EFElection Forecast Lab
Model live · Aug 24
2026 MIDTERM ELECTIONS71 DAYS OUTFORECAST № 01

THE CURRENT READ

Democrats are favored for the House.
The Senate is still a coin flip.

01

Balance of power

Model probabilities with bounded market context

HOUSE CONTROL223.1 expected D seats
Democratic 68.1%31.9% Republican
Model 64.5% · market 87.1%
SENATE CONTROL50.4 expected D-aligned seats
Democratic 48.0%52.0% Republican
Model 48.0% · market 48.4%
TEXAS SENATEPolls D +1.8 · model D +0.4
Talarico 52.2%47.8% Paxton
Model-only 53.1% · market 49.3%
How to read this

Probability is not vote share. A 68% forecast means the same race would produce a Democratic win about 68 times in 100 comparable simulations.

Read the methodology →

RACE EXPLORER

Every seat,
one evidence trail.

507 races
RaceMatchupPollsModelMarketD win
DEl-Sayed
RRogers
D +0.217 pollsR +0.3Ridge + poll/fundamentals61%Weight 26%
52%
DJackson
RCollins
D +2.83 pollsR +0.7Ridge + poll/fundamentals69%Weight 27%
52%
DTalarico
RPaxton
D +1.816 pollsD +0.4Ridge + poll/fundamentals49%Weight 25%
52%
DMary Peltola
RDan S. Sullivan
D +1.85 pollsR +1.0Ridge + poll/fundamentals62%Weight 21%
47%
DTurek
RHinson
R +1.25 pollsR +2.2Ridge + poll/fundamentals41%Weight 26%
36%
DBrown
RHusted
D +4.411 pollsD +3.0Ridge + poll/fundamentals53%Weight 21%
67%
DDemocratic nominee TBD
RPete Ricketts
0 pollsR +5.0external rating priorNo matched market
30%
DAdam Hamilton
RRoger Marshall
R +1.01 pollsR +5.3Ridge + poll/fundamentals20%Weight 26%
24%
DFlanagan
RMichele Tafoya
D +5.02 pollsD +3.7Ridge + poll/fundamentals90%Weight 26%
80%
DShoffner
RCotton
R +7.32 pollsR +4.7Ridge + poll/fundamentals6%Weight 23%
16%
DOssoff
RCollins
D +8.06 pollsD +6.0Ridge + poll/fundamentals93%Weight 27%
88%
DRoy Cooper
RMichael Whatley
D +8.720 pollsD +7.6Ridge + poll/fundamentals91%Weight 27%
91%
DScott Colom
RCindy Hyde-Smith
0 pollsR +18.0external rating prior8%Weight 25%
4%
DCharles Booker
RAndy Barr
0 pollsR +18.0external rating prior4%Weight 20%
3%
DMarquita Bradshaw
RBill Hagerty
0 pollsR +18.0external rating prior4%Weight 24%
3%
DJuliana Stratton
RDon Tracy
0 pollsD +18.0external rating prior96%Weight 23%
97%
DMark Warner
RBert K Mizusawa
0 pollsD +18.0external rating prior97%Weight 22%
97%
DJohn Hickenlooper
RMark Baisley
0 pollsD +18.0external rating prior97%Weight 25%
97%
DChris Coons
RRepublican nominee TBD
0 pollsD +18.0external rating prior97%Weight 0%
97%
DDavid Jordan Roth
RJim Risch
0 pollsR +18.0external rating prior3%Weight 0%
3%
DJamie Davis
RJulia Letlow
0 pollsR +18.0external rating prior6%Weight 0%
3%
DBen Ray Luján
RLarry E Marker
0 pollsD +18.0external rating prior96%Weight 0%
97%
DAnnie Andrews
RRepublican nominee TBD
0 pollsR +18.0external rating prior11%Weight 0%
3%
DDemocratic nominee TBD
RMike Rounds
0 pollsR +18.0external rating prior2%Weight 0%
3%
DDemocratic nominee TBD
RHarriet Hageman
0 pollsR +18.0external rating prior2%Weight 0%
3%
DNixon
RMoody
R +9.04 pollsR +11.3Ridge + poll/fundamentals8%Weight 20%
3%
DRachel Lee Fetty Anderson
RShelley Moore Capito
0 pollsR +18.0external rating prior2%Weight 23%
3%
DCory Booker
RJustin Michael Murphy
0 pollsD +18.0external rating prior98%Weight 23%
97%
DJeff Merkley
RDavid Brock Smith
0 pollsD +18.0external rating prior98%Weight 23%
97%
DPappas
RBrown
D +11.36 pollsD +11.3Ridge + poll/fundamentals89%Weight 0%
98%
DWess
RMoore
R +15.01 pollsR +17.6Ridge + poll/fundamentals4%Weight 23%
2%
DBankhead
RAlme
R +21.45 pollsR +22.5Ridge + poll/fundamentals1%Weight 21%
1%
DMarkey
RDeaton
D +20.14 pollsD +23.0Ridge + poll/fundamentals98%Weight 0%
99%
DJim Priest
RKevin Hern
R +21.01 pollsR +24.7Ridge + poll/fundamentals3%Weight 0%
1%
DReed
RMcKay
D +17.22 pollsD +18.4Ridge + poll/fundamentals97%Weight 0%
99%

SELECTED RACE · SENATE

TX

52.2%Talarico win probability

SIMULATION

Where the chamber
could land.

CONTROL →
223Median D seats
19824890% simulation range
64.5%D control

METHODOLOGY

Built to be
wrong honestly.

A forecast should expose uncertainty, data gaps, and rejected ideas—not just its headline number.
01

Aggregate evidence

Polls decay with time and are weighted by sample size, likely-voter status, sponsor partisanship, pollster frequency, and historically estimated house effects.

02

Blend, don’t crown

Senate forecasts use a fixed 50/50 Ridge and transparent candidate-aware fundamentals blend. House and governor models are selected on independent races and close contests.

03

Gate every adjustment

A probability calibrator was tested and automatically rejected because it failed the 0.001 out-of-period Brier improvement threshold. Markets remain bounded secondary evidence.

04

Simulate shared shocks

50,000 latent-normal draws allow national and office-specific errors to move races together instead of pretending every seat is independent.

VALIDATION

Earlier cycles choose.
Later cycles judge.

84.7%Nested winner accuracy
0.111Nested Brier score
576Independent selected forecasts
3,097Held-out race × horizon forecasts
GOVERNOR

Candidate-aware fundamentals

149 independent races · 60-day Brier 0.045

HOUSE

National environment blend

429 independent races · 60-day Brier 0.188

SENATE

Ridge + poll/fundamentals

196 independent races · 60-day Brier 0.066

DATA PROVENANCE

697 current polls.
161 verified snapshots.

Every poll keeps its original link. Cross-source duplicates are clustered before aggregation, so the same survey cannot gain influence merely by appearing on several sites.

AUTHENTICITY AUDITpass with review flags
POLLING ORGANIZATIONS200
FEC CANDIDATE MATCHES822/847
REVIEW FLAGS1
pollingsource.com 60emersoncollegepolling.com 43d3nkl3psvxxpe9.cloudfront.net 33pro.morningconsult.com 32scholars.unh.edu 29drive.google.com 16static.foxnews.com 16www.politico.com 14quantusinsights.org 13www.270towin.com 13www.nytimes.com 11www.ipsos.com 11www.cygn.al 9napolitannews.org 8www.focaldata.com 8scholarworks.bgsu.edu 8