Updated September 9, 2026 · 22 min read · by Dave Loughran
The headline on this page was written on August 24, and on August 24 it was true: Kalshi listed a market on every scheduled Federal Reserve meeting, 12 of them at once, and the crowd gave the same answer on all 12. Nothing happens. Sixteen days later the crowd has changed its answer on the one meeting that matters most, changed it back, changed it again, and this week widened it during a stretch in which nobody at the Fed was allowed to say a word. As of 10:37 a.m. ET on Wednesday, September 9, the September 16 board prices a quarter-point hike at 55 cents bid, 56 ask. No change sits behind it at 44 to 45. The board got there in four moves: a chair's first Jackson Hole keynote, one governor's "if," an August jobs report that printed roughly three times what economists expected, and a week of oil headlines that landed after the Fed's pre-meeting blackout had taken every official off the microphone.
The other 11 boards still make no change the favorite, from 48.5 cents at the December 2026 meeting up to 77 at July 2027. So the traders are still betting on a quiet Fed, just not this month, and the machines in the headline are the ones who now look too calm about September. That is the honest state of this page's title, and it is why the standing rule here matters more this cycle than any other: the panel's numbers were locked on August 22, six days before Warsh spoke, 13 days before the September 4 payroll print and 18 days before Brent crude crossed $100, and they stay on the page exactly as they were. Nothing gets quietly rewritten. I will come back to the one number the panel leaned on hardest, because the Bureau of Labor Statistics erased it on September 4 without anyone at the Fed saying a word, and to the one input nobody on the panel was handed, because a barrel of oil does not observe a blackout.
This is the standing OddsShopper page for the Fed decision market. Every cycle we run a fresh panel of AI models over the same fetched facts, with the market prices hidden from them, and every cycle we come back and grade the last run in public. This edition puts the repriced board next to the pre-speech panel, walks the two weeks that moved it, adds the blackout week in which crude did the Fed's talking, and marks the last first-tier release left before September 16 settles which side of a 27-point gap was the overreaction.
The Quick Answer
As of 10:37 a.m. ET on September 9, 2026, Fed rate decision odds on Kalshi favor a quarter-point September hike at 55 to 56 cents over no change at 44 to 45, with the cut showing no bid against a 1-cent ask. The hike widened its lead during the Fed's pre-meeting blackout, with Brent crude at $102 on Wednesday morning and a $4.22 national pump average doing the pushing. Polymarket's September board reads the same way, hike 52 to 53. Our eight-model panel, priced blind on August 22, put no change at 71.5%. The full five-way board, the blackout-week tape, the revision that took away the panel's best argument, and the last first-tier release before settlement are below.
The Market
| Venue | Kalshi, a CFTC-regulated event-contract exchange |
| The Event | "What will the Fed do at its next meeting?" listed separately for each scheduled FOMC date, sold as five mutually exclusive yes/no contracts: cut more than 25bps, cut 25bps, no change, hike 25bps, hike more than 25bps |
| Open Boards | Twelve, from September 16, 2026 through January 26, 2028 |
| Scored On This Page | Five of the 12: September 16, 2026 · October 28, 2026 · December 9, 2026 · January 27, 2027 · January 26, 2028 |
| Settles | On the action the FOMC announces that day. Exactly one contract per meeting resolves YES. Per the written rules, if a scheduled meeting is canceled and does not happen on its date, the no-change contract resolves YES and every other contract resolves NO |
| Current Policy | The target range is 3.50% to 3.75%, held on a 9-3 vote on July 29, 2026 |
| Next Decision | Wednesday, September 16, 2026, 2:00 p.m. ET, the second day of the September 15-16 meeting; the September contracts close at 1:59 p.m. ET that day |
| Prices Below As Of | 10:37 a.m. ET, Wednesday, September 9, 2026, quoted from live bids and asks |
Each contract pays one dollar if its outcome happens and nothing if it does not, so a price in cents reads as a rough probability, and the one clause that matters for this board is that a canceled meeting pays the no-change contract; how prediction markets work and Kalshi's fee schedule cover the rest so this page can stay about the Fed.
Hottest Prediction Markets Right Now
- 2028 Democratic presidential nominee · $188M traded
- 2027 Pro Football Champion · $68M traded
- 2028 U.S. Presidential Election winner? · $61M traded
- 2028 Republican presidential nominee · $59M traded
- Pro Baseball Champion · $58M traded
- Bitcoin price at the end of 2026 · $32M 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 Eleven Days That Flipped The Board Three Times
In the 11 days from the last edition on August 24 to the payroll print, this page's list of events that could move these numbers got worked through in order. That edition listed Jackson Hole first, and said a chair who gives markets less guidance "has a blank sheet of paper and a microphone." On Friday, August 28, Kevin Warsh delivered his first keynote as chair, a speech titled "In Our Time," and three sentences from the published text did the work: "Inflation is running above our 2 percent target. So the Fed's predominant focus right now should be on prices." "We must be confident that underlying inflation is moving to our objective, clearly and at sufficient speed." And the line that sums up his approach to guidance: "I stand here today committed to a discipline, not to a decision." No date, no promised move, and CNBC's same-day read was that the September decision had become a coin flip. The September hike contract, 31.5 cents when this page last ran, closed that Friday at 47 and kept walking through the weekend with no new data at all; it touched 62 cents, its high to date, on Monday, and held 61 to 62 into Wednesday as U.S. strikes on Iranian targets around the Strait of Hormuz kept crude above $90 underneath it.
