Senate Race Report: November 1, 2026

Question: Two days before Election Day, the forecaster that hedges longest has made the Democrats the favorites, the Kansas market has moved to the Democrat, and the president is flying to Iowa and Kansas this Sunday after cancelling both trips last week, so what is still undecided?

Answer: Seven seats, six of them Republican-held, and the order in which they fall. Cook Political Report published its closing Senate analysis on October 30, called the Democrats “slight favorites” to win the majority and put their expected gain at two to six seats.¹ Inside Elections had moved nine races in a single update on October 29, every one of them toward the Democrats.² Kalshi spent the week repricing the same map. Kansas went to Adam Hamilton at 51 percent on October 30, after Roger Marshall had traded at 82.8 percent on September 5.³ Texas went to James Talarico at 71 percent.⁴ Michigan went to Abdul El-Sayed at 77 percent.⁵ None of that is a result. It describes a map on which the Republicans are defending six seats no forecaster will call and the Democrats are defending one, with two days left to decide whether a national mood that has favored the Democrats all autumn will override the older gravity of Kansas, Iowa, Alaska and Maine.

What Changed This Week

Cook stopped hedging, and the hedge was the news. Its closing analysis rates seven races Toss Up, six of them Republican-held: Alaska, Iowa, Kansas, Maine, Ohio and Texas, with Michigan the lone Democratic-held toss-up.¹ The seat counts that follow are Solid D 9, Likely D 2, Lean D 2, Toss Up 7, Lean R 2, Likely R 2 and Solid R 11.¹ Work the arithmetic and the Democratic position is narrow but real: hold Michigan, hold everything already leaning their way, take North Carolina, and they still need three of those six Republican toss-ups to reach 51, which is what Cook’s own summary says.¹ Two seats moved toward the Democrats in the same document, Nebraska from Likely to Lean Republican and Florida from Solid to Likely Republican, and both moves are the same admission: seats that were not being contested are now being paid for.¹

Inside Elections moved nine races in one update. Its October 29 export shows Ohio and Texas each shifting from Toss-up to Tilt Democratic, both marked as flips; Georgia from Lean to Likely Democratic; Iowa and Kansas from Tilt Republican to Toss-up; Nebraska and South Carolina from Lean to Tilt Republican, which on its finer scale means the Republican edge narrowed rather than widened; and Louisiana and Mississippi from Solid to Likely Republican.² Nine races, every one of them toward the Democrats, in the last week of a campaign. Raters do not usually move this much this late, because late movement is indistinguishable from guessing.

Kansas crossed over. Kalshi’s Kansas market closed the week with Hamilton at 51 percent and Marshall at 49, a reversal of the 62 to 38 Marshall held on October 8, and both candidates stood at 50 the day before.³ The same day’s Emerson College poll, conducted October 26 to 28 for Nexstar, had Hamilton ahead 49 to 45, with the Libertarian at 3 percent and 3 percent undecided, and after pressing leaners the margin was 50 to 47.⁶ Emerson’s internals are the part that matters: Hamilton led independents 59 to 25, and voters who had already cast ballots broke for him 66 to 32 while those who had not yet voted favored Marshall 48 to 45.⁶ That is a lead built out of votes already banked, which is the kind that survives a bad final weekend. A Washington Post and Schar School survey in the field October 22 to 26 put Hamilton ahead 52 to 43, though the Post itself cautioned that Democrats agreed to be interviewed at a far higher rate than Republicans.⁷ The money explains why anyone is still polling the state at all: ad spending has passed $84.5 million, against a previous Kansas record of $49.1 million in 2020, with the Hamilton campaign at more than $16 million and the Marshall campaign at just over $6 million.⁸

