Eleven states redrew their congressional district lines between August 2025 and June 2026. Ten of those maps survived the courts, and the running total across them is a net Republican gain of 10 seats. Democrats need three. That arithmetic sits underneath every number on this page, and it is the biggest structural difference between the chamber elected in 2024 and the one on the ballot in November.
Kalshi's House control market has barely blinked at it. The exchange still prices Democratic control as a heavy favorite, on 26 million contracts traded, and the price has closed between 84 and 86 cents on every session for more than two weeks. We put the same question to a panel of AI models with no access to any price, gave them the enacted map ledger alongside the national polling, and asked each one to convert a national vote margin into a seat count. They came back well below the market, and they agreed on why.
This is a data-driven analysis of market prices, polling, and the public record. It does not advocate for any party, candidate, or outcome.
The Quick Answer
Kalshi prices Democratic House control at 84.5 cents and Republican control at 15.5 cents. Our price-blind panel lands at 70% Democratic after a revision round, with every model between 67 and 76. The panel is not arguing with the direction. It is arguing with the conversion: how much of a 6-point national polling lead the redrawn map absorbs before it turns into seats. The full ledger, the board, and the exchange's own quiet disagreement with itself are below.
Kalshi event contracts trade on a CFTC-regulated exchange; you must be 18 or older and in an eligible state to participate.
The Market At A Glance
| Venue | Kalshi |
| The Event | "Which party will win the U.S. House?" (Kalshi CONTROLH-2026), two mutually exclusive party contracts |
| The Contract | Per the settlement rules, verbatim: "If the Democratic Party has won control of the House in 2026, then the market resolves to Yes" (mirror wording for the Republican contract) |
| How It Settles | Two paths, and the second one is the fallback. Verbatim from the contract's secondary rules: "This market may be determined early based on a consensus of media calls projecting control of the U.S. House. See full rules for details. Otherwise, victory will be determined by the party identification of the Speaker of the House on February 1, 2027." So an election-night call consensus can settle this early; if the calls do not converge, the Speaker's party on February 1, 2027 decides it |
| Key Dates | Election day November 3, 2026; the new House organizes in January 2027; expected expiration February 1, 2027 (15:00 UTC) |
| Volume | About 26.0 million contracts across the two sides. That is the heaviest 2026 election market on the exchange, more than three times Senate control at 7.7 million. The Republican contract alone took roughly 800,000 contracts in the last 24 hours |
| Prices Below As Of | August 26, 2026, 10:05 p.m. ET (midpoints between the best buy and sell prices) |
Prices are re-read from the exchange every time this page is re-scored, and the stamp above carries the date of the most recent read.
Hottest Prediction Markets Right Now
- 2028 Democratic presidential nominee · $194M traded
- 2027 Pro Football Champion · $72M traded
- 2028 U.S. Presidential Election winner? · $62M traded
- Pro Baseball Champion · $61M traded
- 2028 Republican presidential nominee · $60M traded
- Bitcoin price at the end of 2026 · $33M traded
Every market above links to our full AI model verdict; browse them all on the OddsShopper prediction markets hub, and see how every settled call actually scored on the full graded scoreboard.
The Map Moved Ten Seats Before Anyone Voted
Congressional maps are normally drawn once a decade, after the census. The 2025-2026 cycle broke that habit in 11 states, ten of which now vote on new lines, and the fight is a matter of public record rather than interpretation.
It started in Texas. On July 9, 2025, Governor Greg Abbott called a special legislative session on redistricting, citing a letter from the U.S. Department of Justice that claimed four of the state's majority-minority districts were unconstitutional. The Texas Tribune reported that Texas officials had spent years arguing their maps were race-blind until the DOJ letter arrived. President Trump publicly called for five additional Republican seats out of the state. The legislature delivered a map targeting five Democratic-held seats, and Abbott signed it on August 29, 2025.
Lawsuits followed within days, with plaintiffs alleging the new lines were a racial gerrymander drawn to dilute minority voting power. On November 18, 2025, a federal three-judge panel in El Paso agreed and blocked the map. The Supreme Court stayed that ruling 6-3 on December 4, 2025, allowing the new lines for 2026, and on April 27, 2026 the Court formally reversed the lower court, clearing the map for use. Texas votes on the 2025 lines in November.
