Senate Race Report: September 27, 2026

Question: Six weeks from Election Day, after the Cook Political Report called control of the Senate a true coin flip and moved three races toward the Democrats, with the national generic ballot at its widest Democratic margin of the cycle and Kansas now polling like a race either party can win, is the majority still a Republican-held map, or has it become the toss-up the professionals now describe?

Answer: It is a toss-up, and this week the professionals said so in writing. Jessica Taylor’s September 23 analysis for Cook put control at “a true coin flip” with “viable paths for both parties,” and gave her own range as a Democratic net gain of one to five seats against the four they need.¹ Three races moved toward the Democrats in the same piece: Georgia to likely Democratic, Kansas to lean Republican, and South Carolina to likely Republican.¹ The national numbers moved with them. NBC News released polling averages on September 25 that put the Democrats nine points ahead on the generic ballot, their widest lead of the cycle, with the president at 39 percent approval and independents favoring a Democratic Congress by sixteen.² Kalshi’s own desk had the Democrats at 61 percent to take both chambers on September 22, with its “blue wave” contract up more than nine points in a single week.³⁴

A coin flip is still not a win, and Cook’s own caveat is the honest one: a “very blue environment” does not by itself produce a Senate majority when the seats in question lean Republican by double digits and the other party is spending more.¹ The week handed the Democrats their best polls of the cycle and their first real scandal, in the same seven days. That is what six weeks out looks like.

What Changed This Week

The most consequential document of the week was a ratings note. Cook’s Jessica Taylor wrote on September 23 that control is now “a true coin flip,” the first time she has framed it that way, having rated control a toss-up a month earlier but with the fundamentals still tilting Republican, and moved three races toward the Democrats in the same analysis.¹ Georgia went from lean Democratic to likely Democratic on the strength of Jon Ossoff’s roughly $42 million raised against a Republican who has not kept pace; Kansas went from likely Republican to lean Republican, the first time Cook has rated the state that close this cycle, though not the first time in recent memory, since Cook closed 2020 with Kansas at lean Republican and 2014 with it at toss-up; and South Carolina went from solid Republican to likely Republican, prompted by strong Democratic fundraising for Annie Andrews and Republican worry about Darline Graham’s “soft numbers.”¹⁵ Cook’s range of outcomes is now a Democratic net gain of one to five seats, and the four they need sits in the middle of it.¹ Taylor also named the mechanism plainly: whether “a deluge of ad spending can save Republicans amid a worsening environment as tariffs and the Iran War continue to spike prices.”¹

The national environment kept widening beneath the ratings. NBC News’s polling averages, released September 25, put the Democrats nine points ahead on the generic ballot at 52 to 43 after adjustment to a likely-voter scale, the largest lead of the cycle, with the president at 39 percent approval and 59 percent disapproval and his standing among independents at 26 percent.² The Silver Bulletin’s average rose to Democrats plus 8.1, its own cycle high, from 6.7 at the start of the month.⁶ Emerson’s national survey, in the field September 21 and 22 and released September 24, gave the Democrats eleven points, 53 to 42, with independents breaking 60 to 31.³ USA Today reported on September 25, alongside Cook’s analysis, that the Democrats are now positioned to win control of at least one chamber.⁷

The market’s own record matched the polls. Kalshi’s news desk reported on September 22 that its “blue wave” contract, which requires Democrats to hold at least 218 House seats and 49 Senate seats, had climbed from 73.8 percent to 83.6, a rise of more than nine points in a week, and stood at 83.0 by the close of that article, while the higher-threshold “blue tsunami” contract, which needs 51 Senate seats, rose 6.5 points in five days to 47.0 percent.³ The same desk’s balance-of-power market on the same date priced Democratic control of both chambers at 61 percent, a Democratic House with a Republican Senate at 30 percent, and Republican control of both at 8.2 percent.⁴ Secondary outlets that track the exchange reported its Senate-control contract at about 64 percent on September 25; Kalshi’s own desk has not published that number, so it is reported here as an aggregator’s reading rather than as the market’s record.⁸

