Election Insights Hub

Sample data · Polls as of 2026-10-28RealClearPolitics polling averages (via uspresidentialelectionnews.com), Oct 21, 2026 compilation; RCP component polls fielded ~Sept 14–Oct 11; fresh individual polls released Oct 21–25 added as dated entries; Oct 27–28 releases added as dated entries (12 polls, 8 races); six Oct 22–23 polls added Oct 28 per HK inclusion principle (see poll-inclusion-ledger.md); House generic ballot unchanged at 8.3 (RCP average value not re-verified Oct 28; last verified +8.3)Forecast run ID us-fc-2026-10-28-611ec4f1

US Midterm Forecast 2026

Election Day November 3, 2026 · 40,000 simulations

🗺️ County maps

3,100 counties · 2024 vote share · 2020→2024 swing · diploma divide by state

❓ Ask the Data

Ask in English — AI writes the SQL and answers

P(Democratic Senate majority)

55.9%

Expected 50.5 seats · majority = 51 · 80% range 47–54 seats

Democratic seat distribution

P(Democratic House majority)

97.1%

Expected 241.0 seats · majority = 218 · 80% range 224–259 seats

435 districts simulated individually (2024 presidential baseline + generic ballot)

Democratic seat distribution

Bars = share of simulations ending at each seat count. Shaded = 80% range, dashed = majority line.

Forecast history

Same poll inputs (2026-10-09 vintage); changes reflect modeling improvements, not new polls.

2026-10-09 · Correlated error + τ calibration (#3–#5) → 56.3% / 97.3%
2026-10-09 · Time-weighted poll aggregation (#6) → 58.4% / 97.3%
2026-10-09 · House incumbency adjustment (#7) → 58.4% / 96.6%
2026-10-09 · All 35 Senate seats modeled (#8) → 53.4% / 96.6%
2026-10-18 · Poll refresh Oct 18 (RCP compilation) → 55.1% / 97.1%
2026-10-19 · Poll refresh Oct 19 (CBS/YouGov + UH TX, ActiVote OH) → 54.8% / 97.1%
2026-10-21 · Poll refresh Oct 21 (RCP compilation; Marist OH, ActiVote+Quantus GA, co/efficient IA, NPA NH) → 52.9% / 97.1%
2026-10-23 · Poll refresh Oct 23 (Fox News OH/ME/TX, Emerson TX, TPOR TX, St Anselm NH, CJ/Harper NC, Rasmussen MN, Marist MI) → 52.9% / 97.1%
2026-10-26 · Poll refresh Oct 26 (Trafalgar ME) → 52.7% / 97.1%
2026-10-28 · Poll refresh Oct 28 (12 polls, 8 races: Echelon x5, NYT/Siena MN/FL, Suffolk OH, Glengariff MI, Univision TX, Marist ME, ActiVote FL) → 56.3% / 97.1%
2026-10-28 · Six Oct 22–23 polls added (HK inclusion principle) → 55.9% / 97.1%

Election-night tracker

Counting has not started. Until results arrive the board shows only probabilities and uncertainty ranges. On election night this panel shows called seats and percent reporting.

Runs 2026-11-02 to 2026-11-10 (the day before Election Day through one week after, ET). Refreshes every minute in that window.
Vote counts come only from official state election offices. A race is marked called only when The New York Times has called it (credited); vote share or margin alone never marks a race as decided.
Sources: The New York Times (race calls) · State election offices (vote counts)

Senate battlegrounds

Democratic polling margin (D−R, pts) · dashed line = 4-pt mark

Race details

North Carolina

96.3%

Roy Cooper vs Michael Whatley

RCP average: D+9.990% margin range: D+0.8 ~ D+19.0Leans D

Georgia

95.5%

Jon Ossoff (inc.) vs Mike Collins

RCP average: D+9.490% margin range: D+0.3 ~ D+18.5Leans D

New Hampshire

88.3%

Chris Pappas vs John E. Sununu

RCP average: D+6.790% margin range: R+2.5 ~ D+15.9Leans D

Minnesota

82.9%

Peggy Flanagan vs Michele Tafoya

RCP average: D+5.390% margin range: R+3.9 ~ D+14.5Toss Up

Michigan

70.8%

Abdul El-Sayed vs Mike Rogers

RCP average: D+3.090% margin range: R+6.0 ~ D+12.0Toss Up

Ohio

70.2%

Sherrod Brown vs Jon Husted (appt.)