California responded with a redraw of its own. Proposition 50, a ballot measure redrawing the state's congressional map, passed on November 4, 2025 with 64.4% of the vote and is worth roughly five Democratic seats. Virginia tried the same play and failed: a Democratic-drawn map worth about four seats was blocked in state court, and the appeal died at the Virginia Supreme Court in May 2026.
Here is the full enacted ledger, with the date each map became law:
| Map Enacted | State | Advantage | Status |
|---|---|---|---|
| August 29, 2025 | Texas | R +5 | In effect |
| September 28, 2025 | Missouri | R +1 | In effect |
| October 22, 2025 | North Carolina | R +1 | In effect |
| October 31, 2025 | Ohio | R +2 | In effect |
| November 4, 2025 | California | D +5 | In effect |
| November 10, 2025 | Utah | D +1 | In effect |
| April 21, 2026 | Virginia | D +4 | Blocked in court |
| May 4, 2026 | Florida | R +4 | In effect |
| May 7, 2026 | Tennessee | R +1 | In effect |
| May 29, 2026 | Louisiana | R +1 | In effect |
| June 2, 2026 | Alabama | R +1 | In effect |
Counting only the maps in effect, Republicans gained 16 seats of drawing advantage and Democrats 6, for a published net of Republican +10. Source: Wikipedia's 2025-2026 redistricting ledger, which is where the running total above is maintained.
What that net does is worth stating carefully, because it is easy to overclaim in either direction. It does not hand anyone the chamber. A drawn seat is an expected seat under a neutral national environment, and November 2026 is not shaping up as a neutral environment. What it does is move the bar: the national vote margin Democrats need to reach 218 seats is higher on these lines than it was on the ones that produced the current House.
What Actually Decides House Control
The panel's data card carried no market prices and no punditry. Five blocks did the work, and every fact in them was pulled from a live source at the moment the panel ran.
- The Chamber Is One Bad Night From Flipping. The Clerk of the House currently counts 218 Republicans, 212 Democrats, one independent and four vacant seats. Every seat is on the ballot in November, so the baseline that matters for the election is the 2024 result: Republicans won 220 seats to Democrats' 215. A majority is 218, which puts the Democratic requirement at a net gain of three. That is the narrowest starting margin in modern practice, and it is why the map net matters so much. Ten is a large number next to three.
- The Base Rate, And Its Two Exceptions. In 20 postwar midterms, the president's party lost House seats in 18, averaging a loss of 25.6 seats, with a worst case of minus 63 in 2010 and a best of plus 8. The only two gain years were 1998, the impeachment-backlash midterm, and 2002, the post-9/11 midterm fought with presidential approval above 60. The card asked every model to test whether either cause is present now rather than citing the exceptions as generic caution. Every model ran the test and reported the same answer: neither condition applies, so the exceptions carry almost no weight this cycle.
- The National Environment. The generic congressional ballot averages Democrats 47.8 to Republicans 41.5, a D+6.3 lead, across six public aggregators updated August 20, 2026, with unusually tight agreement between them (the individual readings run from D+5.9 to D+6.6). Presidential approval averages 37.5 approve to 59.2 disapprove across ten aggregators updated August 21, 2026, a net of minus 21.7. The historical record links approval that far under water to midterm losses worse than the 25.6-seat average.
- The Special-Election Thermometer. Across 42 contested 2026 state legislative special elections, Democrats are running a median of 10.4 points ahead of their 2024 presidential baseline, and 12.6 points ahead weighted by turnout, per MultiState's swing tracker. Read that with the limits the source itself names: these are state legislative races, not congressional ones, turnout runs roughly 10 to 20 percent of a November electorate, and the 2018 precedent is that an early swing this size moderates as the general approaches. It is a thermometer, not a forecast.
- The Retirement Wave. Sixty representatives and two non-voting delegates have announced they are not seeking re-election, 25 Democrats and 37 Republicans. That is the second-most retirements in a single cycle in U.S. history, behind only 1992. Thirty-one of them are leaving to run for another office. Open seats swing harder than incumbent-held ones, and a lopsided retirement count is also a read on how sitting members judge their own environment.