Kansas was the week’s clearest single-race development, and it came in two parts. Wedgewood Polls, an independent firm that funded the survey itself, released a September 22 to 24 reading of 500 likely voters showing Adam Hamilton at 50 percent to Roger Marshall’s 48, with the Libertarian at 2 and a margin of error of 4.4 points.⁹ Hamilton’s lead sits inside that margin, so the poll does not establish a leader; what it does establish is a second consecutive independent survey with the Democrat at or above the Republican, after Emerson’s 45 to 43 a week earlier.⁹¹⁰ Cook then moved the race to lean Republican, and for the first time this cycle Kansas earns a row in the ratings table below.¹

Georgia got safer without anyone campaigning for it. InsiderAdvantage’s September 22 and 23 survey of 1,200 likely voters has Ossoff at 50 to Mike Collins’ 42; Big Data Poll’s September 21 to 23 reading of 678 likely voters has it 52.2 to 41.0; YouGov’s September 15 to 18 survey of 3,447 registered voters has it 51 to 41; and Rasmussen’s September 14 poll had Ossoff at 51, the threshold that would spare him a runoff.¹¹ Big Data found 14.3 percent of Trump’s 2024 voters saying they would back Ossoff against 2.9 percent of Harris voters backing Collins, which is what a widening Senate lead looks like underneath a tied governor’s race.¹¹

Maine produced the week’s most honest disagreement. Three polls landed within days of each other and pointed three ways: UMass Lowell and YouGov, in the field August 28 to September 14 and released September 24, had Troy Jackson ahead 48 to 43; the University of New Hampshire had him ahead 51 to 47; and a New York Times, Portland Press Herald and Siena poll released the same day had Susan Collins ahead 49 to 46.¹² The three differ mainly on how independents break and how many Democrats are willing to cross for Collins, which is the whole question in the state.¹² Kalshi’s own desk reported on September 26 that Collins’ price had risen 8.7 points to 36 percent since September 24 while Jackson still led the winner market at 62 percent, and that the exchange’s margin market forecast a Democratic win by 1.8 points.¹³ Kalshi is pricing a narrow Democratic win and a Republican recovering at the same time, which is a fair description of a race this unsettled.

New Hampshire reversed. The UNH Survey Center’s September 17 to 21 poll of 1,418 likely voters, released September 24, put Chris Pappas at 50 to John E. Sununu’s 42, a turnaround from August, when Sununu led 45 to 43.¹⁴ Siena’s September 24 survey had Pappas ahead 50 to 45.¹⁴ Pappas gained eighteen points among progressives since August while Sununu lost seven among moderates, and both men remain underwater in favorability, which is the shape of a race where the environment is doing more work than either candidate.¹⁴

Alaska gave the Democrats their first genuine scandal of the cycle. The New York Times reported on September 23, based on interviews with nine former staff members, that Mary Peltola had repeatedly blown up at aides, that she shoved an aide in 2024, and that she used a homophobic slur and an ableist one in a text this year.¹⁵ Her campaign denied the pattern and attributed her difficulties to grief after losing her mother and her husband within months in 2023; it acknowledged and regretted the ableist term, and Peltola declined to be interviewed and avoided reporters at a town hall the same day.¹⁵ The Sullivan campaign called it a “disturbing track record.”¹⁵ Kalshi’s desk, publishing on September 24, reported that traders had shrugged: Peltola remained at 74 percent to Sullivan’s 29, with the top margin outcome a Peltola win of three to six points at 26 percent.¹⁶ The exchange has priced the story as noise. That is a judgment, not a fact, and it is the kind that can be revised by one poll.

