Kalshi Democratic House Seats 2026: How Many Will Dems Win?

The Quick Answer

Kalshi does not only price whether Democrats take back the House in 2026. It prices how many seats they win, in twelve numbered bands running from "below 210" to "above 249" (each band is a range of seat totals, and exactly one of them will come true). Put the prices together and the board gives Democrats about an 81 percent shot at a majority, with the most likely finish clustered near 231 seats. A five-model AI panel, working without ever seeing a price, agrees on the direction and pushes the confidence higher, to about 90 percent, while centering the outcome a touch lower, near 229 seats, because it weighs the 2026 redistricting maps harder than the crowd does. Where the two divide hardest is at the edges: real money sits on both a Republican hold and a Democratic blowout, and the panel prices both close to the floor. The full twelve-band board, and exactly where the models fade the market, is below.

Free: The Weekly PM Market Brief — the 8-model panel's graded record, the week's biggest market-vs-model gaps, and what's spiking next. One email, Sundays.

Updated August 20, 2026 · 14 min read · by Jake Hari

New to Kalshi event markets? Start here for how these CFTC-regulated contracts settle before you read the board.

The Market

Venue	Kalshi, a CFTC-regulated event-contract exchange (18+; availability varies by state, as of August 2026)
The event	"How many House seats will the Democratic Party win in the 120th Congress?" (ticker KXDHOUSESEATS-27), twelve 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
Prices below as of	August 20, 2026

The twelve bands cover every possible outcome, so exactly one of them pays. If Democrats finish with exactly 224 seats, only the "222-225" band settles YES and the other eleven settle NO; a 233-seat night pays "230-233" instead. Read as raw prices the bands add up to about 105 cents, the usual few points of exchange spread. Scaled back to 100, the bands at "218-221" and above, where Democrats have a majority, carry about 81 cents between them, and the three bands below 218, where Republicans keep the House, carry the other 19. Those prices come from Kalshi's daily closing quotes, because the board's live buy and sell orders sit empty between trades, and this page refreshes them as the news moves.

What The Panel Was Given

Every model saw the same deterministically fetched data card and never saw a market price. The card led with the national environment: a generic-ballot average (the poll that asks which party's House candidate you would vote for) of about D+6.6 across six public aggregators, meaning Democrats lead by 6.6 points, as of August 17, with individual polls since ranging from D+5 to a D+11 Emerson reading; a presidential approval average of net −21.7 (approval minus disapproval); and the postwar base rate that the president's party has lost House seats in 18 of the last 20 midterms, an average of about 26, with only the 1998 impeachment backlash and the 2002 post-9/11 rally breaking the pattern.

Then it carried the chamber math a seat count turns on. In 2024 Democrats won 215 House seats to Republicans' 220, and a run of vacancies has since left the live chamber near 212 Democrats to 219 Republicans. The majority line is 218, so measured from the 215 they won last time, Democrats need a net of about three seats to take control while holding everything they already have. And it carried the wildcard a normal midterm does not have: the mid-decade redistricting fight, a documented, court-tested battle over the maps themselves. Six states drew new congressional lines for 2026 (California, Missouri, North Carolina, Ohio, Texas and Utah). Texas will run under its new Republican map, worth about five GOP seats, after a federal court struck it as an illegal racial gerrymander on November 18, 2025 and the Supreme Court stayed that ruling 6 to 3 on December 4, 2025 and formalized the reversal on April 27, 2026. California's voter-approved Prop 50 map answers it with about five Democratic seats and roughly cancels Texas on its own. Across Texas, Missouri, North Carolina and Ohio, Republicans count about nine seats they think the new lines put in reach; Indiana's legislature failed to pass a map in December 2025. The net is a modest Republican structural edge, and it is the single reason a D+6 environment may not convert into the 26-seat wave the base rate implies.

A disclosure on the panel: our full Model Verdict panel seats eight models. Three of the outside models were unavailable for this run, so this board was scored by five: Claude Fable, Claude Opus, Claude Sonnet, GLM and DeepSeek. Each priced the whole twelve-band distribution in a single pass, then read the others' anonymized reasoning and revised once. The numbers below are that second-round blend, reported as the band-by-band median across the five models.

The Board

Sorted by seat total. The market column is Kalshi's latest daily close as of August 20, 2026; the blend is the median of the five price-blind models after their revision round, so the twelve blend figures need not add to exactly 100. 218 is the majority line.

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 only about a 20 percent shot at holding and, exactly like this board, fattens its open-ended tail (the "below 193 seats" bucket draws the most money at roughly 18¢) · 2026 House Control, the binary who-wins-the-House companion, where Democrats trade at 84¢ · 2026 Senate Control, the other chamber, which prices as close to a coin flip. Prices fetched August 20, 2026.

