State of the Races

Eleven days out, the House projection is unchanged and the Senate is slightly less favorable to Democrats than a week ago, mainly because of South Dakota. In Georgia, Senator Ossoff remains a strong favorite and the governor's race is still a coin flip.

• Headline numbers. The House held at 257 Democratic seats (98.8% chance of a majority). The Senate slipped from 58 to 57 seats and from 94.1% to 92.7%. The environment estimate eased from 11.8 to 11.7 points after an ABC News, Washington Post, and Ipsos likely-voter generic ballot at Democrats +8 (a 9.5-point swing toward Democrats from 2024).
• Georgia. Two new polls kept Jackson narrowly ahead for governor: the University of Georgia for the Atlanta Journal-Constitution by 1 (a 1.2-point swing toward Democrats) and ActiVote by 3 (a 0.8-point swing toward Republicans). All eight pollsters have Jackson ahead, by 2.1 on average; Bottoms's chance is 51.4%. Ossoff leads by 8.4 (97.4%). Early voting reached 1,467,780 ballots through October 22.
• Senate. South Dakota moved from Likely Democratic to a toss-up (82% to 47%) on three polls with Senator Rounds ahead by 8 to 22, swings of 7 to 21 points toward the independent, Bengs, from President Trump's 29-point win. South Carolina rose from 63% to 70%, and Mississippi is now 50% after a poll for a Democratic group had Hyde-Smith ahead by 3 (a 21-point swing).
• Money. Senate outside and party spending on the tracker's board is about $1.06 billion Republican to $547 million Democratic. In Georgia, counting candidate cash, Democrats lead $33.8 million to $22.3 million, 1.5 to 1, down from 2 to 1 as Ossoff spends. The Republican Senate committee drew $20 million on a bank line in September. Money never moves a rating.
• House. Outside money now reaches 27 of the 34 long-shot seats the tracker watches. Two district polls for the main Democratic House super PAC: Lacore ahead by 2 in South Carolina's open 1st (a 15-point swing) and Representative Van Drew ahead by 8 in New Jersey's 2nd (a 5-point swing), 6 points behind the model. District polls never enter the model.
• Governors. Held at 29. Nebraska moved from Lean Republican to a toss-up (38% to 49%), and Oklahoma from Safe to Lean Republican (28%) on its first polls, the Republican ahead by 7 in a Trump +34 state. South Dakota moved to Likely Republican (10%); Democratic-held Kansas rose to 44% on a poll showing it even.

Changes are measured from last Friday evening, October 16. Chances are the model's probabilities. A 20% chance comes true one time in five, so treat none of these numbers as a certainty.

Georgia

Georgia votes on a U.S. Senate seat, an open governorship, every statewide office, all 14 U.S. House seats, and both chambers of the General Assembly. The two top races continue to go in different directions. Senator Jon Ossoff leads Representative Mike Collins by 8.4 points on average, a 10.6-point swing toward Democrats from President Trump's 2.2-point win in Georgia in 2024. His money advantage is narrowing as he spends: in mid-October he had $15.1 million in his campaign account against $1.1 million for Collins. Counting outside and party money as well, Democrats lead about 1.5 to 1.

The governor's race is the closest major race in the state. All eight pollsters in the average have Rick Jackson ahead of Keisha Lance Bottoms, by 1 to 3 points, for an average of Jackson +2.1. That is essentially no swing from 2024, while Ossoff runs about 10 points ahead of the same baseline. The newest, from the University of Georgia for the Atlanta Journal-Constitution, has Jackson ahead by 1, with 8% undecided. The model gives Bottoms 51.4% because it blends the polls with a national environment that favors Democrats. One firm's polls were removed from every average this week after it said most of its releases were not genuine polls.

Down the ballot, Democrats need 10 more seats to control the state House and 6 more to control the state Senate. The model rates the House Lean Democratic, slipping just under 80% this week, and the Senate a longer shot. The first public poll of the lieutenant governor's race has Dolezal ahead of McLaurin by 0.9, a 1.3-point swing toward Democrats. Last November, Democrats won both Public Service Commission races by about 26 points, but only about 1.56 million Georgians voted; a midterm electorate will be two to three times larger. Early voting passed 1.4 million ballots through its tenth day.