The card deliberately carried no district-by-district ratings and no district-level polling, and it told every seat so. That constraint shows up honestly in the panel's reasoning below.
The Board: Market Vs. The Panel
| Wins The House | Market (mid) | AI Blend | Gap |
|---|---|---|---|
| Democratic Party | 84.5¢ | 70.0% | −14.5 |
| Republican Party | 15.5¢ | 30.0% | +14.5 |
The blend is the median of the models' final numbers after a revision round in which each model read the others' anonymized reasoning. Seven of the panel's nine models ran this board; two were unavailable when it ran.
Context for a gap this size, because it needs it: on the panel's own graded history, the models are the worse forecaster when they disagree with a price. Across two independent measurements, the panel has won only about a third of its disagreements of 10 cents or more. The 14.5-point gap in the table above is the panel's public reasoning and its public record, not a recommendation to trade toward it.
Every Model's Number
| Model | P(Democratic control), round 1 | Final (after revision) | Move |
|---|---|---|---|
| Claude Fable | 79.0 | 67.0 | −12.0 |
| Claude Opus | 79.0 | 76.0 | −3.0 |
| Claude Sonnet | 63.0 | 67.0 | +4.0 |
| ChatGPT | 64.0 | 70.0 | +6.0 |
| Gemini Pro | 65.0 | 72.0 | +7.0 |
| Kimi | 70.0 | 68.0 | −2.0 |
| Grok | 58.0 | 71.0 | +13.0 |
Every seat on this panel is graded against real market settlements — records to date: Claude Fable 83% on 779 graded calls · Claude Opus 84% on 849 graded calls · Claude Sonnet 82% on 824 graded calls · GPT 84% on 6,310 graded calls · Gemini 86% on 5,009 graded calls · Kimi 82% on 2,550 graded calls · Grok 84% on 4,307 graded calls. Recomputed daily; the full scoreboard is public. Every number in this piece gets graded against the real settlement once this market resolves.
Stay ahead of the markets.
Daily insights and expert picks on Kalshi, Polymarket, and what's moving markets.
Free forever. Unsubscribe anytime.
More live boards from the same panel: How many House seats Democrats win, where the 230-233 band leads the ladder at 12.5¢ · 2026 Senate control, priced near a coin flip with Democrats at 47.5¢ · 2028 House control, already trading at 67.5¢ for Democrats. Prices fetched August 26, 2026.
What makes the panel's number interesting is not its size but its unanimity of mechanism. Seven models, running independently and blind to every price, all built the same chain: the base rate applies at full weight because neither exception condition is present, the generic ballot can be taken close to face value because midterm polling error has run near zero (minus 0.3 in 2018, plus 0.9 in 2022), and then the map eats the first two to three and a half points of that lead before it starts producing seats. What is left is a cushion of roughly two to three points against a combined uncertainty the models put between three and four and a half. That is a favorite, and a nervous one.
Two models pushed back on the map's own arithmetic in ways worth quoting. Claude Opus argued the redraw is partly self-limiting: "an R+16 redraw that manufactures 16 seats does so by spreading Republican voters thinner, so in a D+6 environment the map's realized absorbency is below its neutral-environment rating. The card carries no district ratings, so I flag that as structural inference, not fact." — Claude Opus. Kimi named the same risk from the other side and declined to resolve it: "the mid-decade maps have no midterm track record — they could underperform via dummymander risk on 2024 low-propensity coalitions, or overperform via efficient packing." — Kimi
A dummymander, for readers meeting the word for the first time, is a gerrymander that backfires: a party spreads its own voters across more districts to win more seats, and then a bad national year sinks all of them at once. It is the specific failure mode that makes a freshly aggressive map risky for the side that drew it.
The Exchange Disagrees With Itself
Kalshi lists three separate ways to price this same question, and they do not all agree.