Texas and Michigan both produced their best Democratic numbers of the cycle. An NPR and Marist survey, in the field September 17 to 20 with 1,139 registered voters and a margin of error of 4.4 points, had James Talarico at 50 to Ken Paxton’s 44, and Cook has noted that third-party candidates in Texas, Iowa and Ohio mean Democrats may not need to reach 50 percent at all.¹¹⁷ Decision Desk HQ’s average had Talarico up 3.8 on September 25 and its model gave him a 56 percent chance, while ad spending in the state has passed $183 million with Republicans outspending Democrats roughly three to one.¹⁷ In Michigan, Big Data’s Wolverine State Poll, in the field September 22 to 24, put Abdul El-Sayed at 45.9 to Mike Rogers’ 40.5 among likely voters, and a Suffolk survey released September 22 had it 47 to 40; Kalshi’s desk reported El-Sayed at 70 percent on September 22, up from a low of 55 in the days after his August 4 primary win.⁴¹⁸ The exchange’s separate margin market forecast a Democratic win by 3.5 points.⁴

Iowa’s polls moved toward the Democrats; its averages did not. A Marist survey in the field September 17 to 20, with 1,050 registered voters, put Josh Turek at 50 to Ashley Hinson’s 42, his widest lead of the cycle, even as 59 percent of the same respondents said they expected Hinson to win.¹⁹ Republican-affiliated firms still show Hinson ahead or tied, and the polling average sits within a point either way, which is why Iowa remains the race where the pollster’s sponsor predicts the result better than the poll does.¹⁹

Florida reverted, and last week’s report should be corrected on the point. Two September surveys now put Ashley Moody back in front: Stetson’s Center for Public Opinion Research, in the field September 14 to 21 with 830 adults, has her ahead 51 to 40, essentially unchanged from its spring reading, and InsiderAdvantage’s September 20 and 21 survey of 600 likely voters has her up 49 to 42.²⁰ Last week’s “first public reading without a Moody lead,” a St. Pete Polls tie at 45, now looks like the outlier rather than the turn.²¹ South Carolina tightened instead: Trafalgar’s September 23 to 25 survey of 1,087 likely voters has Graham at 46 to Andrews’ 45, a one-point race in a state the president carried by about eighteen points, and Cook’s move to likely Republican was a response to exactly that.¹²²

The war kept setting the price of everything else. Brent traded between roughly $102 and $107 a barrel across the week, West Texas Intermediate between $91 and $95, and the American diesel price set a record at $6.51 a gallon.²³ The Strait of Hormuz has now been closed for nearly seven months; an adviser to Iran’s supreme leader said in a video carried by the Fars news agency that the war could spread to the Indian Ocean “or perhaps even beyond,” which pushed Brent above $106, and Iran’s security chief said the strait would not reopen unless its conditions were met, even as Tehran signaled it was reviewing an American proposal.²³ Cook’s Taylor named the connection between that and the Senate map directly, and the Kansas race is the clearest case of it: fuel, fertilizer and beef tariffs are the issues doing the work in a state where the Republican should not need help.¹

The Defense

Michigan is still the most competitive Democratic seat in the country, and the week was its best of the cycle. El-Sayed leads by five and a half in the Wolverine State Poll among likely voters and by seven in Suffolk, and Kalshi’s desk has him at 70 percent after trading as low as 55 in August.⁴¹⁸ The caveat is unchanged: El-Sayed runs ahead of his party’s national numbers in a state that is even on the partisan index, and the model consensus is narrower than the market consensus, with Decision Desk HQ at 67 percent against Kalshi’s 70.⁴ A five-point lead six weeks out is a good week, not a settled seat.

Georgia is the second seat, and it stopped being interesting. Four September surveys have Ossoff between eight and eleven points ahead, Cook moved the race to likely Democratic, and the underlying numbers explain why: the Republican nominee is being outspent by a wide margin and is losing independents by nearly twenty points.¹¹¹ A runoff is now the only realistic Republican path, and Rasmussen’s finding of Ossoff at 51 percent is the number that closes it.¹¹

Minnesota is the defense seat that has not settled. Three September surveys put Flanagan between even and four points ahead, and the polling average sits near three and a half, which is a long way from the nine-in-ten favorite the state was a month ago.²⁴ The Republican, Michele Tafoya, led independents by fourteen in the SurveyUSA reading and has gained while Flanagan has slipped; if the models are wrong anywhere on the defense, this is the likeliest place.²⁴

New Hampshire swung back, and the swing is worth reading carefully. Pappas went from a two-point deficit in August to an eight-point lead in the UNH poll a month later, which is a ten-point move in a state with a two-point Democratic lean and among the most elastic electorates in the country.¹⁴ The mechanism is the national environment rather than the candidates: both men are underwater, and Pappas gained among progressives while Sununu lost among moderates.¹⁴ Cook still rates the race a toss-up, and a ten-point month is the kind of thing that can reverse in a week.

