingshot StrategiesR+1.5
- Quantus InsightsR+1.5
- Public Policy PollingR+1.7
- University of North FloridaR+1.7
- Wedgewood PollsR+1.8
- McLaughlin & AssociatesR+1.8
- Noble Predictive InsightsR+1.9
- Emerson CollegeR+1.9
- Bowling Green State University Democracy and Public Policy Research Network/YouGovR+2.2
- Stetson University Center for Public Opinion ResearchR+2.4
- Targoz Market ResearchR+2.4
- Impact ResearchR+2.7
- Glengariff Group, Inc.R+2.9
- co/efficientR+3.0
▸ House polls
- Data for ProgressD+3.0
- Tavern ResearchD+3.0
- Change ResearchD+1.3
- University of New HampshireD+0.9
- Ragnar Research PartnersD+0.8
- Alaska Survey Researcheven
- FM3 ResearchR+0.1
- Tulchin ResearchR+0.1
- SurveyUSAR+0.3
- Public Policy PollingR+0.6
- Normington, Petts & AssociatesR+0.8
- GBAOR+1.0
- Hart Research AssociatesR+1.1
- Expedition StrategiesR+1.1
- Impact ResearchR+1.2
- co/efficientR+1.3
- Global Strategy GroupR+1.4
- Lake Research PartnersR+1.9
- DCCC Targeting TeamR+1.9
- GQRR+2.2
- St. AnselmR+3.0
▸ Sources // Where the inputs come from
Polls from PollResults.org (New York Times data, CC BY 4.0). Pollster ratings and the 1998 to 2024 polling history from Silver Bulletin, free with attribution. The Senate race list and ratings from Wikipedia's 2026 Senate elections page (Cook, Inside Elections and Sabato). House districts, PVI, incumbents, nominees and the 2022 and 2024 results from Wikipedia's House elections pages. The governor race list, nominees and ratings from Wikipedia's 2026 gubernatorial elections page, the last governor results from its 2022 and 2024 pages, and the governors not on the ballot from its list of current governors. Past results from Wikipedia. The model runs every morning and this page shows whatever it produced today.