RCP average: D+2.990% margin range: R+6.1 ~ D+11.9Toss Up

Texas

67.0%

James Talarico vs Ken Paxton

RCP average: D+2.490% margin range: R+6.6 ~ D+11.4Toss Up

Alaska

64.0%

Mary Peltola vs Dan Sullivan (inc.)

RCP average: D+2.290% margin range: R+7.9 ~ D+12.3Toss Up

Maine

59.4%

Troy Jackson vs Susan Collins (inc.)

RCP average: D+1.490% margin range: R+8.3 ~ D+11.1Toss Up

Iowa

47.1%

Josh Turek vs Ashley Hinson

RCP average: R+0.490% margin range: R+9.6 ~ D+8.8Toss Up

Kansas

45.8%

Adam Hamilton vs Roger Marshall (inc.)

RCP average: R+0.690% margin range: R+10.0 ~ D+8.8Toss Up

Florida

11.7%

Angie Nixon vs Ashley Moody (appt.)

RCP average: R+6.790% margin range: R+15.9 ~ D+2.5Leans R

State	Democrat	Republican	RCP average	90% margin range	Rating	P(Dem win)
North Carolina	Roy Cooper	Michael Whatley	D+9.9	D+0.8 ~ D+19.0	Leans D	96.3%
Georgia	Jon Ossoff (inc.)	Mike Collins	D+9.4	D+0.3 ~ D+18.5	Leans D	95.5%
New Hampshire	Chris Pappas	John E. Sununu	D+6.7	R+2.5 ~ D+15.9	Leans D	88.3%
Minnesota	Peggy Flanagan	Michele Tafoya	D+5.3	R+3.9 ~ D+14.5	Toss Up	82.9%
Michigan	Abdul El-Sayed	Mike Rogers	D+3.0	R+6.0 ~ D+12.0	Toss Up	70.8%
Ohio	Sherrod Brown	Jon Husted (appt.)	D+2.9	R+6.1 ~ D+11.9	Toss Up	70.2%
Texas	James Talarico	Ken Paxton	D+2.4	R+6.6 ~ D+11.4	Toss Up	67.0%
Alaska	Mary Peltola	Dan Sullivan (inc.)	D+2.2	R+7.9 ~ D+12.3	Toss Up	64.0%
Maine	Troy Jackson	Susan Collins (inc.)	D+1.4	R+8.3 ~ D+11.1	Toss Up	59.4%
Iowa	Josh Turek	Ashley Hinson	R+0.4	R+9.6 ~ D+8.8	Toss Up	47.1%
Kansas	Adam Hamilton	Roger Marshall (inc.)	R+0.6	R+10.0 ~ D+8.8	Toss Up	45.8%
Florida	Angie Nixon	Ashley Moody (appt.)	R+6.7	R+15.9 ~ D+2.5	Leans R	11.7%

Latest individual polls: New York Times/Siena

Likely voters · released 2026-10-03 · phone survey · Of 5 states: Democrat ahead in 3, Republican ahead in 1, tied in 1

