2026 Senate Swing State Odds: Kansas Broke First

Published 30 October 2026 · Prices frozen 30 October 2026, 21:14 UTC

Four days out, the 2026 Senate swing state odds have stopped behaving like a national story and started behaving like seven local ones. The Senate control board has barely moved in a fortnight. Underneath it, Kansas has lost a Republican incumbent 14.5 points in eight days and Alaska has handed one 5 back.

That matters because almost every tracker you can read tonight builds its Senate call the same way: take the seven toss-ups, multiply them out, add North Carolina, announce a number. Do that with tonight’s prices and you get 80.88% for a Democratic Senate. Polymarket’s own seven-state board, which prices the races jointly rather than one at a time, says 66.82%. Fourteen points of difference, and not one cent of it is about who is ahead in Maine.

We pulled every leg of all 128 combinations, both chambers of the control market, the eleven-rung seat ladder, seven single-state boards and fourteen Kalshi contracts in one pass, normalised each book against its own sum rather than quoting raw mids, and bracketed every apparent gap at the bid and the ask before calling it a gap. All prices are frozen at 30 October 2026, 21:14 UTC; the live boards are linked throughout.

Where the 2026 Senate swing state odds stand

Start with the seven races both exchanges agree are live. They are the states named on Polymarket’s seven-state swing board and they are also, to the state, the seven that RealClearPolitics rates Toss Up. Nothing else on its Senate page is closer than Leans.

Six of the seven are Republican-held. Michigan is the single Democratic seat in the group, open since Gary Peters announced his retirement, and it is also the one the market thinks is safest — Mike Rogers pays 23.5¢, the lowest Republican price on the board. Iowa and Texas are open on the Republican side, Ohio is a special election for the remainder of the vice president’s old term, and the other three are incumbents defending.

Two things fall out of that last row. Polymarket expects Republicans to win 2.442 of the seven and Kalshi 2.525 — the venues are eight hundredths of a seat apart in aggregate, but Kalshi is the more Republican book in six of the seven races and exactly level in Texas. And Texas alone carries $8.72m, more than Iowa, Kansas and Alaska combined, on a race the market treats as 71.5% decided.

Kansas broke first, and it broke hardest

The national board has been flat. On 22 October a Democratic Senate paid 67.5¢; tonight it pays 70.5¢. Three points in eight days, on a market carrying nearly $13m. Underneath that calm, one of the seven has come apart.

Kansas is the move. Roger Marshall closed at 61.5¢ on 22 October and trades at 47.0¢ tonight — a 14.5-point fall, 4.5 of it in the last 24 hours, and the only one of the seven to cross from favourite to underdog in the window. The polling went the same way: the final Emerson College survey of the race has Democrat Adam Hamilton at 49% against Marshall’s 45%, with a Libertarian on 3%.

Alaska is the counter-trend. Dan Sullivan has gained 5 points over the same eight days while five of the other six fell, which is the single most interesting price on this page after Kansas: the market is pricing the one race it cannot expect to see resolved on the night. Alaska runs a top-four primary and counts the general by ranked choice, so if no candidate clears 50% of first preferences there is no winner until the later rounds are tabulated.

A two-point poll lead is a 70¢ market price

Line the market prices up against the polling averages and the conversion is not smooth. RealClearPolitics has the Democrat ahead in six of the seven, by margins from 0.2 to 3.2 points, and rates all seven Toss Up. The market turns those same margins into prices from 44.5¢ to 76.5¢.

Read that column top to bottom and the market does almost nothing between +0.2 and +0.6, then jumps 25 points between +0.6 and +1.2, then does almost nothing again between +1.2 and +2.6. Alaska at +1.2 and Ohio at +2.6 carry the same price to within half a cent. If you think a polling average is a sufficient statistic for a Senate race, five of these seven prices are wrong; the market plainly does not think it is.

Maine is the one to sit with. The polling average has moved about a point since the start of October — CNN had Jackson +4 in its 20–26 October field, Marist +2 on 22–25 October, Fox tied on 16–20 October — while the Maine board took Susan Collins from 40.0¢ on 26 October to 30.5¢ tonight. Nine and a half points of market move against roughly one point of polling move is not the market reading the polls.

The board nobody quotes: 128 legs on seven races

On 26 October Polymarket opened a market that does not appear in a single Senate tracker we can find. It asks which of the seven states the Republican candidate wins, and it lists every answer: 128 legs, one for each subset of seven states, from None through IA, KS to All 7. It has taken about $447,000 and shows roughly $19.8m of posted order-book depth.

The rules are worth quoting because they are not the single-state rules. A leg resolves Yes only if the Republican wins every state it names and no other state in the market. A state counts as not won by the Republican if anyone else wins it — Democrat, independent or third party — or “if its results cannot be determined by March 31, 2027”. Resolution needs the Associated Press, Fox News and NBC to agree, failing which it goes to official certification.

What the 128 legs say each state is worth

Because every combination is priced, the board carries an opinion about each individual state: add up the 64 legs that contain Texas and you have the board’s implied probability that the Republican wins Texas. Normalise by the book sum so the over-round comes out, and you can put that number next to the single-state board and see whether the two agree. They mostly do not.