Alaska turned from a lawsuit into an investigation, and the price moved. The New York Times reported that operatives aligned with Mary Peltola recruited candidates who shared the incumbent’s name, including Dan J. Sullivan of Petersburg, a retired teacher, in an effort to divert votes from Senator Dan S. Sullivan.⁹¹⁰ Peltola answered at a campaign event, “I knew nothing about any other candidates on the ballot except for me and the incumbent,” and added, “I am not interested in tricking voters.”¹¹ Her campaign said it “has no involvement with either Sullivan campaign,” and the Democratic Senatorial Campaign Committee said, “The DSCC is not involved with Dan J. Sullivan.”¹¹ The Justice Department sent at least three dozen subpoenas and document preservation requests to both campaigns, and Governor Mike Dunleavy ordered the state Department of Public Safety to determine “whether this scheme violated Alaska’s laws and whether criminal charges or civil action are warranted.”¹¹ Alaska Public Media noted of the Times reporting that its reporters “did not find anything that appeared illegal,” which is the sentence that matters most and will be quoted least.¹⁰ Kalshi nonetheless took Peltola from 75 percent on October 26 to 67 percent on October 30, with Sullivan at 33.⁹ Two of the models followed, moving Alaska from Lean Democratic to Toss-up in the same week.²

The final polls landed, and they agreed more than the ratings do. CNN and SSRS, in the field October 20 to 27, found El-Sayed ahead 49 to 42 in Michigan and Talarico ahead 49 to 43 in Texas, both outside the margin of error, while Kansas and Maine were inside theirs: Hamilton 49 to 45 over Marshall, and Troy Jackson 49 to 45 over Susan Collins.¹² The samples were 768, 923, 838 and 783 likely voters.¹² Marist, in the field October 8 to 11, had Josh Turek ahead of Ashley Hinson 48 to 45 in Iowa, and also found that 59 percent of likely voters expected Hinson to win anyway.¹³ A New York Times and Siena survey of Nebraska, conducted October 23 to 29 and released October 31, put Pete Ricketts ahead of Dan Osborn 49 to 43.¹⁴

The president cancelled two rallies and then uncancelled them. Rallies in Iowa and Kansas were called off last week and restored on Friday, and he is scheduled to appear in both states this Sunday.⁷ More than 51 million Americans had already voted as of November 1.⁷ In Texas alone, nearly 6.8 million of 18.8 million registered voters had cast a ballot, against 5.5 million at the same point in 2018.⁴ Early votes are not results, and Republicans have historically voted later, but a party that cancels a rally in a state it has held for ninety-four years and then restores it within the week is telling you something about which states it now believes it must save.¹⁵

The Defense

Michigan is the only Democratic-held seat the forecasters still call a toss-up, and the closing week was spent on a primary that ended in August. El-Sayed’s campaign asked Haley Stevens, his former primary opponent, to skip further joint events after she sought a condemnation of a Rebellion PAC advertisement that she said was “set aside ‘for killing Palestinians,’ and it blames our close ally Israel.”⁵ El-Sayed answered that she was focused on “litigating a primary that ended three months ago,” and Mike Rogers said the reply showed “he does not care about those concerns.”⁵ The numbers have not moved because of it. CNN had El-Sayed up seven, and Kalshi had him at 77 percent, with traders forecasting a margin of 2.8 points.¹²⁵ More than 2 million Michiganders had voted early, 51 percent of them Democratic and 39 percent Republican.⁵

Georgia has stopped being a contest. The polling average sits at Ossoff 53.3 to Collins 43.5, and every forecaster in the table rates the seat Likely Democratic or better, with Silver at Solid.¹⁶² The defense of a seat that no longer needs defending is a different job, and Ossoff has spent the closing week doing it for the rest of the ticket.

Minnesota is the Democratic seat that keeps narrowing without ever quite closing. KSTP and SurveyUSA’s final survey put Peggy Flanagan ahead of Michele Tafoya 48 to 41, with 7 percent undecided and a margin of error of 4.2 points, and the Carleton analyst Steven Schier called it plainly: “A margin of error race. Either candidate could win it. I think it’s really tight.”¹⁷ Sabato moved the race from Leans to Likely Democratic on October 29, which sounds like confidence and is better read as a judgment that the alternative is worse.¹⁸

New Hampshire has been the most stable race on the board for six weeks. The average of four polls is Pappas 49.8 to Sununu 42.5, and the most recent, UMass Lowell’s on October 29, had Pappas ahead 48 to 41.¹⁹ Those four polls span October 6 to October 29, and Sununu has not been closer than six points in any of them.

