Eight AI models looked at Kalshi's October 1 shutdown contract in August, and every one of them priced the risk higher than the market did. Kalshi graded that contract No at 10:07 a.m. ET on October 1. The panel had the side right and the size wrong, and the market was closer by a wide margin: 12 cents against a panel median of 27.5 percent, on an outcome that paid zero. The full grade is below, including the one row in the panel's own August watch table that explains it, because the same panel, plus one seat, now has a live number on December.
The question did not go away; it moved 71 days. The stopgap the President signed on September 2 runs out on December 11 unless full-year bills replace it first, and Kalshi now lists a contract on the morning after. This page reads the live Kalshi government shutdown odds, that contract and the board on how many shutdowns 2026 ends with, next to nine AI models that priced the December count blind (Grok joined the panel for the December round). October was a market about whether Congress would extend funding it had already agreed on. December is a market about whether it can finish 12 bills, nine of which have not cleared the House and none the Senate, and that difference is what every price on this page is carrying.
Kalshi's Government Shutdown Odds Right Now
The live instrument is a binary contract, ticker KXGOVTSHUTDOWN-26DEC12, titled "Will the US government be shut down on Dec 12, 2026?" It pays Yes if the federal government is at least partially shut down from a lapse in appropriations at 10:00 a.m. ET on December 12, the morning after the stopgap expires. Kalshi opened it at 6:30 p.m. ET on September 22, the same day the previous edition of this page said no such contract existed, which was true that morning. The first trade printed at 46 cents at 11:24 p.m. ET on September 22.
Pricing an NFL combo on Kalshi?The free Kalshi Combo Simulator on Stokastic.com plays this week's game 10,000 times and prices any combo next to the Kalshi price.
| Market Read (As Of October 1, 2026, 9:31 p.m. ET) | Value |
|---|---|
| Yes Price (What Buyers Bid / What Sellers Ask) | 33¢ / 36¢ |
| Last Trade | 36¢ |
| Opened | 6:30 p.m. ET, September 22, 2026 |
| Trading Closes | 10:00 a.m. ET, December 12, 2026 |
The price has already made one round trip. From the 46-cent open it slid for five days and bottomed at 20 cents on the afternoon of September 28, a market deciding the lame-duck Congress would find the cheap exit. On September 29 it climbed back to 38, the day Roll Call reported that Senate appropriators were looking at language to answer the President's pocket rescission. It has held between 30 and 39 cents since. The price is the crowd's probability, give or take spread and fees, so the crowd puts a lapse on the morning of December 12 at roughly one in three. That read rests on a thin book, which the next board puts in proportion.
What The December Board Is Quoting
The second December read lives one board over, in the event "How many government shutdowns in 2026?" (series KXNUMSHUTDOWNS, event KXNUMSHUTDOWNS-27JAN01), one contract per count. Two shutdowns have happened by Kalshi's calendar-year count already (the January lapse and the February-to-April Homeland Security lapse), so the count reaches three only if a lapse is in effect on some morning from December 12 through December 31.
| Shutdowns In 2026 | Yes Price (Bid / Ask) |
|---|---|
| Exactly 2 (No December Lapse) | 60¢ / 66¢ |
| Exactly 3 (One December Lapse) | 29¢ / 38¢ |
| Exactly 4 | 6.0¢ / 6.7¢ |
| Exactly 5 | 0.1¢ / 5.8¢ |
Read the first row in reverse for the market's December number. With "exactly two" at 60 to 66 cents, the board puts a third shutdown before New Year's at 34 to 40 percent. Set that beside the 33 to 36 cents on the December 12 contract and the two boards agree, which they should. A shutdown on the morning of December 12 is one way to a third shutdown, not the only one, so the dated contract has to price at or under the count. Both books are thin: about $58,300 of face value has traded across the four rungs over their life, next to $939,600 on the October contract, so one motivated trader can still move the count board several cents. Since the previous edition the "exactly two" bid has slipped from 63 cents to 60, which is the board edging toward more December risk, not less.
