Updated September 10, 2026 · prices as of 9:07 AM ET, September 10 · by Jake Hari
The Quick Answer
The August Consumer Price Index lands Friday, September 11 at 8:30 AM ET, trading on this Kalshi board stops one minute before it, at 8:29, and it is the last inflation reading the Federal Reserve sees before it meets on September 15 and 16. The market's most likely print is still 3.4%, now at roughly a 38% chance, with 3.3% next at about 28%. The forecaster consensus is 3.4% (Kiplinger), and our eight-model panel, which priced this board blind on August 22, also had 3.4% on top by a nose over 3.3%.
Since August 22 the busiest contracts have drifted toward the panel. "Above 3.3%" is 58 cents bid, 61 ask, down from a 64-cent midpoint on August 22, against the panel's 52%. "Above 3.4%" is 20 bid, 23 ask, down from a 31-cent midpoint, against the panel's 25.5%. Then on Thursday morning the August producer price index came in hot on diesel, Kalshi's September rate-hike contract went from 56 cents three minutes before the release to 65 by 8:59, and the two busiest inflation rungs barely twitched: "Above 3.3%" printed 61 thirty seconds before the number and never moved after it, and "Above 3.4%" gave back three cents. What moved since August 22, where the money sits now, and what a 3.3 against a 3.4 does to the hike price, is below.
Stay ahead of the markets.
Daily insights and expert picks on Kalshi, Polymarket, and what's moving markets.
Free forever. Unsubscribe anytime.
The Market At A Glance
| Venue | Kalshi |
| The Event | "Inflation in August 2026 (CPI YoY)," Kalshi event KXCPIYOY-26AUG, 26 yes/no contracts, one per level |
| The Contract | Pays out if the twelve-month change in the all-items Consumer Price Index, before seasonal adjustment, comes in above the stated level, per Kalshi's contract terms. It settles on the one-decimal figure in the BLS news release. The monthly change that leads the headlines is seasonally adjusted and does not count, and neither do later revisions |
| Settles | Friday, September 11, 2026, on the 8:30 AM ET release. Last trade 8:29 AM ET |
| Size | 761,716 contracts of lifetime volume, roughly 575,787 still open, up from 127,130 and 91,368 when this page was first priced on August 22 |
| Prices Below As Of | 9:07 AM ET, Thursday, September 10, 2026 |
The Board
Every rung Kalshi lists, the midpoint this page carried on August 22, the book on Thursday morning, and the panel's blend from the August 22 run, which is the median of eight seats after each read the other seven's reasoning and revised. No model saw a price. Each worked from the BLS release, the weekly gasoline data, the Cleveland Fed's nowcast with its error record, and the contract terms. The panel has not been re-run; its column is frozen so the reader can see which way the market moved against it.
| Contract | Aug 22 mid | Sept 10 bid / ask | AI blend (Aug 22) | Gap now | Contracts traded |
|---|---|---|---|---|---|
| Above 2.5% | 97.0¢ | 99 / 100 | 98% or higher | 32,792 | |
| Above 2.6% | 97.5¢ | 99 / 100 | 98% or higher | 232 | |
| Above 2.7% | 99.0¢ | 99 / 100 | 98% or higher | 42,344 | |
| Above 2.8% | 99.0¢ | 99 / 100 | 98% or higher | 248 | |
| Above 2.9% | 98.5¢ | 99 / 100 | 98% | 3,364 | |
| Above 3.0% | 86.5¢ | 97 / 100 | 95% | −4 | 2,269 |
| Above 3.1% | 87.5¢ | 95 / 98 | 89% | −8 | 20,794 |
| Above 3.2% | 91.0¢ | 86 / 89 | 77% | −11 | 88,229 |
| Above 3.3% | 64.0¢ | 58 / 61 | 52% | −8 | 156,717 |
| Above 3.4% | 31.0¢ | 20 / 23 | 25.5% | +3.5 | 176,617 |
| Above 3.5% | 8.0¢ | 5 / 10 | 11.5% | +3.5 | 121,846 |
| Above 3.6% | 4.0¢ | 1 / 3 | 4% | +2 | 41,212 |
| Above 3.7% | 1.5¢ | 0 / 1 | 4% or lower | 5,457 | |
| Above 3.8% | 2.5¢ | 0 / 1 | 4% or lower | 54,791 | |
| Above 3.9% | 1.5¢ | 0 / 1 | 4% or lower | 1,904 | |
| Above 4.0% | 2.5¢ | 0 / 1 | 4% or lower | 2,686 | |
| Above 4.1% | 3.0¢ | 0 / 1 | 4% or lower | 447 | |