The other nine Democratic seats, Colorado, Delaware, Illinois, Massachusetts, New Jersey, New Mexico, Oregon, Rhode Island and Virginia, are not in play.

The Offense, Tiered

North Carolina remains the anchor, and the anchor is holding. The final High Point University and YouGov poll, released October 28 with fieldwork through October 20, had Roy Cooper ahead of Michael Whatley 54 to 40, with 4 percent unsure and the president at 37 percent approval and 57 percent disapproval in the state.²⁰ A party that has retreated from a state at the national level does not usually get a late surprise there, and the Republicans did exactly that here in the third week of October.¹⁵

Ohio is the race where the money and the polling describe different electorates. Kalshi had Sherrod Brown at 68 percent on October 27, with its own average at 48.3 to 44.0 and $431.7 million spent or reserved on advertising across the race.²¹ More than 1.2 million ballots had been returned, with Democrats about 60,000 ahead of Republicans.²¹ One curiosity from the closing week is worth recording: 25 percent of likely voters believe Brown is the incumbent, and that group splits almost evenly, 44.5 to 43.3.²¹ A quarter of the electorate not knowing which man already holds the seat is the kind of detail that makes a three-point lead feel narrower than three points.

Texas is the largest market in the country and the most stubbornly stable. Kalshi ended the week with Talarico at 71 percent to Paxton’s 29, and CNN had him ahead 49 to 43.⁴¹² Talarico has led in every poll this series has cited since August, and the early vote has already passed 2018’s early vote with days to spare.⁴ The question is no longer whether the lead is real but whether it survives a Texas November, which is the sort of thing Texas voters have heard before and have not yet been asked to do.

Maine may be the majority-making race, and it has become a referendum on one senator rather than a choice between two. CNN had Jackson ahead 49 to 45 and Marist had him ahead 49 to 47, inside a margin of error of 6.6 points, and Marist also found that 58 percent of likely voters expect Collins to be re-elected anyway; Kalshi had him at 64 percent.¹²²² The Wall Street Journal editorial board made the Republican case against its own party’s senator in the closing week, writing that “a Jackson victory, along with one by Mr. El-Sayed in Michigan, will advance the left’s growing strength.”²² Collins has answered by distancing herself from the president on tariffs and by campaigning with a former Democratic colleague, which is the correct instinct and also the tell: an incumbent who needs a Democrat to vouch for her has already lost the argument about which party she belongs to.¹⁵

Alaska is now a question about counting as much as about voting. Peltola entered the final week with three quarters of the market and lost eight points of it in four days, the ranked-choice count means a majority is assembled rather than won, and the Justice Department has asked a court to delay certification if Sullivan loses by fewer votes than the second Sullivan received.⁹¹¹ The state’s own investigation is running in parallel.¹¹ A market collapse that steep, this late, is not nothing. But the ballots are already cast, and no subpoena reaches a ballot box.

Iowa is the race where the market and the polls disagree and both sides can point to a number. The polling average is Hinson 45.8 to Turek 45.2;²³ Marist had Turek ahead 48 to 45;¹³ Kalshi had Hinson at 55 percent.²⁴ Marist also found that 59 percent of likely voters expect Hinson to win, which is either a forecast or a habit, and in a state that has not elected a Democratic senator since 2008 the difference matters.¹³ When the average says half a point and the market says ten, the market is pricing the state’s history rather than its polls, and history has been the better bet here for two decades.