The Verdict: Where Nine Blind Models Landed On December
We put the December count to nine AI models under the same protocol as August: none was shown a price, none was told what the others thought, and each received an identical briefing on the legislative record and settlement rules. Each returned a probability distribution, then read the other eight analyses with the names stripped and could revise once with a stated reason. The panel ran on September 2 and has not been re-asked since; the market rows are marked to the October 1 read.
| The Board | Value |
|---|---|
| Market: A Third Shutdown In 2026 (From "Exactly 2" At 60¢ / 66¢) | About 34% to 40% |
| Market: Shutdown On The Morning Of December 12 (Bid / Ask) | 33¢ / 36¢ |
| Panel Median After Revision | 37% |
| Gap | None; the panel sits inside the market's range |
| Panel Range After Revision | 33% to 40% |
| Market On "Exactly 3" (Bid / Ask) | 29¢ / 38¢ |
| Panel Median On "Exactly 3" After Revision | 32% |
| Model | Round 1: Third Shutdown Or More | Final: Third Shutdown Or More | Final: Exactly 3 |
|---|---|---|---|
| Claude Fable | 38% | 40% | 34% |
| Claude Opus | 36% | 37% | 32% |
| Claude Sonnet | 53% | 38% | 32% |
| GPT | 32% | 37% | 32% |
| Gemini | 26% | 33% | 30% |
| Grok | 35% | 38% | 32% |
| Kimi | 38% | 36% | 31% |
| GLM | 41% | 40% | 35% |
| DeepSeek | 38% | 37% | 32% |
| Blend (Seat Median) | 38% | 37% | 32% |
When this page's September 4 edition went up, the panel sat at the top of the market's range, then 32 to 37 percent. The market has since come up to meet it: "exactly two" has slipped three cents on the bid while the panel's 37 percent has not moved, because nobody has re-asked it. Gemini's 33 percent is the low seat, a point under the count board's range and level with the December 12 bid. The panel sits inside the "exactly three" spread too, 32 percent against 29 to 38. Where it leans is the tail: 37 minus 32 leaves 5 percent for four or more shutdowns, a touch under the 6.0 to 6.7 cents the market prices "exactly four" at on its own.
Every seat on this panel is graded against real market settlements — records to date: Claude Fable 88% on 5,981 graded calls · Claude Opus 87% on 6,126 graded calls · Claude Sonnet 86% on 6,070 graded calls · GPT 85% on 13,424 graded calls · Gemini 86% on 12,259 graded calls · Grok 84% on 9,942 graded calls · Kimi 86% on 7,272 graded calls · GLM 84% on 5,553 graded calls · DeepSeek 84% on 5,346 graded calls. Recomputed daily; the full scoreboard is public.
These are estimates, not predictions of fact and not financial advice. Every number here gets graded in public once the market settles. The October grade is the next section, the full graded scoreboard shows how each seat has done over every settled call, and you can browse every live model-verdict board on the prediction markets hub.
The Record: How The October Call Graded
The contract this page was built on, KXGOVTSHUTDOWN-26OCT01, asked whether the government would be shut down at 10:00 a.m. ET on October 1. Trading closed at 9:36 a.m. ET that morning, and Kalshi graded it No half an hour later.
| The October Grade | Value |
|---|---|
| Panel Median, Round 1 (Asked August 21) | 29% |
| Panel Median After Revision | 27.5% |
| Panel Range After Revision | 24% to 28% |
| Market When The Panel Was Asked (Bid / Ask) | 12¢ / 13¢ |
| Last Trade Before The Close | 1¢, at 11:59 p.m. ET on September 30 |
| Result | No, graded 10:07 a.m. ET on October 1 |
Both columns said no. The market said it at 12 cents, the panel said it at 27.5 percent, and the answer was zero. So the market missed by 12 points and the panel missed by 27.5. Right on the side, wide on the number, and beaten by the market on this contract: that is the grade, and it goes on the scoreboard like every other call.