The Big Agreement: Democrats Are Favored, But Not For A Wave

Start with what the crowd and the machines share, because it is the headline. Both make Democrats clear favorites to retake the House, the market at about 81 percent for any total of 218 or more, the panel near 90, and both put the center of the outcome well short of a classic midterm blowout. The market's prices imply an average near 231 Democratic seats; the panel's blend lands around 229. Either way that is roughly 14 seats better than the 215 Democrats won in 2024, not the 26-seat swing the raw postwar average would predict.

The reason both sides shave the number is the same, and every model named it independently. The national wind is real: an underwater president and a six-point generic-ballot edge are exactly the conditions that cost a governing party its House majority. But the map has fewer places for that wind to move seats than it did in 2018, and Texas's new lines against California's voter-approved counter leave the terrain tilted a hair toward Republicans. DeepSeek put the mechanism 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."

So the consensus is a majority, not a mandate. Democrats most likely land somewhere in the 222-to-237 corridor, enough to run the chamber, short of the numbers that would let them govern without their own narrowest members.

Where The Panel Fades The Market: The Blowout Tails

Now the disagreement, and it is a clean one. The panel is more concentrated than the market in the moderate-majority core. It pays 10 to 15 percent for each band from 218 to 237, several points above the crowd, and it is far less generous to the extremes.

Look at the two ends of the board:

• Above 249 seats: market 10.5¢, panel 4 percent. This is the widest gap on the page. Better than ten cents says there is a real chance Democrats win a blowout of 250-plus seats, a landslide-sized House. The panel says that with the president underwater but no collapse event on the card, and with protective Republican maps in the biggest states, a wave that large is a low-single-digit outcome, not a one-in-ten shot.
• Below 210 seats: market 8.6¢, panel about 3 percent. The mirror image. Nearly nine cents on Republicans not just holding the House but gaining seats, which would require the polling to be wrong by more than the Trump-era midterm error has ever run. The card's own poll-error history shows the two midterms without Trump on the ballot were nearly unbiased, so the panel treats a miss that large as a distant tail, not a live scenario.

Here is the tell that this is a pattern and not noise. On Kalshi, those two open-ended tail bands are the highest-volume contracts on the entire board (each has drawn more than 200,000 in traded volume, roughly triple the bands in the likely middle). More money sits in "below 210" and "above 249" than in the ranges the panel thinks actually decide the seat count. That is the classic shape of a catch-all bucket: the open ends are where casual money piles onto a cheap long-shot on a blowout, and the price fattens accordingly. The panel, pricing from the distribution rather than from the drama, keeps its probability in the middle where the seats are. Reasonable traders can weigh those tails differently. The point is that the crowd's money and the models' math diverge most exactly where the outcome is most extreme.

Every Model's Number

The full per-model board. Each model priced all twelve bands as one distribution; cells are rounded for display.

Seats (Democrats)	Market	AI blend	Claude Fable	Claude Opus	Claude Sonnet	GLM	DeepSeek
Below 210	8.6¢	3%	4	1	3	1	5
210-213	5.0¢	3%	5	2	3	2	5
214-217	6.0¢	5%	7	5	4	4	7
218-221	7.0¢	10%	10	10	7	6	10
222-225	10.5¢	12%	12	15	10	9	12
226-229	11.5¢	13%	14	19	13	13	13
230-233	12.5¢	15%	14	19	15	16	13
234-237	11.5¢	14%	12	14	14	16	11
238-241	9.4¢	9%	9	8	12	13	9
242-245	7.5¢	6%	6	4	9	10	6
246-249	5.2¢	4%	4	2	5	6	4
Above 249	10.5¢	4%	3	1	6	4	5

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.

Dark Horses The Panel Won't Dismiss

The prose so far is about the center of the board. A few off-center bands drew real numbers from at least one seat, and each is a specific, checkable story rather than a hedge.

• 214-217 seats, Republicans keep it barely (blend 5 percent). This is the strongest realistic Republican-hold path. The mechanism: the redistricting net proves to be the Republican-tilt end of the range, Democrats' own count of toss-up seats to defend costs them a couple on offense, and a normal polling miss leaves them a hair short of 218. It is the only sub-majority band any model kept above a rounding error.
• 234-237 seats, a wave-sized majority (blend 14 percent). The upside dark horse, and the panel's second-heaviest band. If the D+6.6 edge behaves like 2018's larger margin, if a court unwinds a challenged Republican map before Election Day, and if the redistricting net comes out closer to a wash, a mid-230s finish moves from possible to likely. GLM was highest here at 16 percent.
• Above 249 seats, a historic blowout (blend 4 percent, Sonnet 6). The panel floors this band, but it does not zero it. Sonnet kept the widest distribution on the panel and left real weight on a 2018-plus wave in case the map drag proves smaller than the pessimists think.