The headline two-contract market says 84.5 cents. But the exchange also runs two seat-count ladders, one on how many seats Democrats win and one on how many Republicans win, Each of those can be read as a control probability by adding up the bands on the winning side of 218 and normalizing for the fact that every band trades in its own order book. On the Republican ladder that produces a Republican-control figure, 19.4%, which subtracts from 100 to give the Democratic number in the table.
| Kalshi Surface | Implied Democratic control | What is being read |
|---|---|---|
Two-contract topline (CONTROLH-2026) | 84.5% | Two party contracts |
Democratic seat ladder (KXDHOUSESEATS-27) | 80.6% | Twelve seat bands |
Republican seat ladder (KXRHOUSESEATS-27) | 80.6% | Eleven seat bands |
Ladder figures are the sum of the control-side bands divided by the ladder's total, which is how a book that sums to 104.3 and 105.4 cents respectively gets converted back to a probability.
Read independently, on separate order books with different band structures, the two ladders land on the same number to the tenth of a point, and both sit about four points below the topline. Read as distributions, they also agree on where the seat count lands: the Democratic ladder's median falls in the 230-233 band, and the Republican ladder's median falls in the 203-207 band. Those describe nearly the same chamber from opposite ends, since 230 to 233 Democratic seats leaves Republicans 202 to 205.
That is not proof the topline is wrong. Normalizing a thin ladder is an approximation, and the two-contract market carries far more volume than either ladder. But it means the panel's direction of travel is not an outside opinion. The exchange's own more granular pricing is already sitting below its own headline, and the panel is sitting below that. Our Republican House seats board scores that ladder rung by rung.
Where The Panel Changed Its Mind
The revision round compressed the panel from a 21-point spread to 9 points, and the two biggest movers went in opposite directions, which is the useful part.
Grok moved up 13 points, from 58 to 71, after reading the others and concluding it had been double-counting the map's drag on the seats-neutral margin, meaning the national vote lead Democrats would need simply to break even on seats: "Others' 63-65 cluster overweights conversion opacity; the 79s underweight that seats-neutral may sit near D+4-5 with no district ratings on the card. Dominant uncertainty is vote-to-seats on brand-new lines, narrowly ahead of ten-week environment drift." — Grok
Claude Fable moved 12 points the other way, from 79 to 67, and was explicit about which line of argument did it. Its cushion is the margin by which it expects Democrats to clear the raised bar, and its sigma is the standard deviation around that estimate, meaning how wrong it expects to be: "A ~3-point cushion over a ~4.5-point sigma prices near 70; I shave to 67 because the fresh gerrymanders could outperform my absorption estimate and the specials swing historically fades toward November." — Claude Fable
Claude Opus barely moved, from 79 to 76, and held the high end on a reading of how fast the national environment can actually change: "approval this deep rarely recovers 15 points in ten weeks, but ten fresh maps have never been run." — Claude Opus
Three models reading identical data, converging on an identical mechanism, and still landing nine points apart is a fair description of what this market is actually uncertain about.
What Would Change The Panel's Mind
Every model was asked to name the checkable things that would move its number. Five came back repeatedly, and they are what this page gets re-scored on.
- The Generic Ballot Falling Below D+4, And Especially Below D+3. The models put the map's absorption band at 2 to 3.5 points, so a fade to D+3 erases the cushion entirely. Pushes Republican.
- Presidential Approval Recovering Past Net Minus 12, Or Any Rally-Scale National Event. Net minus 21.7 is what shuts off the 1998 and 2002 exception cases. Recovery re-opens them and weakens the base rate directly. Pushes Republican.
- The Autumn Polling Wave, From September 1 Onward. Midterm polling volume multiplies after Labor Day, and September specials are the first real test of whether the D+10.4 swing is holding or fading on the 2018 schedule. Direction depends on what it shows.
- Any Further Court Action On An Enacted Map. Reinstating Virginia's blocked map, or striking one of the large Republican redraws, moves the bar by one to two points in the corresponding direction. Pushes Democratic if it lands that way.
- District-Level Ratings On The Redrawn Seats. The card carries none, and several models named that gap as the largest single source of their uncertainty. Ratings showing the new Texas, Florida and Ohio seats holding comfortably would push Republican; the same seats drifting toward toss-up would push Democratic.