The other nine Democratic seats, Colorado, Delaware, Illinois, Massachusetts, New Jersey, New Mexico, Oregon, Rhode Island, and Virginia, are not in play.²⁵

The Offense, Tiered

North Carolina remains the anchor of every path, and it had a quiet week that confirmed the shape of the race. Common Cause North Carolina’s survey, in the field September 17 to 20 with 857 likely voters, put Roy Cooper at 51 to Michael Whatley’s 41, with leaners allocated 52 to 44; High Point University’s earlier September poll had it 50 to 42; and no public survey this cycle has shown Whatley ahead.²⁶ The more telling item was a Politico report that the Senate Republicans’ campaign arm no longer counts North Carolina as a priority and intends to move money to defending incumbents in Iowa, Texas, Maine and Alaska. Whatley disputed it, and Cook’s own note that Republicans retain a financial advantage is the reason the report matters more than the denial.¹²⁶

Ohio and Texas are the two coin flips, and both now lean left of even. Rasmussen’s September 22 and 23 survey of 1,115 likely voters put Sherrod Brown at 46 to Jon Husted’s 43, and the averages sit near Brown plus three, with Brown leading independents by eighteen and holding a large fundraising edge.²⁷ No Senate debate has been scheduled in Ohio, and Husted’s campaign has said it would be shocked if Brown agreed to one, which is where that stands.²⁷ Texas is the week’s strongest Democratic read: Talarico led the Marist survey by six, the Decision Desk HQ average has him up 3.8, and Kalshi’s own Texas market on September 24 put him at 58 percent to Paxton’s 42 with a forecast Democratic margin of 0.2 points.¹⁶¹⁷ A margin market at two-tenths of a point is the exchange saying it does not know, and the Republican super PAC money now reserved in the state is the party saying it agrees.

Maine, Alaska, Iowa and Nebraska are the bench, and the week made three of them live. Maine now has a set of September polls pointing three ways, an incumbent whose price has recovered 8.7 points since Thursday, and a debate schedule that begins after early voting opens on October 5, which is the arrangement Collins wanted.¹²¹³²⁸ Alaska has a candidate under a New York Times investigation from September 23 and a market that priced it as noise within a day, which leaves the race where it was: a ranked-choice contest with a same-named spoiler where the Democrat won the primary by eight and the AARP survey’s reallocation gave her the general by six.¹⁵¹⁶ Iowa is now the tightest of the four in the averages and the widest in the polls, with Marist at Turek plus eight and the Republican-affiliated firms still ahead or tied, and the models have barely moved on it all cycle.¹⁹ Nebraska remains the wildcard, essentially even in the averages with one survey showing Osborn up four and another showing Ricketts up four, in a state the president carried by twenty points, and an independent win there is still the cheapest fourth seat available because it needs no party apparatus at all.²⁹

Kansas is the margin seat, and this week it stopped being a curiosity. Two independent surveys in eight days have the Democrat at or above the Republican, Cook has moved the state to lean Republican for the first time this cycle, and the ratings table now carries a Kansas row.¹⁹¹⁰ The honest caution has not changed and deserves repeating: Kansas produced competitive September polls in 2014 and 2020, and the Republicans won both Novembers by double digits.⁵ What is different this time is not the September poll but the money and the mechanism. Hamilton raised more than $3.5 million in his first two months, a record for a Kansas Senate candidate, and the issues driving the race, fuel, fertilizer and beef tariffs tied to the Iran war, are the ones Cook cited when it moved the state.¹³⁰