Alaska

D+7.0

DMary Peltola50%

RDan Sullivan43%

n=504 · ±5.1 pts · Sep 24–Oct 1

RCP average in model D+2.2

Texas

D+6.0

DJames Talarico51%

RKen Paxton45%

n=615 · ±4.5 pts · Sep 21–30

RCP average in model D+2.4

Ohio

D+3.0

DSherrod Brown49%

RJon Husted46%

n=616 · ±4.5 pts · Sep 22–Oct 1

RCP average in model D+2.9

Kansas

Tied

DAdam Hamilton45%

RRoger Marshall45%

n=605 · ±4.6 pts · Sep 22–30

RCP average in model R+0.6

Iowa

R+1.0

DJosh Turek47%

RAshley Hinson48%

n=606 · ±4.6 pts · Sep 22–Oct 1

RCP average in model R+0.4

North Carolina

D+14.0

Different pollster: SurveyUSA/WRAL, 2026-10-13

DRoy Cooper54%

RMichael Whatley40%

n=643 · ±5.0 pts · Oct 7–11

RCP average in model D+9.9

North Carolina

D+7.9

Different pollster: Trafalgar Group, 2026-10-10

DRoy Cooper48.8%

RMichael Whatley40.9%

n=1,091 · ±2.9 pts · Oct 8–10

RCP average in model D+9.9

Texas

D+4.0

Different pollster: Siena College, 2026-10-13

DJames Talarico48%

RKen Paxton44%

n=617 · ±4.0 pts · Oct 13

RCP average in model D+2.4

Maine

D+2.0

Different pollster: Suffolk University, 2026-10-12

DTroy Jackson49%

RSusan Collins47%

n=500 · ±4.4 pts · Oct 8–11

RCP average in model D+1.4

Texas

D+2.0

Different pollster: CBS News/YouGov, 2026-10-18

DJames Talarico51%

RKen Paxton49%

n=1,492 · ±3.5 pts · Oct 8–16

RCP average in model D+2.4

Texas

D+2.0

Different pollster: UH Hobby School of Public Affairs, 2026-10-18

DJames Talarico48%

RKen Paxton46%

n=2,000 · ±2.2 pts · Oct 10–14

RCP average in model D+2.4

Ohio

R+0.4

Different pollster: ActiVote, 2026-10-18

DSherrod Brown49.8%

RJon Husted50.2%

n=400 · ±4.9 pts · Sep 26–Oct 18

RCP average in model D+2.9

Ohio

D+4.0

Different pollster: Marist College (NPR/PBS), 2026-10-20

DSherrod Brown48%

RJon Husted44%

n=1,176 · ±3.9 pts · Oct 15–18

RCP average in model D+2.9

Georgia

D+8.0

Different pollster: ActiVote, 2026-10-20

DJon Ossoff52%

RMike Collins44%

n=400 · ±4.9 pts · Oct 6–20

RCP average in model D+9.4

Georgia

D+6.9

Different pollster: Quantus Insights, 2026-10-20

DJon Ossoff51.8%

RMike Collins44.9%

n=755 · ±4.0 pts · Oct 15–17

RCP average in model D+9.4

Iowa

R+3.0

Different pollster: co/efficient, 2026-10-20

DJosh Turek45%

RAshley Hinson48%

n=863 · ±4.0 pts · Oct 15–19

RCP average in model R+0.4

New Hampshire

D+7.0

Different pollster: National Public Affairs, 2026-10-20

DChris Pappas50%

RJohn E. Sununu43%

n=600 · ±4.0 pts · Oct 8–12

RCP average in model D+6.7

Ohio

D+5.0

Different pollster: Fox News/Beacon Research/Shaw & Company, 2026-10-22

DSherrod Brown52%

RJon Husted47%

n=1,592 · ±2.5 pts · Oct 15–19

RCP average in model D+2.9

Maine

Tied

Different pollster: Fox News/Beacon Research/Shaw & Company, 2026-10-22

DTroy Jackson49%

RSusan Collins49%

n=1,514 · ±2.5 pts · Oct 16–20

RCP average in model D+1.4

Texas

Tied

Different pollster: Fox News/Beacon Research/Shaw & Company, 2026-10-22

DJames Talarico49%

RKen Paxton49%

n=1,398 · ±2.5 pts · Oct 15–19

RCP average in model D+2.4

Texas

D+2.0

Different pollster: Emerson College/The Hill, 2026-10-22

DJames Talarico48%

RKen Paxton46%

n=1,000 · ±3.0 pts · Oct 17–19

RCP average in model D+2.4

Texas

D+4.0

Different pollster: Texas Public Opinion Research/Slingshot Strategies, 2026-10-22

