The Quick Answer
Democrats are favored to win the House in 2026. Kalshi's seat-count board gives them about an 81 percent chance of a majority, with the most likely finish near 231 seats. A five-model AI panel that never saw a price agrees, putting a majority near 90 percent and its center near 229 seats.
Both stop well short of a wave. The 2026 redistricting maps tilt the terrain a little toward Republicans, and that trims the usual midterm swing against the president's party. The one real split is at the extremes: traders leave real money on both a Republican hold and a Democratic blowout, and the panel prices both near the floor. The full board, and exactly where the models fade the market, is below.
Stay ahead of the markets.
Daily insights and expert picks on Kalshi, Polymarket, and what's moving markets.
Free forever. Unsubscribe anytime.
Prices as of August 20, 2026 · by Jake Hari
The Market
| Venue | Kalshi |
| The Event | "How many House seats will the Democratic Party win in the 120th Congress?" (ticker KXDHOUSESEATS-27), 12 mutually-exclusive seat-count bands |
| The Contract | Each band pays out if Democrats hold a seat total inside that range on February 1, 2027, per Kalshi's settlement rules |
| Settlement Date | February 1, 2027, the date named in Kalshi's rules for counting Democratic seats (the 120th Congress convenes January 3, 2027). Election Day is November 3, 2026 |
| How The Bands Add Up | About 105 cents read raw, the usual few points of exchange spread. Scaled to 100, the majority bands from "218-221" up carry about 81 cents; the three bands below 218 carry the other 19 |
| Prices Below As Of | August 20, 2026 |
Exactly one band pays. A 224-seat finish settles only "222-225" YES and the other 11 NO; a 233-seat night pays "230-233" instead.
The Board
Sorted by seat total. The market column is Kalshi's daily close as of August 20, 2026. The blend is the median of five price-blind models after one revision round, so the 12 blend figures need not add to exactly 100. 218 is the majority line. Our full Model Verdict panel seats eight models, but three outside models were unavailable for this run, so this board was scored by five: Claude Fable, Claude Opus, Claude Sonnet, GLM and DeepSeek.
| Seats (Democrats) | What it would mean | Market | AI blend |
|---|---|---|---|
| Below 210 | GOP holds and gains | 8.6¢ | 3% |
| 210-213 | Republicans keep the House | 5.0¢ | 3% |
| 214-217 | Republicans keep it, barely | 6.0¢ | 5% |
| 218-221 | Narrowest Democratic majority | 7.0¢ | 10% |
| 222-225 | Comfortable narrow majority | 10.5¢ | 12% |
| 226-229 | Solid majority (net +11 to +14) | 11.5¢ | 13% |
| 230-233 | Solid majority (net +15 to +18) | 12.5¢ | 15% |
| 234-237 | Wave-sized majority | 11.5¢ | 14% |
| 238-241 | Big wave | 9.4¢ | 9% |
| 242-245 | Very big wave | 7.5¢ | 6% |
| 246-249 | Near the modern ceiling | 5.2¢ | 4% |
| Above 249 | Historic blowout | 10.5¢ | 4% |
More live boards from the same panel: Republican House Seats 2026, the mirror market, where the crowd gives Republicans about a 20 percent shot at holding · 2026 House Control, the binary who-wins-the-House companion, where Democrats trade at 84¢ · 2026 Senate Control, the other chamber, which prices close to a coin flip. Prices fetched August 20, 2026.
Every seat on this panel is graded against real market settlements. Records to date: Claude Fable 84% on 508 graded calls · Claude Opus 85% on 578 graded calls · Claude Sonnet 84% on 556 graded calls · GLM 81% on 2,367 graded calls · DeepSeek 80% on 2,392 graded calls. Recomputed daily; the full scoreboard is public.
Model estimates generated August 20, 2026, price-blind, from a five-model panel. These are model estimates, not predictions of fact and not financial or trading advice. Models are frequently wrong; the market price reflects real traders' money. Every number here gets graded in public once the market settles, on our full model-verdict scoreboard. Kalshi is a CFTC-regulated exchange; 18+, availability varies by state.
Why Isn't This A Wave?
Either center is about 14 seats better than the 215 Democrats won in 2024. The postwar base rate says the president's party loses about 26 House seats in a midterm, and the conditions for that kind of loss are in place. The generic-ballot average sat near D+6.6 across six public aggregators as of August 17, and presidential approval at net −21.7.
What holds the number down is the map. Six states drew new congressional lines for 2026. Texas's new Republican map, which the Supreme Court cleared for this cycle, is worth about five GOP seats, and California's voter-approved Prop 50 map answers with about five Democratic seats. Across Texas, Missouri, North Carolina and Ohio, Republicans count about nine seats the new lines put in reach. The net is a modest Republican edge, with fewer competitive districts than 2018 for a national wind to move.