Florida is not a race again yet, and last week’s report overstated it. Moody leads by eleven in the Stetson survey and by seven in InsiderAdvantage, she outspends Nixon by more than ten to one, and the tie that looked like a turn was one pollster.²⁰²¹ Louisiana and Mississippi remain long shots with real activity: a pollster working for Jamie Davis has Letlow ahead by four in a survey the Davis campaign paid for, which is a smaller lead than the money would suggest, while Scott Colom has raised more than Hyde-Smith in two consecutive quarters and the two spent September trading attack websites, with the incumbent still declining a debate.³¹³² Montana and South Carolina are Republican-favored and no longer comfortable: Kurt Alme leads a split field at 38 to Seth Bodnar’s 30 in the Montana Free Press reading, with the Democratic vote splitting badly enough that 53 percent of Democrats told pollsters they preferred Bodnar to their own nominee, while Graham leads Andrews by one in Trafalgar’s new survey after Cook moved the state to likely Republican.²²³³

The nine safe Republican seats, Alabama, Arkansas, Idaho, Kentucky, Oklahoma, South Dakota, Tennessee, West Virginia, and Wyoming, are not in play.²⁵

The Math

The forecasters moved this week, which is the change worth naming. Cook moved three races toward the Democrats on September 23 and described control as a coin flip; Sabato’s Crystal Ball, which had not updated since August 26, moved Iowa to toss-up and Michigan to lean Democratic when it refreshed on September 22; RealClearPolitics moved New Hampshire to lean Democratic and Nebraska to toss-up on its September 24 update; Decision Desk HQ made Alaska lean Democratic on September 25, the first forecaster to put a flip on the state; and Silver’s model moved Iowa to lean Democratic, Michigan to likely Democratic, and Kansas to lean Republican, while stepping Georgia back from solid to likely Democratic.³⁴ Five of the six columns in the table below refreshed within the week, and every move among them was toward the Democrats except Silver’s Georgia revision, which was a model acknowledging that an eight-point lead is not a fifteen-point one.

The forecasters and the markets are now telling nearly the same story, which is itself the story. For two weeks the tables sat still while the models and markets moved, and this report said so. That gap closed this week: Cook’s own range, a Democratic net gain of one to five seats, contains the market’s implied outcome, and the market’s own prices agree with it once they are read properly. Kalshi’s balance-of-power market prices the four possible outcomes separately on September 22, at 61 percent for Democratic control of both chambers, 30 percent for a Democratic House with a Republican Senate, 8.2 percent for Republican control of both, and 0.8 percent for a Republican House with a Democratic Senate.⁴ Add the two Democratic-Senate outcomes and the exchange’s own numbers put Democratic Senate control at 61.8 percent, which is a price the Democrats would take and roughly the same place Cook’s range points toward.⁴ The disagreement that remains is not between the professionals and the exchanges but inside the ratings columns, and it is concentrated in four states: Georgia, Iowa, Ohio and Texas, where the models and the markets are more Democratic than the analyst columns are.

The market path is North Carolina, Ohio, and Texas, plus one of Maine, Alaska, Iowa, or Nebraska. That is four flips, and it is what the markets are pricing.

The consensus path is North Carolina, Ohio, Maine, and Alaska or Iowa. It requires winning the top four.

The nightmare is any loss on the defense, and the defense is now the harder half of the map. Minnesota sits near three and a half points in the averages against a nine-in-ten model, New Hampshire moved ten points in a month, and Michigan’s market price runs three points ahead of its model. Every defensive loss raises the requirement to five flips, and five flips means winning Kansas or Nebraska or Florida on top of everything else.³⁵

The Kansas connection is worth stating plainly again. If the Democrats sweep the top three, Kansas is not needed. If any of the three slips, Kansas, or Nebraska through Osborn, becomes the margin seat. This week Kansas got its second independent poll with the Democrat ahead, its first lean-Republican rating from Cook this cycle, and a row in the table below for the first time this cycle, in the same week that Cook named tariffs, fuel and fertilizer as the reason.¹⁹ The margin seat is no longer a theory that requires a Kansas poll to sustain it. It is a race the forecasters have moved, in a state where the party in power sent the vice president two weeks ago and has not elected a Democratic senator since the 1930s.¹⁰

The ratings table

Ratings as of September 27, 2026, with each column on its own schedule: Cook Sep 23; IE Sep 17; Sabato Sep 22; RCP Sep 24; DDHQ Sep 25; Silver Sep 23.