Then the favorite changed hands in ten minutes on the morning of Thursday, September 3. Governor Christopher Waller, in a Reuters NEXT interview in Washington, said that if the next two weeks of data keep showing progress on inflation, "I would be inclined to support holding the target for the federal funds rate at its current setting." He also named the data his vote hinges on. "If the incoming data for August show this improvement has been fleeting, then it may be appropriate to raise the policy rate when the FOMC meets on September 15 and 16." And he said which data: "So my decision on the appropriate stance of policy will be heavily influenced by what we learn about August inflation." CNBC's headline was that Waller "indicates he will support holding rates steady," and the August inflation report he was pointing at lands September 11. The no-change contract went from 47 cents to 53 and 54 in the ten minutes after his remarks crossed, and buyers kept lifting it all evening: by 4:40 a.m. Friday the hold was 56 bid, 57 ask, the hike 42 to 43. One governor's "if" had undone a chair's keynote and a war premium in a day.
It lasted until 8:30 a.m. on Friday, September 4. The Bureau of Labor Statistics reported that nonfarm payrolls rose by 162,000 in August, the unemployment rate held at 4.1%, and, in the line that matters most for the panel below, July's loss of 23,000 jobs was revised to a gain of 21,000. Yahoo Finance had the consensus near 55,000, wages up 0.3% on the month and 3.1% on the year, and the CME FedWatch reading for a September hike moving from about 50-50 before the print to about 60% after it. Kalshi's own payroll ladder, which closed at 8:29 a.m., had "above 150,000" last traded at 5 cents; it resolved YES. Almost nobody was positioned for it.
The tape shows what a 5-cent event does to a 57-cent favorite. In the single minute after 8:30, the no-change contract printed from 56 down to a low of 41 on about 113,700 contracts, then bounced; the quarter-point hike went the other way, 44 to 53 in the same minute, touched 56 by 8:34, and gave a few cents back as the morning wore on. Roughly 302,000 contracts changed hands on the hold in the half hour after the release, against about 74,000 on the hike. By 9 a.m. the board had found the level it would hold into the weekend: hike 52 to 53, hold 46 to 47. Note the size of that resting point. A print three times consensus moved the favorite, but it did not restore the hike's 62-cent high from earlier in the week, because the market has read Waller's "data for August" as next Friday's inflation report, which is how Bloomberg headlined it, and Principal Asset Management's Seema Shah said the quiet part to Yahoo that Friday, that "next week's CPI report is still likely to be the key swing factor" for the September vote. That week is now this week.
The board after payrolls, in one line: the crowd priced the payroll surprise as secondary to the September 11 inflation report Waller pointed at, which is why the hike was the favorite again on September 4 but short of the previous week's high.
The Blackout Week: No Fed Voice, And Oil Did The Talking
Federal Reserve blackout periods begin the second Saturday before a meeting and run through the Thursday after it, so from Saturday, September 5 through Thursday, September 17, nobody who votes on September 16 is speaking about it in public. The first two weeks of this story were priced off microphones. This one had none, which makes it the cleanest measurement this page has taken of what moves a Fed board when the Fed is silent.
The tape first. The hike did not hold its Friday level. With the committee quiet and no first-tier data due, the hold's buyers crept back over the weekend on thin volume: by Saturday morning, September 5, the two contracts were a cent apart, the hike quoted 49 to 50 and no change 48 to 49, and they sat there through Sunday.
What broke the tie was not in Washington. On Labor Day Monday, Al Jazeera had Brent near $97 a barrel, up 9% in five days and 19% in a month, after U.S. strikes on three Iranian tankers over the weekend and a second strike inside a month on Saudi Aramco's Jizan facilities; the hike closed Monday evening at 53 on its busiest hour since the payroll print, about 36,000 contracts. On Tuesday, September 8, Iran-aligned Houthi forces hit Aramco facilities at Abha, Jazan and Najran and an air base at Khamis Mushait, wounding at least 73 people and suspending operations at several facilities, and Brent went to $98.61; the hold traded more than 100,000 contracts in the hour before noon and nearly 80,000 more in the hour before 3 p.m., and closed the day at 45 to 46. That evening U.S. Central Command said it had destroyed five IRGC-linked oil tankers, four in the Gulf of Oman and one near Kharg Island, and Iran answered with missiles at U.S. targets in Jordan. A pre-dawn block of about 85,000 hike contracts went through Wednesday before 5 a.m., Fortune had Brent at $102.05 at 7 a.m., up $2.20 on the day, and the hike printed 57 in the 9 o'clock hour, its highest since Waller spoke, before settling to the 55-to-56 quote at the time of writing.
The consumer end of that tape is on every corner. AAA's national average is $4.22 this morning, up 7 cents from Tuesday's $4.15, 10 cents in a week and $1.03 above a year ago, after AAA called this Labor Day the most expensive on record at the pump, $4.14 against a prior Labor Day high of $3.82 set in 2012; its diesel average set an all-time high of $5.94 today. This matters to the board for one specific reason. Gasoline is inside the CPI that lands Friday, but the August reading Waller named as his test was collected in August, so Friday's print is the last consumer-inflation number the committee sees that does not carry a $100 barrel. The futures market has read it the same way: Yahoo Finance's Tuesday roundup had CME FedWatch at nearly 56% for a September hike, which puts the rate-futures crowd and the event-contract crowd within a few points of each other on the same question.
Measured the way this page measures everything, the blackout week took 2 cents off the hold's Friday quote and added 3 to the hike's, and all of it came in three sessions that map onto three oil headlines, with no first-tier data, no speech and no interview in the window; the week's only employment print, ADP's August read of 38,000, had landed on September 2, before the blackout began. That is smaller than the keynote's 15.5 cents or the payroll print's 10, and it is the only leg of the move so far that the panel could not have priced from documents, because it had not happened yet. Whether a barrel of oil belongs in a rate decision is a question the committee answers next Wednesday. What the crowd has already answered is that it moves the price of the decision.
The blackout week, in one line: with every Fed official off the microphone from September 5, the September hike went from a 49-cent dead heat to a 55-to-56 favorite on three oil headlines and no first-tier data.