DJames Talarico51%

RKen Paxton47%

n=969 · ±3.9 pts · Oct 15–18

RCP average in model D+2.4

New Hampshire

D+7.0

Different pollster: Saint Anselm College Survey Center, 2026-10-22

DChris Pappas50%

RJohn E. Sununu43%

n=1,672 · ±2.4 pts · ended Oct 20

RCP average in model D+6.7

North Carolina

D+10.0

Different pollster: Carolina Journal/Harper Polling, 2026-10-22

DRoy Cooper49%

RMichael Whatley39%

n=600 · ±4.0 pts · ended Oct 20

RCP average in model D+9.9

Minnesota

D+3.0

Different pollster: Rasmussen Reports, 2026-10-23

DPeggy Flanagan48%

RMichele Tafoya45%

n=1,182 · ±3.0 pts · completed Oct 21

RCP average in model D+5.3

Michigan

D+2.0

Different pollster: Marist College, 2026-10-21

DAbdul El-Sayed50%

RMike Rogers48%

n=1,185 · ±4.3 pts · Oct 15–18

RCP average in model D+3.0

Individual polls are shown for reference and are not model inputs. The model reads RCP averages only, so these polls reach the probabilities once they enter the average. Most gaps sit inside the margin of error, so no single poll settles a race.

sri.siena.edu · usatoday.com · newsmax.com · kratko.news · newsweek.com · 270towin.com · focalnews.org · activote.net · ellisinsight.com · actionnetwork.com · political.org · newsmax.com · texaspublicopinionresearch.substack.com · politicalwire.com · politicalwire.com · pollingsource.com · 270towin.com

House

Generic ballot average

D+8.3

RealClearPolitics generic ballot average

Competitive-district polls (37)

49 : 47

Sept 8–11 · team-compiled district polls

435-district simulation

97.1%

80% range 224–259 seats

Each of the 435 districts uses a 2024 presidential baseline plus the generic ballot (D+8.1) uniform swing. District polls are not ingested yet, so σ=5.5 for all districts.

Methodology

Each simulation draws one national error s_nat ~ N(0, τ_nat²) and one error per Census region s_reg ~ N(0, τ_reg²), then scores every race as mᵢ + s_nat + s_reg + σᵢ·εᵢ. The national error moves all races together; the regional error moves races within a region together.

P(D wins i) = Φ(mᵢ / √(σᵢ² + τ_nat² + τ_reg²)); race outcome = mᵢ + s_nat + s_reg + σᵢ·εᵢ

τ_nat = 3.1 pts (national), τ_reg = 1.2 pts (regional) — calibrated on 2010–2022 Senate polling error
σᵢ = √(4.1² + 5.45²/n) by poll count, calibrated
40,000 simulations · seed 20261103 (same inputs give the same result)
The generic ballot is not mapped 1:1 to districts and is not used directly
Poll bias: backtest found a +1.4pt Democratic lean, left unadjusted due to cycle-to-cycle instability (a fixed correction would be judgmental).

Model assumptions

Race-level error σᵢ = √(4.24² + 5.45²/n) by poll count, calibrated (RCV races wider)
National error τ_nat = 3.1 pts, regional error τ_reg = 1.2 pts (decomposed from 2010–2022 Senate polling error; see scripts/calibrate-regions.py)
AK/ME Senate use RCV: +2.0 pts added to σ (first-choice polls ≠ final round)
Senate: all 35 seats modeled — 12 battlegrounds (polls) + 23 non-battlegrounds (fundamentals: state pres + incumbency, σ=6.5). 34 seats not up for election.
House: 435 districts modeled individually (2024 presidential baseline + generic-ballot uniform swing + 2.0-pt incumbency bump, σ=5.5). No seat cap.

Sources

Editorial charter

This board supports no candidate or party; its purpose is information only.
Every number carries a source. Polling averages come from public aggregates; probabilities are reproducible from the run ID.
Probabilities are not results. A 60% favorite loses four times in ten.
No future data is used. Before results, the board shows only probabilities and uncertainty ranges.
The model is in beta for its first cycle and will be scored against actual results after the election.