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 it is denominated in vote share for one party rather than margin, which is why it runs about 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. Cook says so himself, that 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, with Pew finding 158 of 176 Senate races since 2012 going the way of the presidential carry, but applied to exactly these ten races it calls six correctly and misses the same four it has missed all cycle: Georgia, North Carolina, Ohio and Texas, every one of which it would call Republican and every one of which the forecasters now favor for the Democrat. 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. The counter-dimension is elasticity, how hard a state swings when the nation swings, and it is nearly orthogonal to lean, correlating at only r = +0.12 across these races. Maine and New Hampshire are among the most elastic states in the country, which is part of why their polls moved ten points in a month, while Georgia and North Carolina are among the most inelastic, which is part of why forecasters rate them Democratic despite Republican-leaning PVI numbers. The rule that follows is simple: never argue from PVI against the forecasters, because where the two disagree, the forecasters are the point of the exercise. IE is Inside Elections, Nathan Gonzales’s nonpartisan newsletter, one forecaster among several. The rating scale runs solid, likely, lean, tilt, toss-up, and Tilt is Inside Elections’ own rating; Cook, Sabato and DDHQ do not use it, so a Tilt appearing only in the IE column is not a disagreement but a finer scale. (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, and the difference is worth two sentences. PVI alone can be reproduced from certified FEC totals, which we have done, matching Cook’s published figures; Cook, Inside Elections, Sabato and RCP are analyst judgment, which means their ratings cannot be independently reproduced from any underlying data because the rating is the data, and that is expert opinion in a table rather than a measurement. RCP’s method in particular is unverified, because the site blocks automated reading, so treat that column as unlisted rather than as a seventh opinion. DDHQ and Silver are published models, and DDHQ is the most independent column here: it uses neither expert ratings nor Cook’s PVI, 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, currently Georgia, Iowa, Ohio and Texas. Freshness differs per column and is stated above rather than averaged, and this week that matters, because five of the six moved.

The Close

The report lands six weeks from Election Day with the last holdouts having conceded the premise. For two weeks this series described a market and a model set that had moved while the rating tables would not. This week the tables moved: three races toward the Democrats from Cook, Iowa and Michigan from Sabato, New Hampshire and Nebraska from RealClearPolitics, Alaska from Decision Desk HQ, four races from Silver, and a Kansas row where a week ago there was none. Cook’s own range now contains the outcome the Democrats need.

The honest cautions are the same ones, and they are not smaller for being familiar. The ratings moved because the polls moved, and September polls in Kansas have been wrong before, in 2014 and 2020, by double digits. Minnesota and Michigan both poll closer than the models say, which means the defense is doing less work than the map implies. Florida went back to Moody within a week of tying, which is a reminder of how quickly a single pollster’s reading can look like a trend. And Cook’s central caveat stands: a blue environment does not flip a Senate seat on its own, and the party defending the majority is still the party with more money.

But the direction is no longer seriously in dispute, and the shape of the endgame is clear. Hold the defense, flip North Carolina, win two of the coin flips, and find a fourth seat somewhere in the long tail. This week the forecasters moved Kansas into the marginal column, Kalshi’s own balance-of-power prices sum to about 62 percent for a Democratic Senate, the generic ballot reached its cycle high, and the Kansas margin seat went from a curiosity to a rated race. Six weeks is a long time, and it is also, at this point, the only thing standing between the Democrats and a majority they are now priced to win.

PRH | huffmanwrites.org | © Philip Huffman

Notes