| Above 4.2% | 3.5¢ | 0 / 1 | 4% or lower | 722 | |
| Above 4.3% | 3.0¢ | 0 / 1 | 4% or lower | 3,204 | |
| Above 4.4% | 3.0¢ | 0 / 1 | 4% or lower | 721 | |
| Above 4.5% | 3.0¢ | 0 / 1 | 4% or lower | 750 | |
| Above 4.6% | 2.5¢ | 0 / 1 | 4% or lower | 1,038 | |
| Above 4.7% | 1.5¢ | 0 / 1 | 4% or lower | 695 | |
| Above 4.8% | 2.5¢ | 0 / 1 | 4% or lower | 666 | |
| Above 4.9% | 0.5¢ | 0 / 1 | 4% or lower | 750 | |
| Above 5.0% | 2.5¢ | 0 / 1 | 4% or lower | 1,221 |
More live boards from the same panel: Fed rate hike odds (the September 16 decision: a quarter-point hike 64 bid, 65 ask, a hold 34 to 35) · how Kalshi's CPI markets work (the monthly-change board has "Above 0.3%" at 58 to 61 cents) · core CPI odds (August core "Above 0.2%" is 22 bid, 28 ask). Prices fetched 9:07 AM ET, September 10, 2026.
The gap column takes the panel's number minus the midpoint of Thursday's book, rounded half up, so "Above 3.3%" reads as 60 and "Above 3.4%" as 22. The panel forecast the printed value directly, so the eight rungs from 2.9% to 3.6% get a number. At the ends it gave a lump instead: 2% on anything at 2.9 or lower, which makes 98% a floor for the bottom five contracts, and 4% on anything at 3.7 or higher, which makes 4% a ceiling for the top 14. The ladder is also finally in order. On August 22 "Above 3.2%" had a higher bid than the easier "Above 3.0%"; on Thursday every rung's bid sits at or below the rung beneath it, which is what six times the volume does to sloppy quotes.
The traded column has changed hands. On August 22 "Above 3.3%" was the busiest contract on the board with 33,659 traded. It has traded roughly 123,000 more since, taking it to 156,717 lifetime, and "Above 3.4%" has passed it at 176,617 lifetime with 149,860 contracts still open, the largest open position on the ladder. The three middle rungs, 3.3 through 3.5, account for 455,180 of the board's 761,716 contracts, and 78,262 of those traded in the 24 hours before this pull. That is where the money has an opinion, and it is still exactly where the panel disagrees.
What Moved Since August 22
The four-cent drop in "Above 3.3%" understates what happened to it. The contract closed August 22 at 68 cents, slid through the last week of August and was 52 by the close on August 27. It printed as low as 43, on a 20-contract sale before dawn on Thursday, September 3, and as high as 75 on a few dozen contracts at 11:00 the next morning, after the jobs report; the real money that Friday was a single 5,000-contract buy at 67 at 4:22 PM. Then the buyers came back. On Tuesday, September 8, 86% of the day's 18,447 contracts were bought at the ask rather than sold at the bid, the biggest of them 5,123 at 62 at 2:03 PM; on Wednesday it was 92% of 20,646, led by 6,000 at 62 at 1:28 PM. The contract has been bought back from its September 2 close of 51 to the low 60s in the week before the print, and about 123,000 contracts have changed hands on it since this page was first priced.
"Above 3.4%" went the other way and stayed there. It averaged 31 cents on August 22, was 20 by August 26, and has closed between 17 and 27 every day since. Its biggest trade since this page was first priced was a 7,424-contract buy at 25 on Monday evening, September 7, and its busiest day was Wednesday, when 34,474 contracts traded and the largest order was a 4,000-contract sale at 21 at 3:31 PM. "Above 3.5%" is unchanged at 8 cents on the midpoint but not for lack of trying: two buys of 4,894 and 3,846 contracts went through at 10 cents three minutes apart on Wednesday afternoon, and by evening the contract had been sold back to 3. The one rung that did move in the last day is the cool one: "Above 3.2%" last traded 88 against a 94 print twenty-four hours earlier, on 8,394 contracts, and "Above 3.1%" 95 against 97. On Wednesday the cool tail, not the hot one, got bid.