A move this size is exactly the kind of thing this page refuses to take at face value. Three checks. It is more than one trader's order: the five September contracts have put up about $40.9 million of face value in combined lifetime volume (Kalshi counts one dollar per contract), roughly $2 million of it in the last 24 hours, and the no-change contract alone accounts for $15.9 million. It shows up on a sibling series: the exchange's separate ladder on where the rate itself ends September prices "above 3.75%," which is what a hike means when the current range tops out at 3.75%, at 56 to 57 cents, up from a 54-cent prior close. And it holds across venues: Polymarket's September board, fetched the same morning on more than $105 million of lifetime volume, up from about $94 million on September 4, has the quarter-point hike at 52 to 53 and no change at 45 to 46, three cents under Kalshi on the hike and within a cent on the hold. Two venues, a sibling series, and the tape itself agree: the repricing is real, and a September hike has been the market's modal outcome in every session since the payroll print.
What a repricing cannot tell you is whether the crowd is right, or whether a market that moves 30 cents on speeches is pricing the Fed or pricing its own nerves. For that, this page keeps a panel that never heard the speeches: eight models handed the documents alone, so that when the two disagree, the disagreement measures exactly how much of the price is words.
The Case The Panel Built Before Seeing A Price
Eight models were handed the same statement, minutes, inflation and payroll numbers and yield curve, with every market price withheld, then run through a second round in which each read the others' anonymous reasoning and could revise. The panel priced on August 22, six days before the keynote and 13 days before the September 4 jobs report. Its September numbers are a pre-speech, pre-print artifact, preserved on purpose, because a graded record you quietly rewrite is not a record. Two panel findings do the work on this page, both from documents anyone can read, and one piece of context about the man running the meetings frames why the two sides can read the same documents differently.
The Hawks Are A 2026 Story, And The Calendar Says SO
The reason anyone was pricing a hike at all, even before Jackson Hole, traces to a single roll call. On July 29 the Fed held its target range at 3.50% to 3.75%, and three officials voted no because they wanted a quarter-point increase: Beth Hammack of the Cleveland Fed, Neel Kashkari of Minneapolis, and Lorie Logan of Dallas. It was the first time since September 2016 that three policymakers dissented in the same direction. Then the minutes of that meeting came out on August 19 and showed the hawkishness ran past the three dissenters: participants said that "policy tightening would likely be necessary if inflation did not decline," and some argued that financial conditions "might not currently be sufficiently restrictive" to get inflation back to 2%.
Claude Opus, which came in at the panel's lowest September hike number, went looking at who those three people actually are, and found the thing this page had never said out loud. Four Reserve Bank presidents rotate onto the voting committee each year alongside the Board of Governors and New York, and the Fed's own committee page lists the 2026 four as Hammack, Kashkari, Logan and Anna Paulson of Philadelphia. Three of the four dissented. In other words, the entire Board of Governors, plus New York, plus Philadelphia, voted to hold.
"The dissent count is a composition illusion: those three names are the 2026 rotating presidents, so the actual signal is that the full Board of Governors and New York voted to hold.", Claude Opus
And the same page says the 2027 rotation goes to Chicago, Richmond, Atlanta and San Francisco. Hammack, Kashkari and Logan do not lose an argument when the calendar turns. They lose their votes: the rotation seats at the first scheduled meeting of 2027, which is the January 26-27 one. Whatever the hawkish case is worth, it is worth it inside a window that closes on a published date. When this page last ran, the market priced that window at a 10.5-cent drop in hike odds between December 2026 and January 2027. Today that cliff measures about 21 cents: a quarter-point hike is 42 cents at the December meeting and 20.5 at the January 2027 meeting. (Read the direction more than the decimal: the January 2027 board is thinly quoted, so the second leg of that comparison is soft.) Two weeks of repricing made the window more valuable, and the market steepened the calendar cliff this page described in August without anyone announcing it. Waller's September 3 remarks are the first time a member of the hold majority has argued the hold case out loud, a statement from the bloc that voted to hold that this page has had to price, a condition and a date rather than a roll call, which is why one governor moved a board that three dissenters could not.
The Argument The Panel Had About A Scheduling Rule
Buried in those August 19 minutes, well past the policy section, is a paragraph about the shape of the calendar itself. Chair Kevin Warsh, who took office on May 22, 2026, floated cutting the published eight meetings a year to six. The minutes record it in the Board's own careful phrasing:
"The Chairman observed that six scheduled meetings per year, held roughly every two months, would allow more information to accumulate between meetings than under current practice and provide policymakers and the staff more time to consider strategic monetary policy issues. The Chairman asked for input from the Committee on these issues, but no decisions regarding possible changes in the meeting schedule were made, and the Chairman indicated that any change in practice would not affect the schedule over the balance of 2026.", Minutes of the FOMC, July 28-29, 2026
Now put that next to the cancellation clause in the table above. A calendar change would not be a coin flip on the economy. It would be a payout. Gemini Pro found that connection on its own in the first round and swung hard at it, writing that "the trajectory of this contract ladder is mechanically hijacked by the FOMC's active debate about shifting to a six-meeting annual schedule." [Editor's note: Gemini's first-round version named the October 2026 meeting as at risk. The Chairman's own sentence in the minutes rules that out. The panel's second round narrowed the claim to the 2027 and 2028 dates, the only ones the argument can reach.] Five of the other seven seats went the other way, and said so bluntly.
"I rejected [the] claim of 'massive cancellation risk' because the minutes explicitly recorded 'no decision' on the six-meeting proposal.", GLM
"I rejected [the] cancellation-hijack thesis: the minutes show an undecided discussion, tentative labels are routine, and a six-meeting format would not obviously delete these specific dates, so the clause earns only 1pt at 2027-01 and 2-3pt at 2028-01.", Kimi K3
The tiebreaker is Federal Reserve housekeeping: the Board's calendar carries eight meetings for 2026, eight marked tentative for 2027, and exactly one date for 2028, January 25-26, sitting there as a footnote, with each date confirmed at the meeting before it. So the honest reading is the narrow one Kimi landed on: the clause is dead for the rest of 2026 because the Chairman said so, and it is worth a point or two on the 2027 and 2028 contracts, which is small but not zero and is more than the price sheet shows anyone paying for it. A rule that pays out on a scheduling decision only surfaces when you make eight models read the settlement language instead of the headline.