Georgia's U.S. House seats

Democrats hold 5 of Georgia's 14 U.S. House seats. Georgia did not redraw its map this cycle, so the 2024 results below are on current lines. The model's median is 6 Democratic seats (80% range 5 to 8), with a 68% chance that Democrats gain at least one seat and a 35% chance they gain two or more. All five Democratic seats are Safe, including the open 13th. No single Republican seat is favored to flip, but five of them sit between 13% and 38%, and the same national swing moves all of them together, so in most simulations at least one falls.

The best Democratic opportunities are two Lean Republican seats: the 12th, held by Rick Allen (Trump +14), at 38%, and the open 1st on the coast (Trump +16), at 37%. Next come the open 10th (Trump +21, 20%), the open 11th (Trump +23, 15%), and Rich McCormick's 7th (Trump +22, 13%). In last November's low-turnout Public Service Commission races, the Democrat carried the 1st, 10th, and 12th and came within 1.5 points in the 7th and 11th. Three Democratic candidates have released their own polls showing close races in deep-red seats: Chris Harden ahead by 2 in the open 11th, Maura Keller behind Brian Jack by 5 in the 3rd, and Kozycki ahead of McCormick by 2 in the 7th, swings of 24 to 25 points from 2024. Polls paid for by a candidate are discounted 3 to 4 points, and the tracker records district polls but never enters them in the model.

The Senate

Democrats hold 47 seats and need 51 for control; a 50-50 Senate stays Republican because the vice president breaks ties. The model's median is 57 Democratic seats, with a range of 51 to 62. It rates nine Republican-held seats Lean Democratic or better, from North Carolina (99%) through Ohio, Texas, Alaska, Maine, Kansas, Iowa, and Nebraska to South Carolina (70%). South Dakota, Idaho, and Mississippi are toss-ups at 46% to 50%. Every Democratic-held seat is rated Likely or Safe.

Texas again drew the most polling. The Democrat, Talarico, leads the average by 2.3 points, a 16-point swing from 2024, and the race is Likely Democratic at 85%. Republicans lead outside and party money about two to one, but Democratic candidates hold more cash in most of the close races, and candidates buy television at the lowest rates.

The model blends each race's polls with the national environment, which is why South Carolina rates Lean Democratic with Republicans ahead by 0.9 points. Osborn, Bodnar, and Achilles are independents, counted with Democrats. Week: change since October 16.

The House

Democrats need 218 seats. The model's median is 257, with a range of 233 to 292, and a 98.8% chance of a majority. The projection rests on a national environment of about 12 points in Democrats' favor. That figure blends three things: the likely-voter generic ballot average (Democrats +8.2), special elections this cycle (+14), and primary turnout (+13).

District polls often run behind the model, and this week was no exception in several seats. A Republican-side poll in California's open 48th has the Democrat, Wilpert, ahead by 3, no swing from Vice President Harris's 3-point win and well below the model's median of about 16. Others run closer: the Democratic super PAC poll in South Carolina's 1st sits about 2 points ahead of the model. District polls are a check, never an input.

Week = change in the chance since October 16, in points. The model's wider set of competitive seats (20% to 80%) is listed with every seat drawing $1 million or more in outside money.

Governors nationwide

Thirty-six states elect governors this year. The model projects 29 Democratic governors after the election, within a range of 25 to 32, up from 24 today. It rates four Republican-held governorships Lean Democratic or better: Alaska (89%), Iowa (88%), Ohio (77%), and Nevada (62%), where the polls are even. Vermont, Georgia, Texas, and Nebraska are toss-ups, as is Democratic-held Kansas, where a new poll has the race even and the Republican leads the average by 3.6 points.

Two races cut against their states' usual lean. Vermont's Republican governor, Phil Scott, leads by about 2 points in one of the most Democratic states in the country. And in Pennsylvania, Governor Josh Shapiro leads by about 23 points in a state President Trump carried.

Governor chances apply the tracker's Georgia governor model to every state's polling average: the polls, plus half of the national environment's lean, with wide uncertainty. The races with no tracked polling (Colorado, Hawaii, and Wyoming) use the 2024 presidential margin in place of polls. Safe races are left out of the table.