Nebraska is the cheapest fourth seat and the one the final polls have probably settled. Siena’s October 23 to 29 survey put Ricketts ahead of Osborn 49 to 43, and the finding underneath it is the whole story: asked whether Osborn is more a Democrat, a Republican or an independent, 55 percent said a Democrat and 33 percent said an independent, and among voters who backed the president in 2024 the first figure rises to 78 percent.¹⁴ An independent who cannot persuade voters he is independent, in a state the president carried comfortably, does not usually close a six-point gap in two days.

Kansas is the margin seat and now the most expensive election in the state’s history. The spending passed $84.5 million, the Emerson poll put Hamilton ahead by four with early voters breaking for him more than two to one, and the market moved the race to the Democrat for the first time.⁸⁶³ The caution is unchanged and should be stated plainly: Kansas produced competitive September polls in 2014 and 2020 and elected Republicans by double digits in both Novembers, and Marshall won by 11.4 points in 2020 against polls that looked much like these.¹⁵ What is different this year is not the polling. It is that the spending is coming from Republicans trying to hold a seat they never expected to defend.

Florida has tightened without becoming competitive, which is the distinction Cook drew on October 30 when it moved the seat from Solid to Likely Republican.¹ The average of six polls is Moody 50.2 to Nixon 44.5, and the final Stetson survey had it 49 to 45.²⁵ Inside Elections made the same move on October 15.² A four-point race in a state the president carried comfortably is a warning about the national environment, not a prediction about this seat.

Louisiana is tied in the only poll anyone has released, and that poll belongs to the Democrat. Hart Research surveyed 500 likely voters on October 28 and 29 for the Davis campaign and found the race 45 to 45 with 10 percent undecided.²⁶ Every forecaster still rates the seat Republican, and two of them tightened it in the past week, Sabato and Inside Elections each moving it to Likely Republican.¹⁸² The honest reading is that a tied partisan internal in a state the president carried easily tells you the Democrat is alive, not that he is ahead.

Mississippi is the quietest of the long shots and the one with the largest gap between its polling and its rating. A Change Research survey of 842 likely voters, fielded October 21 and 22 and sponsored by Government that Works PAC, had Hyde-Smith ahead of Colom 43 to 40, with the independent Ty Pinkins at 6.²⁷ The modelling site 535.wtf read the same poll and left Colom at a 1 in 100 chance of winning, which is the honest translation of a three-point poll in a state the president carried comfortably.²⁷ Cook moved the seat from Solid to Likely Republican this month, which was the correct response to $1.2 million of late Democratic spending, and also a reminder that moving a rating is not the same as moving a seat.²¹⁵

Montana has a three-way race that has not resolved into a two-way one. The Montana State University Billings Mountain State Poll, released October 29 with 720 responses and a margin of error of 3.6 points, had Kurt Alme ahead of Seth Bodnar by four points and Alani Bankhead by nine.²⁸ Bodnar’s independent candidacy is the reason the seat is not entirely safe and the reason it remains Republican-favored: a four-point deficit with a Democrat still taking double digits is not a path to 50.

South Carolina is the race where three forecasters moved toward the Democrats in one week and the market barely moved at all. Cook kept it Lean Republican, Inside Elections and Sabato each moved it a notch toward the Democrats, and RealClearPolitics went to Toss-up.² The polling underneath is genuinely tight: a South Carolina Policy Council survey in the field October 16 to 25 with 1,200 registered voters had Darline Graham ahead of Annie Andrews 47 to 46, and Kalshi’s average is 45.6 to 45.3.²⁹ Kalshi still prices Graham at 75 percent, which is what a state the president carried by nearly eighteen points does to a two-point race.²⁹

The nine safe Republican seats, Alabama, Arkansas, Idaho, Kentucky, Oklahoma, South Dakota, Tennessee, West Virginia and Wyoming, are not in play.

The Math

The requirement has not changed since September and it is worth restating in the last edition of this series. Republicans hold 53 seats. Democrats need a net gain of four for 51, because a 50-50 Senate belongs to the vice president.¹ That means holding Michigan and winning four of the six Republican-held toss-ups, or five of the six if Michigan slips.