What the panel priced was the record and the calendar. Three shutdowns since October 2025, two stopgaps that did not match (the House's ran to December 4, the Senate's to December 11 with riders attached), and a Senate not due back in session until September 14. Its own August watch table put the escape route in the top row: the House takes the Senate's text unchanged, with a landing zone of 8 to 10 percent. That is what happened, the day after the House returned. The House came back on August 31 and passed the Senate's H.R. 6500 untouched on September 1, 370 to 48, and the President signed it on September 2 as Public Law 119-103, riders included, the delay of his own budget office's grants rule through December 11 among them. The contract traded at 4 cents on August 31 and at 1 cent by the close of September 1. The market did not stop in the panel's landing zone of 8 to 10 percent. It went through it to 1.
Claude Opus, tied for the lowest seat at 24 percent, had named "presidential-signature risk since the blocked rule is the administration's own" as one reason to stay above the market. The signature came within a day of the vote, rider attached. That is the piece of the grade worth carrying into December: the panel treated the record of this Congress as the base rate, and the market treated a 90-to-6 Senate vote as the base rate. In October the vote count was the better guide.
The contract's last four weeks were a market in No at 97 to 99 cents. From September 2 it never traded above 3 cents, and from September 7 never above 2. The final trade was at 1 cent, at 11:59 p.m. ET on September 30, ten hours before the contract was graded.
What December Has That October Did Not
October asked Congress to extend current funding levels both chambers had already voted for in different forms, and that produced 90 to 6 in the Senate and 370 to 48 in the House. December asks it to do one of two things in the 14 days both chambers are scheduled to sit together between the November 3 midterms and December 11, per the House and Senate calendars. The first is to pass the 12 fiscal 2027 bills, of which the House floor had passed 3 and the Senate none, per the Committee for a Responsible Federal Budget's tracker, last updated September 3. The second, another stopgap into 2027, is the cheap escape, and on the count board it is also the settling condition: a stopgap into 2027, passed before any lapse, ends 2026 at exactly two.
One thing is on the table now that was not in August. On the night of Friday, September 25, the President announced a pocket rescission of $810 million in appropriated money, a cancellation sent so late in the fiscal year that Congress had virtually no time to vote to block it, and Susan Collins told Roll Call, in a September 29 report, that she is "already looking at language" to answer it in the fiscal 2027 bills; Mike Rounds told the same reporters the move was wrong and made Congress's job harder. An appropriations fight with the power of the purse itself at issue is a different fight from a one-week date mismatch; that it feeds the December price is our read, not anything Roll Call wrote. The panel's 37 percent predates all of it. The count board has since risen to 34 to 40 percent and now sits on top of the panel's number. In October the vote count was the better base rate. December has no 90-to-6 vote to anchor on, no stopgap text at all yet, and 14 shared days to produce one, which is why the count board climbing past the September panel reads, on this page, as the market catching up to the record rather than running ahead of it. The December 12 contract settles first, at 10:00 a.m. ET that morning; the panel's call grades when the count board settles after New Year's, in public, on this page.
Stay ahead of the markets.
Daily insights and expert picks on Kalshi, Polymarket, and what's moving markets.
Free forever. Unsubscribe anytime.
More on this: England Vs Pakistan Test Odds: Is The Kalshi Draw Cheap? · Florida Governor Odds: Donalds Vs Jolly After The Primary · 2026 Governor Races: Every Kalshi Board We Track · GTA 6 PC Release Date: The Pattern Rockstar Never Breaks · Jets Vs Titans Odds Week 1: Spread, Total & Kalshi Prices
Disclosure and fine print.Stokastic Inc. trades on Kalshi and holds positions in its markets; this series is an open research log of an approach we have not proven, and we do not publish our own trading returns. The model records quoted above are a separate thing: those are the panel's public forecasting scorecard, graded against settled markets. The shutdown contracts on this page are CFTC-regulated event derivatives traded on a designated contract market, not sportsbook wagers, and a position can lose its full value. Selling an unlikely outcome collects a small premium and risks most of a dollar; at a penny, one loss erases the premiums from about 99 wins. Position size determines how much of that asymmetry a holder is exposed to. We have no affiliate or commercial relationship with Kalshi. 18+, available where Kalshi operates; the risk of loss is real. Nothing here is trading advice, a political forecast, a pick, or a recommendation.
Hero illustration: OddsShopper, in the house collage style. United States Capitol photo by Beethoven, licensed CC BY-SA 4.0; photos cropped, toned, and composited.