What Warsh Is Being Criticized For
Start with a measurement this page can make from its own tape. Between the August 24 edition and 4:40 a.m. on September 4, the September no-change contract lost 11 cents, from 67.5 to a 56-57 quote, and no first-tier inflation or payroll print landed in that window; the inputs were a chair in Wyoming, a governor in Washington, and a war premium in crude. The jobs report then opened a 15-cent air pocket inside the first minute and, after the bounce, left the hold 10 cents below its pre-dawn quote by 9 a.m. So of the 21 cents the hold gave up in those 11 days, more than half was priced off speeches. The blackout week then added a third input to the list, crude, and took another 2 cents off the hold with no Fed voice anywhere in the window, which is the vacuum measured in its purest form; the panel and the crowd, 4 points apart on September when this page ran in August, now sit 27 apart on the same row. When a chair removes forward guidance, the market loses the input it leans on hardest, and this ladder has to be priced from votes, minutes, data, and now microphones. That is what a guidance vacuum looks like when you measure it.
The vacuum is the documented substance of the criticism of Warsh's short tenure, and the critics predicted the tape above in August: Bloomberg's opinion desk criticized his unwillingness to say how policy adjusts, Goldman Sachs on reduced guidance amplifying instability, and Bloomberg's August 20 report that Treasury Secretary Scott Bessent is at odds with him. Measured in cents per sentence, the difference between the two Fed voices is the number nobody else is printing: Warsh's keynote moved the hike contract about 15.5 cents on the day and 30 by Monday without naming a condition; Waller's one conditional sentence moved the hold 7 cents in ten minutes and told the market exactly which release to wait for. At Jackson Hole the chair answered none of his critics: "committed to a discipline, not to a decision" is the ambiguity his critics describe, restated as a philosophy. Waller's September 3 interview, by contrast, is what guidance looks like: a condition, a meeting date, and the data he is waiting on. The market rewarded the governor's sentence with a clean 7-cent repricing in ten minutes, and the chair's with a session of guesswork that kept running for three days, and the difference in shape is the whole critique in miniature.
We are not scoring that argument, only what it does to a price, and it is why the six-meeting proposal is more than trivia: fewer meetings would mean fewer statements, and less of the guidance the critics already say is missing. That is the build: a hawkish bloc on a timer, a settlement clause nobody prices, and a chair who will not say what he is watching until he is standing at a podium.
Where The Panel Settled Its Own Argument
Every model read the other seven's anonymous reasoning and could revise; the full tables these final numbers land in are in the board section below. The movement was not random: the panel converged toward calm on September and toward action further out. The biggest single move belonged to Gemini Pro, which came into round one as the outlier hawk with September at a straight 45-45 split between no change and a hike, and left round two with no change at 73%.
"I am integrating Forecaster G's structural observation that the three hawkish dissenters are rotating 2026 voters who lose their franchise in 2027, which severely caps long-term hike risk. I held my near-term 'Hold' numbers high because the -23k payrolls print effectively paralyzes the hawkish minority from hiking immediately." — Gemini Pro
Every seat saw the others only as Forecaster A through H, so nobody could defer to a brand; Forecaster G was Claude Opus. The payroll print Gemini leans on no longer exists in the official data; the section after the board explains what happened to it. Kimi K3 moved for the same rotation reason.
"I had weighed the dissents as a standing hawkish bloc and underpriced how mechanically its influence ends on January 1, 2027, so I cut [the quarter-point hike at the January 2027 meeting] to 17% and made [it] the first meeting where cuts rival hikes." — Kimi K3
[Editor's note: the rotation actually seats at the first scheduled 2027 meeting, January 26-27, not literally on January 1. The quote is verbatim; the mechanism it describes is right, the date is loose.]
And the seat that supplied the rotation argument trimmed its own September hike to 15%, the lowest figure anywhere on the board, while stating the case against itself rather than ignoring it. The peer prices it names are first-round numbers, from before the revision pulled the panel together.
"Held: my September hike at 15%, against three peers pricing 36-45%. Their strongest argument is that the hawks' concern is the inflation LEVEL (core still ~50bp over target) and not its trend, which does not expire when a monthly print improves." — Claude Opus
[Editor's note: Opus also wrote that it discounted its own voter-rotation argument because "the incoming 2027 cohort includes St. Louis and Kansas City." The Fed's published table lists the 2027 rotating voters as Chicago, Richmond, Atlanta and San Francisco (New York votes every year). The rotation fact holds; that particular hedge on it does not.]
The panel's second round moved it toward calm on September, from Gemini's 45-45 round-one split down to a 71.5% panel median on no change, six days before a speech sent the market the other way and 13 days before a jobs report did it again. Round one's hawk, before it revised, was closer to where the crowd lives today than any final number the panel shipped. Whether the revision round sharpened the panel or sanded off its best guess is precisely the kind of question the September 16 settlement answers in public.
Stay ahead of the markets.
Daily insights and expert picks on Kalshi, Polymarket, and what's moving markets.
Free forever. Unsubscribe anytime.