The forecasters now agree on the direction and disagree only about the size. Cook calls the Democrats slight favorites with a gain of two to six seats.¹ Sabato’s analysis of October 27 puts the Senate at 49 at least leaning Democratic, 46 at least leaning Republican and five toss-ups, Maine, Iowa, Kansas, Ohio and Texas, with Alaska and Michigan at least leaning Democratic, and concluded that the Democrats have “the inside track to a Senate majority, albeit narrowly at 51 or 52 seats.”¹⁸ Decision Desk HQ gives the Democrats 60 percent.¹

The market is further out than any of them. Kalshi put Democratic control at 68 percent in its own reporting on October 29, and its seat market projects 52 Democratic seats.³⁰³¹ The gap between 60 and 68 percent is not a disagreement about the direction. It is the width of the last mile, and both numbers describe a Senate that turns on three or four states.

The paths run through the same door. Hold Michigan, Minnesota and New Hampshire, take North Carolina, and win any three of Alaska, Iowa, Kansas, Maine, Ohio and Texas. That is 51, which is what Cook describes.¹ Win four of the six and it is 52, which is the outcome Kalshi’s seat market and Sabato’s range both describe.³¹¹⁸ The defensive path is the one nobody wants to write down: a loss in Michigan or Minnesota or New Hampshire raises the requirement to five flips, and five flips means winning a state that has not sent a Democrat to the Senate in ninety-four years.¹⁵

The Kansas connection is the argument in miniature, and it has finally moved. In September the state was a curiosity. In October it became the most expensive race in its history. On October 30 Kalshi, whose own news desk publishes its prices, moved the race to the Democrat, and a Republican incumbent who raised $6 million against a challenger’s $16 million is now the underdog in a state his party has held since 1932.³⁸¹⁵ If Kansas holds Republican, the Democratic path still runs through Ohio and Texas and Maine. If Kansas falls, the Democrats can lose one of those and still reach 51. That is what a margin seat looks like two days out: not a symbol, but a spare.

The ratings table

The columns are refreshed on different schedules and the differences are real, so the dates are given per column rather than averaged: Cook October 30; Inside Elections October 29; Sabato October 29; RealClearPolitics October 31; Decision Desk HQ October 31; Silver Bulletin October 31.² Georgia has dropped out of this table under its own rules, since no column now rates the seat closer than Likely Democratic; its numbers are in the defense section above. Eighteen cells in this table differ from the edition of October 25. Not all of that movement happened in the last seven days, because Sabato, RealClearPolitics, Decision Desk HQ and Silver were between one and three weeks stale a week ago and have since refreshed, which is itself part of the story; across all 35 races, the moves carrying published dates are Inside Elections’ nine on October 29, Cook’s two on October 30, and Sabato’s five between October 27 and 29.² The largest single move is Alaska, where Decision Desk HQ and Silver each went from Lean Democratic to Toss-up in the same week the state’s investigation opened.²

A word on what the columns mean. PVI is the Cook Partisan Voting Index, and it is the only backward-looking number here. It measures how far a state leans from the nation as a whole, computed from the 2020 and 2024 presidential two-party vote shares weighted toward 2024, and Cook says of it that the race ratings reflect an outlook for which party will win the next election while the PVI takes a longer view. It earns its place as contrast, which is what makes “Tossup in an R+6 state” legible, and not as a forecast. Lean is a strong prior across the whole map: Pew found 158 of 176 Senate elections since 2012, 89.8 percent, went the way of the state’s presidential carry. But applied as the sole predictor to the races this series tracks, lean gets several of them backwards, calling Georgia, North Carolina, Ohio and Texas for the Republicans while every forecaster in the table now favors the Democrat in each. That is not a contradiction but a selection effect: a race earns its way onto this table precisely by being a case where the baseline does not settle it. Never argue from PVI against the forecasters, because where the two disagree, the forecasters are the point of the exercise.