Settlement Timeline
| Date | What happens |
|---|---|
| September 1, 2026 Onward | Midterm polling volume multiplies; the autumn special-election calendar resumes |
| November 3, 2026 | Election day, all 435 seats |
| Election Night, November 3-4, 2026 | If the media calls converge on a projected House majority, the contract can be determined early on that consensus |
| January 2027 | The 120th Congress convenes and elects a Speaker |
| February 1, 2027, 15:00 UTC | The fallback and the expected expiration: control decided by the Speaker's party identification on this date |
This page is re-scored when the story moves, with the as-of stamp above carrying the date of the most recent read.
How To Read A Two-Contract Political Market
An 84.5-cent YES price is not a settled fact wearing a price tag. It implies the market thinks the other outcome happens about one cycle in six from this position. The two sides here sum to 100 cents, so unlike a 33-name field there is no big re-scaling step, but the gap between the best price to buy and the best price to sell still matters: buyers of the favorite pay the higher of the two, and on a months-long hold that gap plus the time your capital is tied up is the real cost of the position.
Reading the topline against the seat ladders, the way this page does above, is simply reading the same event priced on adjacent shelves. Where those shelves disagree, the disagreement is information about how confident the market really is, not an instruction to do anything about it.
If the mechanics are new to you, our guide to politics markets for sports bettors walks through how an event contract actually settles and what it costs to hold one.
The honest frictions: you must be 18 or older and in a state where Kalshi operates, availability varies and changes, the contract can settle early if the media calls converge on election night but otherwise holds capital until the Speaker rule decides it on February 1, 2027, and settlement is party control of the chamber rather than the winner of any individual race.
Questions Readers Actually Ask
When does the 2026 House control market settle? Possibly on election night, and at the latest on February 1, 2027. The elections are November 3, 2026. The contract's own secondary rules allow it to be determined early on a consensus of media calls projecting control of the House; if that consensus does not arrive, control is decided by the party identification of the Speaker of the House on February 1, 2027, which is also the contract's expected expiration (15:00 UTC).
How many seats do Democrats need to win the House in 2026? It depends which baseline you count from, and the two answers differ. Every seat is on the ballot in November, so the baseline that governs the election is the 2024 result: Republicans won 220 seats to Democrats' 215, and 218 is a majority, which puts the Democratic requirement at a net gain of three. Counting instead from the chamber as it sits today, the Clerk of the House records 218 Republicans, 212 Democrats, one independent and four vacancies, which would make it six. Any figure quoted without its baseline is incomplete.
Does the 2025-2026 redistricting decide the 2026 House? It moves the bar rather than settling the question. Eleven states redrew congressional lines between August 2025 and June 2026, and ten of those maps are in effect; the running total across them is a net Republican gain of 10 seats. A drawn seat is an expected seat under a neutral national environment, and the current generic ballot is D+6.3 rather than neutral. What the redraw does is raise the national vote margin Democrats need before their lead converts into seats, which is exactly the point where the panel and the market disagree.
More on this: The 72-Year Streak Traders Are Betting Against · Who Will Win the House in 2026? Where the Big Money Went · Michigan House Races 2026 Odds: The Nominee Who Quit And Won · 2026 House Popular Vote Odds: The AI Calls Kalshi Too Blue · TX-15 Odds: Pulido And De La Cruz Trade The Lead
The Bottom Line
The market and the panel agree that Democrats are favored to win the House in 2026, and they disagree about how comfortably. Kalshi's headline contract says 84.5 cents. The exchange's own two seat ladders both say 80.6%. Seven price-blind models, handed the enacted map ledger alongside the polling, say 70%.
The reason all three numbers point the same way is that the disagreement is not about the political environment. Nobody on this page disputes a D+6.3 generic ballot or a president 21 points under water. The disagreement is entirely about conversion: how many of those points the redrawn map absorbs before they become seats, on lines that have never been run in a midterm electorate. The panel's answer is two to three and a half points, which leaves a real but thin cushion. The market's answer is implicitly less.
The one number worth checking weekly between now and November 3 is still the generic ballot, and the threshold to watch is D+4. Below it, the panel's own reasoning says the cushion is gone. When it moves, or when the autumn polling wave lands, this page gets re-scored on the same URL and graded in public either way.