The Board: Five Meetings, And One Of Them Broke Ranks
The Kalshi column is the midpoint between the live bid and the live ask at 10:37 a.m. ET on September 9, 2026, with the move since this page's August 24 edition beside it. The panel column is the median of eight independent model estimates, generated from a card of fetched public data with every market price withheld, locked on August 22, before the keynote and before the jobs report. The gaps now measure two things at once: where the models and the crowd read the same documents differently, and how far a speech, an interview, one payroll print and a $100 barrel moved the crowd while the panel stood still. These are the round-two numbers.
| Meeting | Kalshi: no change (Sept 9) | Move since Aug 24 | The panel: no change (Aug 22) |
|---|---|---|---|
| September 16, 2026 | 44.5¢ | −23 | 71.5% |
| October 28, 2026 | 68.5¢ | −3.5 | 63.5% |
| December 9, 2026 | 48.5¢ | −17 | 59% |
| January 27, 2027 | 56.5¢ | 0 | 64% |
| January 26, 2028 | 52¢* | n/a* | 51% |
*The January 2028 no-change contract is quoted 34 bid against 70 ask this morning, on fewer than 100 contracts traded since September 4; that midpoint is a sketch, so no move is scored on it this edition.
More live boards from the same panel: the market on whether the Fed cuts at all in 2026 prices any cut at about 10 to 11 cents even with a hike favored for September · the monthly core inflation board is the one Waller's test actually runs on · our September Fed board tracks the front meeting alongside the deadline markets. Prices fetched September 9, 2026.
Read the move column before anything else, because its shape is the story. Sixteen days of headlines did not lift the whole ladder; they moved two of the five boards this page scores. September has taken 23 cents off no change, December 17, and the other three barely moved, with January 2027 unchanged and January 2028 not trading at all; its last real reading, 65 cents on September 4, was within two cents of the August board. The market's implied narrative is oddly specific: a hike in September, a quiet October to watch it land, a real fight over a second hike in December while the hawks still have their votes, and then the 2027 rotation washes it all out. The crowd, in other words, has adopted the panel's calendar argument wholesale. It just disagrees, by 27 points now, about how fast the hiking starts. The row I keep coming back to is December, because it is the one that went to a dead heat during the post-speech run, at 46.5 cents each way, and has since drifted back to a slim no-change lead at 48.5 against 42 for the hike. The blackout week did not touch it: the December hike rung traded 23 contracts in the last day, so this week's oil bid went into September alone. The market is treating September and December as one question asked twice: the December hike caught about two-thirds of the September move, 26 cents to 42 against 31.5 to 55.5, so the crowd is pricing one hiking decision with two possible dates and a hardening preference for the earlier one.
September, Rung By Rung
When this page ran on August 24, the front meeting was the least contested number on it, panel and market 4 points apart. Today that same meeting is the widest gap anywhere on this page, and it widened again this week.
| Outcome | Kalshi (Aug 24) | Kalshi (Sept 4) | Kalshi (Sept 9) | Move since Aug 24 | The panel (Aug 22) |
|---|---|---|---|---|---|
| Cut More Than 25bps | 0.5¢ | 0.5¢ | 0.5¢ | — | 0% |
| Cut 25bps | 1.5¢ | 0.5¢ | 0.5¢ | −1 | 3.5% |
| No Change | 67.5¢ | 46.5¢ | 44.5¢ | −23 | 71.5% |
| Hike 25bps | 31.5¢ | 52.5¢ | 55.5¢ | +24 | 24% |
| Hike More Than 25bps | 0.5¢ | 1.5¢ | 1.5¢ | +1 | 1% |
An honesty note, because not every row is really a quote. The bigger-cut and bigger-hike contracts still sit at zero-to-one-cent bids against one-to-two-cent asks; they have traded plenty over their lives, nearly $9.5 million on the jumbo hike alone, but at today's quotes the spread is the whole position. The quarter-point cut tells its own small story about the 16 days: at the August 24 update it had crept up to a real 1-cent bid, the keynote took that bid off the board, and neither a 162,000 payroll print nor a $100 barrel is the kind of number that puts it back. It quotes no bid against a 1-cent ask, on a contract still carrying about $6.6 million of face-value open interest (outstanding positions, not size waiting at the bid). No change carries the volume, about 650,000 contracts in the last day, and the hike row carries the price action on about 223,000; even the quarter-point cut, quoted at no bid, still turned over about 114,000 contracts. The odd row is the jumbo hike: quoted 1 to 2 cents, it turned over more than a million contracts in the last day, more than the quarter-point hike and the cut combined, which at those prices is penny-tail churn rather than anyone's forecast of a half-point move. This is as deep as event markets get: the five September contracts combined have put up roughly $40.9 million in lifetime volume.
Here is the wide version for September, with each model's full distribution after the revision round, all locked before the speech and before the print.
| Outcome | Kalshi (Sept 9) | Panel | Claude Fable | Claude Opus | Claude Sonnet | ChatGPT | Gemini Pro | Kimi K3 | GLM | DeepSeek |
|---|---|---|---|---|---|---|---|---|---|---|
| Cut >25bps | 0.5¢ | 0% | 0% | 1% | 0% | 1% | 0% | 1% | 0% | 0% |
| Cut 25bps | 0.5¢ | 3.5% | 3% | 4% | 6% | 5% | 2% | 6% | 3% | 1% |
| No Change | 44.5¢ | 71.5% | 71% | 79% | 65% | 69% | 73% | 72% | 63% | 82% |
| Hike 25bps | 55.5¢ | 24% | 25% | 15% | 27% | 24% | 24% | 20% | 30% | 16% |
| Hike >25bps | 1.5¢ | 1% | 1% | 1% | 2% | 1% | 1% | 1% | 4% | 1% |
Every seat on this panel is graded against real market settlements — records to date: Claude Fable 87% on 1,549 graded calls · Claude Opus 87% on 1,621 graded calls · Claude Sonnet 85% on 1,597 graded calls · GPT 85% on 7,942 graded calls · Gemini 86% on 6,624 graded calls · Kimi 84% on 3,119 graded calls · GLM 82% on 2,941 graded calls · DeepSeek 81% on 2,985 graded calls. Recomputed daily; the full scoreboard is public.
