Senate control: Democrats 77 in 100, Republicans 22 in 100
Democrats were 78 in 100 on Oct 12: down 1 point in a week.
Our model simulates the Nov. 3 election 40,000 times using polls, each race’s partisan lean and the national mood. Updated Oct 18, 9:14 PM ET.
Democrats win the Senate in 77 of 100 simulations. Republicans win 22; 1 end with no majority.
7 likely to flip to D 7 · 2 more could
Democrats win the House in 98 of 100 simulations. Republicans win 2.
30 likely to flip to D 26 · to R 4 · 14 more could
Each dot is a seat, colored by how likely each side is to win it; lighter end dots in the Senate aren’t up this year. A yellow ring marks a seat likely to change parties. Select a dot to open its race. Odds: 77 and 98 in 100 for Democrats.
Democrats were 78 in 100 on Oct 12: down 1 point in a week.
Unchanged from a week ago (Democrats 97 in 100).
Marshall: 39 → 48 in 100. Chance of winning, from the simulations.
Poll ended Oct 12, RV, 1,403 respondents. Our adjusted figure: Democrats +7.
We simulate the election 40,000 times. “77 in 100” means that side won 77 of every 100 runs — the other side still wins sometimes.
Each race blends its polling average with the state’s partisan lean and the national mood. The closer to Election Day, the more polls count.
If polls miss in one state, they usually miss in similar ones too. That’s built in, so upsets come in bunches like they do in real life.
The closest Senate and governor contests right now.
Seats where the party favored to win is not the party that holds the seat today, listed in the order the model is surest. 44 are likely to change parties (50 in 100 or better); 18 more could (25–49 in 100). The count covers both directions.
“Holds the seat today” is the party of the current officeholder. In the ten states with new House maps (AL, CA, FL, LA, MO, NC, OH, TN, TX, UT) it is the party of the sitting member, whose district lines have changed.
In a typical simulation Democrats and allied independents hold 53 seats; in 80 of 100 they land between 48 and 56. They need 51, since Vice President Vance breaks 50–50 ties for Republicans.
Each dot is one outcome: how many seats Democrats and allied independents hold. Blue dots reach 51.
The race most likely to hand one party its majority, across simulations.
From the Democrats’ best chance to the Republicans’ best chance.
| Race | Rating | Non-Republican side | Republican side | Projected margin | Polls |
|---|---|---|---|---|---|
| Kansas | Toss-up Likely flip · R → D | Adam Hamilton 52 in 100 | Roger Marshall 48 in 100 | Hamilton +0.3 | 3 |
| Iowa | Toss-up Likely flip · R → D | Josh Turek 53 in 100 | Ashley Hinson 47 in 100 | Turek +0.3 | 14 |
| Nebraska | Lean R Could flip · R → I | Dan Osborn (I) 40 in 100 | Pete Ricketts 60 in 100 | Ricketts +1.1 | 2 |
| South Carolina | Lean R Could flip · R → D | Annie Andrews 38 in 100 | Darline Graham 62 in 100 | Graham +1.2 | 2 |
| Alaska | Lean D Likely flip · R → D | Mary Peltola 66 in 100 | Dan S. Sullivan 34 in 100 | Peltola +1.6 | 7 |
| Texas | Lean D Likely flip · R → D | James Talarico 69 in 100 | Ken Paxton 31 in 100 | Talarico +1.8 | 26 |
| Maine | Likely D Likely flip · R → D | Troy Jackson 81 in 100 | Susan Collins 19 in 100 | Jackson +3.3 | 14 |
| Ohio (special) | Likely D Likely flip · R → D | Sherrod Brown 82 in 100 | Jon Husted 18 in 100 | Brown +3.5 | 13 |
Democrats win a median of 240 seats (218 is a majority); in 80 of 100 simulations they win 226–255. District lines reflect the 2026 maps, including the ten states that redrew mid-decade.
Seats won by Democrats in each simulation. Blue dots reach 218.
36 states elect governors. Democrats win an average of 22.8 of them and Republicans 13.2.
| Race | Rating | Non-Republican side | Republican side | Projected margin | Polls |
|---|---|---|---|---|---|
| Georgia · runoff 40 in 100 | Toss-up Likely flip · R → D | Keisha Lance Bottoms 54 in 100 | Rick Jackson 46 in 100 | Bottoms +0.4 | 9 |
| Nevada | Lean D Likely flip · R → D | Aaron Ford 65 in 100 | Joe Lombardo 35 in 100 | Ford +1.7 | 2 |
| Nebraska | Lean R Could flip · R → D | Lynne Walz 32 in 100 | Jim Pillen 68 in 100 | Pillen +2.4 | 1 |
| Vermont | Lean D Likely flip · R → D | Amanda Janoo 70 in 100 | Phil Scott 30 in 100 | Janoo +3.2 | 3 |
| Texas | Lean R Could flip · R → D | Gina Hinojosa 29 in 100 | Greg Abbott 71 in 100 | Abbott +2.0 | 23 |
| Oregon | Likely D | Tina Kotek 76 in 100 | Christine Drazan 24 in 100 | Kotek +3.0 | 2 |
Chance of Democratic control (polls + fundamentals). Use the buttons to zoom in or out, like a stock chart. Points before Oct 18, 2026 are recomputed from the polls available on each date, since the site launched then.
Prediction markets are bets, not polls, and they are not an input to this forecast. They’re shown so you can see where traders disagree with the model.
Market prices normalized so both parties sum to 100. Fetched with the forecast.
Biggest movers over the past week, by change in win probability.
No race moved by 3 or more points since the last update. The log fills in as daily runs accumulate.
The generic-ballot average is D+8.2 and the president’s net approval is -24.6. Blending the two, the model expects a national environment of D+9.6, give or take about 2.4 points. See the trackers →