Two traps make that column look inconsistent with numbers quoted elsewhere. The first is that PVI is denominated in vote share for one party, not margin, which makes it roughly half the size of a margin-style lean: Cook calls Texas R+6 while Silver’s partisan lean calls the same state R+11, and both are right, because the denominators differ. The second is that the scale flattens elasticity, how hard a state swings when the nation swings, and elasticity is orthogonal to lean, correlating with it at only r = +0.12 across these races. New Hampshire, at 1.15, and Maine, at 1.13, are among the most elastic states in the country; Georgia, at 0.90, is among the most inelastic. A state can be strongly partisan and highly elastic at once, which is why a lean column can never stand in for a competitiveness column.

IE is Inside Elections, Nathan Gonzales’s nonpartisan newsletter, one forecaster among several. The rating scale runs, strongest to weakest, solid, likely, lean, tilt, and toss-up, and Tilt is Inside Elections’ own rating: Cook, Sabato and Decision Desk HQ do not use it, so a Tilt appearing only in the IE column is not a disagreement but a finer scale. Decision Desk HQ publishes win probabilities that map onto the same ladder: above 95 percent is solid, 80 to 95 likely, 65 to 80 lean, and 35 to 65 a toss-up for either party. (flip) marks a rating that forecasts a change of party from the current holder.

A note on the sources. One column in that table is arithmetic and six are not. PVI alone can be reproduced from certified FEC totals, which we have done, matching Cook’s published figures; everything else is somebody’s assessment. Cook, Inside Elections, Sabato and RealClearPolitics are analyst judgment, which means their ratings cannot be independently reproduced from any underlying data, because the rating is the data. That is expert opinion in a table rather than a measurement, and it is not a scandal: election analysis is not a solved problem. RealClearPolitics is worse than unaudited, because it blocks automated reading, so we cannot say whether that column is independent judgment or a composite of the others; treat it as unlisted rather than as a seventh opinion. Decision Desk HQ and Silver are published models, and Decision Desk HQ is the most independent column here, using neither expert ratings nor Cook’s PVI and deriving its prior from the most recent presidential margin adjusted for the national swing. Silver’s FLIPR is not independent in the same way, because in its default build it feeds the Cook, Inside Elections and Sabato ratings in as an input at roughly one-sixth weight, so where his column agrees with theirs the agreement is partly built in and is not corroboration. All six read the same polls and talk to the same operatives, which means six agreeing columns are not six confirmations; the signal is where they disagree, and two days out that is Kansas, Iowa, Nebraska and South Carolina. Freshness differs per column and is stated above rather than averaged.

The Close

Two days is the part of a campaign in which nothing that has not already happened can happen. Every ballot that will decide Kansas has probably been cast. The subpoenas in Alaska will not be answered before Tuesday, and no court will rule on a certification that has not been requested. The final polls are in the field or out of it, and the last undecided voters are being asked to break a tie that has not moved since Labor Day.

What moved instead was everyone watching. Cook spent the autumn refusing to call this Senate and called it on Friday. Inside Elections moved nine races in one update. A market that had Marshall at 82.8 percent on September 5 finished October with him behind. None of that is a vote. All of it is a judgment, made by people who read the same polls and talk to the same operatives, and the fact that they now agree is less evidence than it looks, because late agreement is one of the things a campaign is designed to produce.

The country is not being asked on Tuesday whether it likes this president. That was answered months ago, in the thirties and staying there. It is being asked whether a midterm electorate angrier about prices than about anything else will spend that anger on the party holding the Senate, and whether four states with long memories, Kansas and Iowa and Alaska and Maine, will follow a national mood or their own older gravity. Kansas has voted Republican for ninety-four years. Iowa has not sent a Democrat to the Senate since 2008. Maine re-elected Susan Collins after the polls said she would lose. Alaska counts in rounds, and this one may not finish on election night.¹⁵ Two days is enough time for any of those habits to hold, and enough time for all of them to break at once.

Notes

PRH | huffmanwrites.org | © Philip Huffman