These are model estimates, not predictions of fact and not financial advice. The models were scored without seeing market prices, from data available on August 22, 2026, six days before the Jackson Hole keynote and 13 days before the August jobs report; prices in the tables were re-fetched September 9. One inflation report, the August CPI on Friday, September 11, still lands before the September 16 decision.
The gap on the no-change row is 27 points, and there is no way to soften what that means: one side of this table is going to look bad on September 16. If the Fed hikes, the panel's 71.5% hold, built on the composition illusion and a negative payroll print that has since been revised positive, will have been overtaken by the three most consequential inputs of the cycle, a chair's first Jackson Hole speech, a jobs report three times consensus and a $100 barrel, none of which existed when the numbers were locked. If the Fed holds, the market will have paid 56 cents plus fees for a speech that named no date, a payroll number that, on the market's reading of Waller, is not the report that decides his vote, and an oil price no voter has been allowed to comment on, and the panel's refusal to move off the documents will look like discipline instead of blindness. Every number in this piece gets graded in public once each market settles, and you can check the running tally on the full graded scoreboard. Panel medians are taken outcome by outcome, so they do not necessarily add to 100; the individual model columns do.
A Worked Example: What The 56-Cent Hike Contract Pays
Concrete numbers, because a coin flip with a favorite confuses people. A YES contract on "Hike 25bps" at the 56-cent ask costs $0.56, so a 100-contract position is $56, plus Kalshi's trading fee. If the FOMC announces a quarter-point increase on September 16, the position settles at $100, a $44 gross profit before fees; any other action and the $56 is gone. The same math on the other side is what makes this board interesting: no change at the 45-cent ask pays $100 on a hold, a 122% pre-fee return, on an outcome the same market priced at 67.5 cents 16 days ago and at 57 cents at breakfast on September 4. Strip the board's small overround, its five midpoints add to 102.5 cents, and the market's cleaner read is about 54% hike and 43% hold; your own break-even is the price you actually pay, plus fees. The market has left itself almost no margin for error, and Friday's inflation report is the last scheduled input that can hand one side of the book a very good morning.
The Number The Panel Leaned On Just Got Revised Away
The seat that moved furthest toward a September hold cited one fact directly: the July jobs report, published August 7, showed payrolls falling by 23,000, and Gemini said the print "effectively paralyzes the hawkish minority." The panel's flagship long-dated call, a January 2028 cut, was built on the series that print belonged to, payrolls reading +148,000, +63,000, +20,000, minus 23,000 from April through July. The last edition of this page said the same thing in its own voice, listing the August jobs report as the input most likely to put a cut back on the board if it printed negative again, and, in the same breath, that "a sharp bounce above 100,000 hands the dissenters their trigger instead."
The bounce was 162,000, and the negative print went with it. The Bureau's release revised July from a loss of 23,000 to a gain of 21,000 and June from 20,000 to 31,000, so the series the panel read now runs +148,000, +63,000, +31,000, +21,000, +162,000, which reads as a slowdown that bottomed. That does not make the panel's hold call wrong. Its two structural arguments, the composition of the dissent and the rotation cliff, are untouched by a revision, and Waller's stated condition was disinflation continuing in the August data, which the market has read as the inflation report rather than payrolls. But it removes the panel's best cyclical argument at exactly the moment the market needed a reason to stop at 52 cents rather than run back to 62, and a graded record is supposed to say so. If the September row settles against the panel, the post-mortem is already written: the models were right about who votes and wrong about what the economy was doing, because the economy's own first draft was wrong.
The revision, in one line: the July payroll loss the panel built on is now a gain of 21,000, so the panel's structural case (who votes, and until when) survives the revision and its cyclical case does not.
The market's restraint was the more interesting half of that morning, and it lasted through the weekend. A hike at 52 to 53 after that print was a smaller number than the 62 the same contract touched on a Wyoming speech, with an oil bid behind it, which meant the crowd had decided a governor's stated condition outranks a payroll surprise. What it has decided since is narrower: with the governor and everyone else silenced by the blackout, the oil bid alone was worth another 3 cents, and the hike at 55 to 56 is still 6 short of its high. That is a defensible read and a checkable one: Bloomberg's headline on Waller was that the September decision "hinges on August CPI," and the Bureau's schedule puts that release at 8:30 a.m. ET on Friday, September 11. Between now and then the board is pricing a report that does not exist yet, which is the same thing it was doing at 4:40 a.m. on September 4, before payrolls landed and briefly took 15 cents out of the hold.
January 2028: The Thin Board Still Leans Toward The Panel
While everyone watched September, the far end of the ladder kept drifting toward the panel, with the loudest caveat on the page attached: the January 26, 2028 contracts have put up about 7,400 contracts of lifetime volume, so everything in this section is drawn in wide spreads. The board settles on the announcement of the only 2028 meeting the Fed has put on paper, and it was where the disagreement was widest at the August 24 update.
| Outcome | Kalshi (Aug 24) | Kalshi (Sept 4) | Kalshi (Sept 9, quote) | The panel (Aug 22) |
|---|---|---|---|---|
| Cut More Than 25bps | 3.5¢ | 4.5¢ | 5 to 6 | 5.5% |
| Cut 25bps | 5.5¢ | 12¢ | 9 to 16 | 21% |
| No Change | 63.5¢ | 65¢ | 34 to 70 | 51% |
| Hike 25bps | 11.5¢ | 15.5¢ | 6 to 30 | 15.5% |
| Hike More Than 25bps | 11.5¢ | 9.5¢ | 10 to 21 | 4% |
This week the board did not trade: fewer than 100 contracts changed hands in five days, and what moved were the quotes, not the prices. The no-change contract is now quoted 34 bid against 70 ask, which is not a market so much as a market maker keeping a seat warm, so the September 9 column shows the raw quotes rather than midpoints that would only launder that width. The September 4 numbers remain the last real reading; two rows carried the whole section on them, and both had moved toward the panel.