Early voting

Early-vote counts measure volume, not choice, and pace often reflects changes in voting rules or calendars. The tracker compares a state only against the same day in a comparable earlier election, after checking the rules match. Early-vote figures never move a rating or a chance.

Two states clear that check: North Carolina, matched day for day, at 1.37 times its 2022 pace, and Virginia at 1.10 times 2025 (a 2023 mail-ballot change rules out 2022). Georgia has cast 1,467,780 ballots in ten days, but its only same-day comparison is 2024, a presidential year. Texas is reporting with its checks still open; South Carolina has only a 2024 comparison, and New Hampshire is not in the data set.

Pace = 2026 ballots divided by ballots cast by the same day before the election in the comparison year; 1.00 is the same pace. Georgia, North Carolina, and Virginia through October 22; Pennsylvania through October 21; others October 22 or 23. Data: ElectIndex; Virginia from VPAP (Virginia Department of Elections).

AL, MO, NH, NY. Most have not begun early voting or have not yet posted data; reasons are on the tracker.

Party registration as each state publishes it (ElectIndex). Registration is not a vote: registered Democrats and Republicans can cross over, and voters with no party split both ways; early voters also differ from Election Day voters. These shares describe who has voted so far, not how a state will vote. States without party registration, or whose files omit the split, are left out rather than estimated.

Georgia: who is voting early

Through ten days of early voting, Black voters make up 33.4% of Georgia's early voters on the registration file, against 30.4% on the same day in 2022, the last midterm, and white voters 53.4% against 57.5%. The gap has narrowed on recent weekdays: on Thursday, October 22, the Black share of that day's voters was 28.9% against 25.9% on the matching weekday of 2022. The one Georgia poll that publishes its sample, YouGov's, assumes 29% Black on a self-identified basis, roughly a 2022- or 2024-like electorate on file terms. In 2022, Election Day voters were more heavily white than early voters, so Election Day voting can narrow the gap. The rows below are conditional: what the model gives if part or all of the early difference survives to the final count.

2026: Georgia Votes, which counts early voters by the race on their registration (Georgia Votes, last updated 2026-10-23); the Georgia Secretary of State's data hub shows the same pattern. 2022: the same measure, by weekday (2022 voting opened a day earlier, on a Monday, which is included in its running share). 2024 is not used: a presidential year draws a different electorate. Who has voted is not how they voted. Extra margin is rough arithmetic holding vote choice fixed, from the approximate pattern of past Georgia exit polls. The rows do not change the published chances.

What could change the picture

The headline rests on one estimate of the national environment. The scenarios below change that one assumption and hold everything else constant. Each is run through the same model.

House, Senate, and Georgia legislature cells: median Democratic seats, then the chance of Democratic control. Ossoff and Bottoms: chance of winning, with the polling margin for Bottoms.

Even in the polls-only scenario, Democrats remain favored for both chambers of Congress, though the Senate falls to about three in four. The race that moves most across the scenarios is Georgia's governor, from about 44% to 51% for Bottoms, followed by the Georgia state Senate, which ranges from a long shot to close to even.

• The model simulates the election 20,000 times. Its inputs are 2024 presidential results in every district and state, incumbency, and an estimate of the national environment. It also allows for a wave larger than any in the historical record.
• Every rating in this memo comes from the model's chance of a Democratic win: Safe 95% or more, Likely 80% to 95%, Lean 60% to 80%, and Toss-up 40% to 60%, with the same bands on the Republican side. A seat is marked as changing party when the model rates it Lean or stronger toward the party that does not hold it.
• Polling averages use likely-voter results, including voters leaning toward a candidate, where pollsters publish them. Each pollster and sponsor counts once, at its newest poll. Polls paid for by a campaign or party are discounted 3 to 4 points. Polls of individual House districts are recorded but never enter the model.
• Spending is tracked from Federal Election Commission filings, but money never moves a rating.
• "Swing from 2024" compares a race's polling margin with President Trump's 2024 margin in the same place.
• This is a personal analysis for friends and family, not a forecast from any news organization, campaign, or party.