Every model named the map on its own. DeepSeek put it plainly, writing that a D+6-to-7 national margin "translates to roughly a 25-30 seat gain on neutral maps, but I shade that down for two structural factors: the mid-decade redistricting round leaves Republicans a modest net edge... and the starting point of 212 means fewer low-hanging pickups than in 2018."
Where The Panel Fades The Market
The panel is more concentrated than the crowd in the moderate-majority core. It pays 10 to 15 percent for each band from 218 to 237, several points above the market, and far less at the two ends of the board.
- Above 249 Seats: Market 10.5¢, Panel 4 Percent. The widest gap on the page. Better than ten cents says there is a real chance Democrats win a blowout of 250-plus seats. The panel calls a wave that large a low-single-digit outcome: the president is underwater, but there is no collapse event on the card, and the biggest states run protective Republican maps.
- Below 210 Seats: Market 8.6¢, Panel About 3 Percent. The mirror image. Nearly nine cents on Republicans holding the House and gaining seats, which would need the polls to miss by more than any Trump-era midterm has. The two midterms without Trump on the ballot were nearly unbiased, so the panel treats a miss that large as a distant tail.
Those two open-ended bands are also the busiest contracts on the board, each with more than 200,000 contracts traded, roughly triple the bands in the likely middle. That is the shape of a catch-all bucket: cheap long shots on a blowout draw casual money, and the price fattens. The panel keeps its probability in the middle, where the seats are.
Could Republicans Still Hold The House?
Any total below 218 keeps the House Republican. The market prices that near 19 percent and the panel near 10. The only sub-majority band any model kept above a rounding error is 214-217, at a 5 percent blend. The path there is specific: the redistricting net lands at the Republican-tilt end of the range, Democrats drop a couple of their own competitive seats on defense, and a normal polling miss leaves them a hair short of 218. It takes no Republican wave, only a map that fully offsets a favorable national environment.
What Changed When The Models Read Each Other?
Each model priced the field once, then read the others' anonymized reasoning and could revise. The round tightened the tails and pulled the two outliers toward the middle without moving the center.
The biggest move came from Opus, the most cautious model in round one. It had priced a Democratic majority at 71 percent and raised that to 92:
Democrats start only 6 short and the environment is a clear wave, so losing the chamber requires a 4+ point collapse or a polling miss the last two midterms didn't produce. — Claude Opus
DeepSeek moved the other way for the same discipline, trimming from 99 percent to 83 after peers pressed that the map and the vote-to-seats conversion carry real, two-sided error. Fable held the middle it started on:
Forecaster A's 99% majority treats the map as riskless; Forecaster C's 221 center under-credits the 2018 translation. Neither moved me off the middle. — Claude Fable
What Moves This Board Before November?
- A redistricting ruling. Several 2026 maps still face live challenges. A ruling that unwinds a challenged Republican map pushes the center up; one that reinstates a blocked Republican map or strikes California's counter pushes it down. This is the largest swing factor on the board.
- The generic ballot leaving D+6. A drift to D+8 or better makes the 234-plus bands live. That is the panel's wave path, its second-heaviest band at 14 percent for 234-237, with GLM highest at 16. A tightening back toward D+3 revives the Republican-hold path.
- A batch of new race ratings. Republican-held seats sliding toward Toss-up or Lean Democratic at the Cook Political Report or Sabato's Crystal Ball, the direction ratings ran through mid-August, raises the center. Democratic-held seats sliding the other way lowers it.
More on this: Republican House Seats 2026: How Many Will The GOP Hold? · Kalshi 2026 House Odds: Which Party Wins The House? · 2026 House Popular Vote Odds: The AI Calls Kalshi Too Blue · TX-15 House 2026 Odds: De La Cruz Vs Pulido On Kalshi
The Bottom Line
Between now and Election Day, the courtrooms matter more than the polls: one ruling on a contested 2026 map can move this board further than a month of surveys. Every number here gets graded on the public record once the seats are counted in early 2027.
To be explicit: these are model estimates for information, not financial advice, and not a recommendation to buy or sell any contract. The models are graded against real settlements and are wrong regularly. Event contracts are for adults 18+ where legal; Kalshi is a CFTC-regulated exchange and availability varies by state. Political markets are covered here as neutral probability questions, with no endorsement of any party, candidate or outcome.
Prices on this page are Kalshi's book. If you also trade on Polymarket, code OS4 gets new users a $50 trading bonus on a $10 deposit — affiliate link; terms as stated by Polymarket; 18+, availability varies by state.