The quarter-point cut was the panel's flagship call and the widest gap anywhere on this page: 21% against a 5.5-cent price, with seven of the eight seats in double digits and DeepSeek highest at 28%. GLM called the payroll series "an unusually sharp deterioration" and argued that by 2028 the Fed is more likely cutting than hiking even if it hikes first. The market has since moved in GLM's direction: it favors a September hike at 55.5 cents today, and on September 4 it lifted the January 2028 cut to a 12-cent midpoint, up from 5.5, while still pricing 2028 hikes over cuts on that last real reading, 25 cents to 16.5 across the two rungs on each side, and paying the panel's flagship only about 57% of its 21%. A caution before anyone gets excited on the panel's behalf: that midpoint sat on a 9-cent bid against a 15-cent ask, the ask has since widened to 16 with nothing traded, and the September 4 revision took away the deterioration GLM was describing. The direction of the drift is real; the size of it is a wide-spread sketch, and the argument under it just got weaker.
The jumbo hike ran the other way. At the August update the market had "hike more than 25bps" for January 2028 at the same 11.5-cent midpoint as the ordinary quarter-point hike on the same board, effectively pricing any 2028 hike as a coin flip to be a half-point panic move. Our models put the jumbo at 4%, with six of the eight seats at 5% or under, on the argument that half-point hikes are what a Fed behind the curve makes, and nobody could date such a moment. Once a standard quarter-point path was live and priced, the jumbo faded to a 9.5-cent midpoint on a one-cent spread by September 4, 9 bid against 10 ask, while the ordinary hike firmed to a 15.5-cent midpoint, on a wide 10-to-21 quote, level with the panel's 15.5%. The market was unwinding the crisis premium as the panel said it should, without a single data point that dates a 2028 decision having printed. This week both quotes blew out, the jumbo to 10 against 21 and the ordinary hike to 6 against 30, on no trades worth the name, so that unwind is a September 4 reading until someone trades the rung again.
Both numbers still deserve the thin-board caveat. The January 2028 contracts have traded about 7,400 contracts against nearly 16 million on the September no-change contract alone, and this week's quotes are wide enough that the midpoints stop meaning anything, which is what wide spreads and few traders always look like. Treat the long end of this ladder as a sketch, not a photograph, and re-run the numbers at whatever price is actually on your screen.
Dark Horses The Panel Won't Dismiss
Two outcomes the price sheet has mostly written off that at least one seat refuses to. A third, a September cut, was on this list at the last update; the September 4 payroll print and the revision above put it down, and it quotes no bid against a 1-cent ask.
A half-point hike in December 2026, at a 2.5-cent midpoint. GLM alone gives it 6%, and its logic runs as a chain: if core inflation reaccelerates through the autumn, a committee that already told itself tightening "would likely be necessary" would be behind and would move in larger steps than a quarter. The ordinary December hike went from 26 cents at the last update to 42 today, and the half-point rung quotes 2 bid against 3 ask behind it. Friday's CPI is GLM's test as much as Waller's, and a $100 barrel is the kind of input that can turn its "if" into a reading, first in headline inflation and then, if it lingers, in core.
A big cut in January 2028, quoted 5 bid against 6 ask. GLM again at 10%, DeepSeek at 8%. This is the recession tail nobody wants to write down, the mirror image of the market's jumbo-hike bet on the same board, and since August the market took the jumbo hike from 11.5 cents to 9.5 and then, this week, to a quote 11 cents wide, while the big cut's 2-cent drift, inside spreads like these, is not a move anyone should read. The panel's own numbers pick a side: it prices the big cut at 5.5% against a 5-to-6-cent quote, near fair, and the jumbo hike at 4% against a quote whose bid alone is 10. If the panel is right, the crisis premium on this board is still on the hike rung, not the cut rung.
What Moves These Numbers Next
Three inputs have now moved the front board, two of them by double digits, and the third, crude, was on nobody's list. One first-tier release remains before the September settlement, then the settlement itself, and after that the one date that ends the argument entirely.
- The August CPI Report, Friday, September 11, 8:30 A.m. ET. Now the fulcrum of the whole board: the August inflation data Waller said would weigh heaviest on his vote, and the release the market has taken as his test. Core inflation ran at 2.48% over the year in July. Warsh's Jackson Hole bar was confidence that underlying inflation is moving to target "clearly and at sufficient speed"; Waller's bar is that the recent progress not prove "fleeting." A hot core print five days before the vote satisfies the hawks and would likely push the hike back toward the high end of its range; a soft one hands Waller his hold, and the 44-cent no-change contract becomes the side the market abandoned three times. One wrinkle the board has already priced: the August gasoline component was collected before this week's move above $100, so whatever Friday says about August, the September CPI, due October 14, is the first one that carries the barrel.
- Crude, With No Release Time. The input that arrived without a date. Brent went from about $97 on Monday to $102 by Wednesday morning, the pump average to $4.22, and the hike added 3 cents on it in a week when no Fed official could speak. Another leg up before Wednesday pushes the same rung; a ceasefire headline is the one thing that could put the hold back in front before the data does, and it is not on any calendar either.
- The FOMC Decision, Wednesday, September 16, 2:00 P.m. ET. The September contracts close at 1:59 p.m. and settle on the announced action. This page grades both sides that afternoon, and the July precedent below says exactly how.