Where The Panel Changed Its Mind

Each model priced the field once, then read the others' anonymized reasoning and could revise. The revision round tightened the board's tails and pulled its two outliers toward the middle, rather than moving the center, and the arcs are worth reading because they show the panel policing its own overreach.

The biggest move came from Opus, the most cautious model in round one. It had priced the majority at just 71 percent; after the round it raised that to 92, writing:

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 opposite way for the same discipline. It had been the panel's boldest model at 99 percent for a majority and trimmed to 83, widening its tails after peers pressed that the map and the vote-to-seats conversion carry real, two-sided error it had underweighted.

And Fable held the middle it started on, which is the point of a revision round done right, since revision is licensed but conformity is not:

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 Would Change The Panel's Mind

This board re-scores when the story moves. The concrete triggers, and the direction each pushes the seat count:

• Redistricting litigation resolves. 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 single largest swing factor on the board.
• The generic ballot moves off D+6. A drift toward D+8 or better makes the 234-plus bands live and thins the below-218 tail; a tightening back toward D+3 does the reverse and revives the Republican-hold path.
• A batch of new race ratings. The nonpartisan handicappers (the Cook Political Report, Sabato's Crystal Ball) grade each district on a scale from safe to toss-up. Republican-held seats sliding toward Toss-up or Lean Democratic, the direction the last month has run, raises the center; Democratic-held seats sliding the other way lowers it.
• A national shock. The base rate's only two exceptions were an impeachment backlash and a rally-around-the-flag. A comparable event before November would be the one thing that could break the whole model, in either direction.

When a trigger fires, the panel re-scores and this page updates, with the "prices as of" stamp carrying the date.

Settlement Timeline

Market	KXDHOUSESEATS-27, Democratic House seat count, 120th Congress
Settles	February 1, 2027, the date named in Kalshi's rules for the Democratic seat count (the 120th Congress convenes January 3, 2027)
Next catalysts	Redistricting court rulings through the fall of 2026; the final pre-election generic-ballot and ratings updates in late October; Election Day, November 3, 2026; results and any recounts through late November
Re-score policy	This page is re-scored whenever the story moves, with the "prices as of" stamp carrying the date

Frequently Asked Questions

How many seats do Democrats need to win the House in 2026? Democrats won 215 seats in 2024 and need a net gain of about three to reach 218, the number that makes a majority in the 435-seat chamber. Republicans won 220, and a run of vacancies has since left the live chamber near 212 Democrats to 219 Republicans, but the target is the same: everything at 218 or above is a Democratic majority. Kalshi prices the full range of possible Democratic totals in twelve bands.

Will Democrats win the House in 2026? They are favored. Kalshi's seat-count board implies about an 81 percent chance Democrats reach 218 or more, and our price-blind AI panel puts it near 90 percent. Both center the most likely outcome around 229 to 231 seats, a solid majority but short of a historic wave, because the 2026 redistricting maps and a small pool of competitive districts cap the gain below the usual midterm swing against the president's party.

How does the Kalshi Democratic House seats market settle? The board lists twelve mutually exclusive bands, from below 210 seats to above 249. Exactly one settles YES, based on how many House seats Democrats hold on February 1, 2027, the date named in Kalshi's rules (the 120th Congress convenes January 3, 2027). The seat total is decided by the November 3, 2026 election, so any recount or contested seat resolved before the chamber organizes counts toward the settled result.

What could keep Republicans in control of the House? Republicans keep the House in any outcome below 218 Democratic seats, which the market prices near 19 percent and the panel near 10 percent. The realistic path is a combination: the 2026 redistricting maps net toward Republicans, Democrats lose a few of their own competitive seats on defense, and the polls overstate Democrats by a normal margin, leaving the party just short of 218. It does not require a Republican wave, only for the map to fully offset a favorable national environment.

Hottest Prediction Markets Right Now

• 2028 Democratic presidential nominee · $179M traded
• FedEx St. Jude Championship Winner · $70M traded
• 2027 Pro Football Champion · $62M traded
• 2028 U.S. Presidential Election winner? · $59M traded
• 2028 Republican presidential nominee · $57M traded
• Pro Baseball Champion · $53M 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 Bottom Line

Two independent reads, the crowd's money and five models that never saw it, land in the same place: Democrats are on track to win the House, most likely finishing in the high 220s or low 230s. That is control, not a landslide, and the reason is the map. The 2025-2026 redistricting round handed Republicans just enough of a structural cushion to blunt a national mood that would otherwise point to a much bigger Democratic year. The one honest disagreement is about how wild the night could get: traders leave real money on both a Republican hold and a Democratic runaway, while the models bet the result lands closer to the middle. All of it gets settled, and graded on our public record, once the seats are counted in early 2027. Between now and then, watch the courtrooms, not the polls: a single ruling on a contested 2026 district can move this board further than a month of surveys.

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.