Six of the seven read more Republican on the combination board than on the single-state board, Ohio reads five points less, and the sum of the marginals is 2.680 expected Republican wins against 2.442 from the single boards. That is a quarter of a seat of disagreement between two markets on the same exchange, about the same seven elections, open at the same moment.

And then look at the last column, because it is why nobody has closed it. Buying Texas synthetically means lifting 64 separate offers for 73.1¢, against 29¢ for the same exposure on one board. Selling it raises 29.3¢ against a 28¢ bid. Every gap in that table sits inside its own bracket, several times over. The disagreement is real and it is not a trade.

Seven coin flips would make this a 4% event

The interesting thing about the combination board is not the marginals. It is that pricing all 128 outcomes forces a view on how the seven races move together, and that view is emphatic. Take the seven single-state prices and multiply them out as independent events — the standard tracker method — and compare.

Both columns have the same mean, 2.68 against 2.44, and nothing like the same shape. The board’s standard deviation is 1.94 seats against 1.23 under independence. It pays four times as much on a clean Democratic sweep of all seven and sixty-six times as much on a clean Republican one, and it takes the money out of the middle: 2 of 7 and 3 of 7 together are 58.1% if the races are independent and 35.9% on the board.

That is a market saying these seven races share a national error term, which is the one thing every post-mortem of every recent polling miss has said too. It is also, for what it is worth, the only place on Polymarket where that belief is priced rather than assumed.

Four routes to a Democratic Senate, 14 points apart

Now roll it forward. Republicans hold 53 seats against 45 Democrats and two independents who caucus with them, so Democrats need a net gain of four for the 51 that the control board requires. Outside the seven, the market expects exactly one flip: North Carolina pays 97.65¢ for Roy Cooper. Georgia, New Hampshire and Minnesota are all Democratic holds above 92¢. So the swing seven have to supply three net seats — remembering that Michigan is a Democratic seat, so a Republican win there costs one back.

Last night the four books that answer this question directly agreed to within 0.67 points, which we wrote up in the Polymarket vs Kalshi comparison. Tonight they still do. The disagreement has moved one level down: the two ways of building the Senate out of its component races are 14.06 points apart, and the whole of that gap is the correlation assumption. Assume the seven are independent and you are 10.38 points more bullish than the market that trades the question.

The honest reading is that the combination board is the young, thin one — four days old, $447,000 traded, 58 legs without a bid — and the control board is the $12.95m one. But the direction of the disagreement is the part worth keeping. Every tracker that multiplies toss-ups together is sitting 10 points above the deepest market on the subject, and it is not because it likes the Democrats’ chances in Maine.

What would move this before 3 November

Kansas, and then Kansas again. It is the only one of the seven that has moved more than six points in a week, it is the cheapest Democratic pickup still priced as a coin flip, and it is the seat whose fall would make the seat ladder — which already pays 48.29% on Republicans holding 47 or fewer — look conservative rather than aggressive.

After that, watch whether the combination board’s spreads tighten. Fifty-eight dead legs on a market with $19.8m of posted depth is a book that has not been arbitraged against its own single-state boards, and the first serious participant to do that work will move seven prices at once. If the 128-leg marginals converge on the single boards in the next three days, the 66.82% converges on 70.50% and this entire disagreement quietly closes.

And if you take one number from this page, take the one that is not a probability. The seven races on this board have a combined Polymarket volume of $22.88m. The market that prices them jointly, and is the only thing on the exchange with an explicit view on how they move together, has taken $447,000 — about 2% as much. The most important assumption in every Senate forecast published tonight is also the least traded.

Related analysis

• Polymarket vs Kalshi 2026 midterms — two rulebooks, one price
• 2026 midterms odds: the Senate flipped in 30 days
• 2026 Senate odds and oil — five conditional boards, five prices
• 2026 House odds: markets 90¢, models 76-93%
• 2026 governor race odds: 36 boards vs one ladder
• 2026 midterms hub — every Senate and governor race we track

Sources: the Polymarket Gamma and CLOB public APIs (events 1262243, 32224, 106193, 769205, 57664 and the seven state boards 59233, 57647, 57648, 57650, 57658, 57682 and 57672) and the Kalshi public markets API (SENATEAK/IA/KS/ME/MI/TX-26 and SENATEOHS-26, plus CONTROLS-2026-D), all read at 30 October 2026, 21:14 UTC. The combination board’s resolution terms are quoted from the seven-state board itself and Kalshi’s from the SENATEKS-26-R contract, in both cases the operative text the venue publishes rather than a summary. The Kansas polling figures are Emerson College’s final Kansas survey; the RealClearPolitics averages are as republished on 29 October 2026. The 53-45-2 going-in split, the open seats and the nominees are as described here. Kalshi volume and open interest are contract counts; Polymarket volume and liquidity are dollars, and the two are never added on this page. Prices here are a frozen snapshot; the live numbers are on the market pages, which refresh every 10 minutes. Nothing here is financial or betting advice.