- The 2027 Voter Rotation, Seated At The January 26-27 Meeting. Nothing to forecast: the three dissenters stop voting, four different Reserve Bank presidents start, and everything on this page about the hawkish bloc expires. The market prices that cliff at 21 cents of hike odds between December and January, roughly double what it priced in August.
| Date | What lands |
|---|---|
| August 28, 2026 | Jackson Hole keynote. Fired: the September board went from a 67.5¢ no-change favorite to a coin flip within a session and a hike favorite by Monday |
| September 3, 2026 | Waller's Reuters NEXT remarks. Fired: the hold retook the lead at 53-54¢ in ten minutes |
| September 4, 2026 | August jobs report, +162,000. Fired: the hike retook the lead in one minute, settling at 52-53¢ by 9 a.m. |
| September 5-9, 2026 | Fed blackout from September 5, no first-tier data. Fired anyway: Brent $97 to $102, and the hike from a 49-50¢ dead heat on Saturday to 55-56¢ on Wednesday |
| Friday, September 11, 2026 | August CPI report, the August inflation data Waller said would "heavily" influence his vote |
| September 16, 2026 | The FOMC decision. The September contracts settle that afternoon and this page grades both sides |
| October 28, 2026 | Next scheduled decision |
| December 9, 2026 | Final 2026 decision, the last one the current voting bloc decides |
When any of these fires, this page gets re-scored, and the graded results stay on the page rather than being quietly replaced.
Last Fed Decision: How The Models Did
The July 2026 contracts settled on July 29. The Fed held, so no change resolved YES and everything else expired worthless. We scored that board on July 24, five days out, with the market price on the run date next to the panel's number.
| Outcome | Kalshi (Jul 24) | Panel | Result | Grade |
|---|---|---|---|---|
| Cut More Than 25bps | 1¢ | 2% | NO | Both right |
| Cut 25bps | 1¢ | 17% | NO | Market right, panel wrong |
| No Change | 75¢ | 66% | YES | Both right; market more confident |
| Hike 25bps | 26¢ | 12% | NO | Panel right, market overpriced it |
| Hike More Than 25bps | 1¢ | 2% | NO | Both right |
Both sides made no change the favorite and no change happened, with the market's 75 cents closer to the truth than the panel's 66%. The contested rungs split: the panel's hike fade was vindicated, 12% against a 26-cent price on a contract that died, while the cut was an embarrassment, a blended 17% on something the market correctly priced at a penny. (A no-change buyer at 75 cents was paid a dollar five days later, less Kalshi's trading fee; had the Fed moved, the whole 75 cents was gone. There is no partial credit in event contracts.)
That July hike fade is exactly why the September table above deserves your attention. Last cycle the panel faded a 26-cent hike and was right. This cycle it is fading a 55.5-cent hike, the same lean at twice the price, except that this time the fade was priced before the chair stood up at Jackson Hole, before a governor named his test, and before the payroll series the panel was reading got rewritten. Same instinct, radically different information set. Whether a document-driven fade survives a speech-driven repricing and a data revision is the cleanest test this page has ever gotten to run.
| Cycle | Market favorite (at run) | Panel favorite | Result | Read |
|---|---|---|---|---|
| July 2026 | No change, 75¢ | No change, 66% | No change | Both hit; panel better on the hike rung, market far better on the cut rung |
One cycle is not a sample; the table exists so that when it is six rows deep you will not have to take our word for how the panel performs. The September row grades itself on September 16, and four open calls on later meetings stay on the record.
One note about this site: OddsShopper's odds screen and comparison tools cover sports markets, not Fed contracts, so this page pitches no tool. It runs the panel, shows the board, and keeps score. A rules note worth carrying with you: the meeting contracts scored here have unusually detailed written terms, but not every rate market on the exchange is written that tightly, so read a contract's settlement terms before treating its price as gospel. Our guide to Kalshi's economic markets covers the wider family of contracts, our ranking of prediction market platforms compares the venues that list them, and our Fed rate hike odds page carries the minute-by-minute tape on the September board.
More on this: Fed Rate Cut in September? Traders Are Betting on a Hike · Fed Rate Cut Odds: A Hike Is Priced Higher At Every Meeting · NFL Spread Predictions: The Panel Doubts Almost Every Favorite · Colorado Vs Georgia Tech Picks: Two Untested QBs, One Team That Can Run · Kalshi Fed Odds: Trump's Own Pick Might Hike On Him
The Bottom Line
The crowd's answer is still stillness at 11 of the 12 meetings on the ladder, which is what the headline on this page has always said. The twelfth is the one that settles first, and the market has compressed the entire Warsh era's drama into it and the December meeting after it, the four months its dissenters have left. The panel, handed the same raw material before the speech, the interview, the print and the barrel, is on the record the other way, and one of those two positions is about to be graded, in public, on this page.
Both sides are leaning on something real, and not the same something. The market is leaning on the clearest guidance this chair has given, a payroll number nobody forecast, a barrel of oil that crossed $100 in a week when no Fed official was allowed to answer it, and $2 million of face value traded in the past day. The panel is leaning on the documents: a hold vote that was 9-3, a hawkish bloc whose votes expire in January, and a negative payroll print that the Bureau has since revised into a positive one, which is the honest weak point in its case and this page says so. The tiebreakers arrive on a schedule: the inflation report Waller named on September 11, then the decision itself on September 16, when the September contracts settle, this page grades both sides, and four open calls stay on the record. One more number for the road, from the exchange's separate ladder on where the rate itself lands: the contract that pays if the funds rate sits above 4.00% after the December meeting, which takes two quarter-point hikes from here, trades at 36 to 37 cents. That is the market's price on the Warsh era being a cycle rather than a single move. If you would rather have something that settles tonight, our experts post free picks every day.
Prices quoted are live Kalshi and Polymarket prices as of 10:37 a.m. ET on Wednesday, September 9, 2026, except where another time is given, and will keep moving until these markets settle. This article is informational and does not constitute financial, investment, or trading advice; the model figures are estimates, not predictions of fact.
Hero illustration: OddsShopper, in the house collage style. Kevin Warsh photo by Federalreserve, licensed Public domain; photos cropped, toned, and composited.
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.