Put the rungs together and the market's distribution looks like this. On August 22 it gave a printed 3.5 about 23 points and a printed 3.4 about 33. On Thursday morning it gives 3.5 about 14 and 3.4 about 38, with 3.3 at 28 and a 3.2-or-lower print at 12, up from 9. The 3.5 rung lost nine points, five of them to 3.4, three to a cool print of 3.2 or lower, and one to 3.3. The market did not narrow onto one number so much as give up on the hot tail. So it moved toward the panel on the two busiest contracts, closing the gap on "Above 3.3%" from 12 points to 8, and stayed away from it on the shape: the panel put 26.5 points on a 3.4 print and spread the rest wide, and the market has 38 on the one number everyone is forecasting.
Why One Decimal Decides Everything
A twelve-month inflation rate looks like a forecast about a year. Eleven of those months are already printed and cannot change. The index stood at 333.918 in July 2026 and at 323.976 in August 2025, both final, both in the same government file. The only unknown is the move from July to August, and every contract on the board turns on that one monthly change.
| If August's Monthly Change Is… | The twelve-month rate becomes… | BLS prints |
|---|---|---|
| 0.00% | 3.069% | 3.1 |
| 0.10% | 3.172% | 3.2 |
| 0.20% | 3.275% | 3.3 |
| 0.30% | 3.378% | 3.4 |
| 0.40% | 3.481% | 3.5 |
| 0.50% | 3.584% | 3.6 |
BLS prints one decimal, so the board is really a set of rounding thresholds. Printing 3.4 or higher takes a monthly change of at least 0.273%, and printing 3.5 takes 0.370%. Less than a tenth of a point of monthly movement separates outcomes that pay out completely differently.
For scale, the last 20 Augusts have a median monthly change of 0.129%. Last August was 0.287%. Repeat it exactly and the twelve-month rate barely moves, which is why the board is priced where it is.
The consensus does not map onto that table directly, and the difference matters. Forecasters quote the seasonally adjusted monthly change, which is the number that leads the headlines, and the consensus for it is 0.4%. The contract settles on the unadjusted index. Last August the adjusted figure was also 0.4%, and the unadjusted index rose 0.287% (BLS, September 2025). If that gap repeats, a consensus month lands at about 0.29% unadjusted, which is a twelve-month rate of 3.365% and a printed 3.4 with 0.015 points to spare over the line. The Cleveland Fed's nowcast, the panel's other anchor, has nudged up from 3.364% on August 20 to 3.38% as of September 9, or 0.03 points over the line. Its median miss when graded at the August 20 point in the month was 0.063; this page has no error record for a read two days out, and it should be tighter. Even so, every independent estimate of Friday's number sits within three hundredths of the 3.3-or-3.4 boundary: the 3.38 nowcast, the consensus mapped onto the unadjusted index at 3.365, and Nowflation's 3.36 sealed call. That was the panel's whole case on August 22 and nothing published since has moved it.
The Two Gasoline Readings The Panel Was Waiting For
This page said on August 22 that the last two weekly gasoline readings of the month were the main input the panel had not seen, and that a full-month average back near July's $4.06 would take about 0.08 points off the monthly change, enough on its own to drop the print under 3.4. Both readings are in, and they went the panel's way.
| Week Ending | U.S. average, regular |
|---|---|
| July 2026 (Full Month) | $4.064 |
| August 3 | $4.211 |
| August 10 | $4.141 |
| August 17 | $4.182 |
| August 24 | $4.085 |
| August 31 | $4.071 |
| August, five weekly readings | $4.138 |
The two late readings came in at $4.085 and $4.071 (EIA), the lowest two prints of the month, and they pulled the simple average of August's five Monday readings down to $4.138. That is a rise of 1.8% on July, not the 2.8% the panel worked with in the third week of August, and on gasoline's 2.895% weight it adds about 0.05 points to the monthly change instead of 0.08. The 0.03 points that came off is, on the table above, the whole distance between the consensus row and the line that separates a printed 3.4 from a printed 3.3.
It is not the whole story, because the CPI does not read the EIA's pump average. Wells Fargo, in Kiplinger's roundup, has gasoline prices "increasing a little over 4%" on the way to a 0.40% adjusted month. It does not say whether that 4% is the adjusted index or the pump, and the difference is the whole argument here: on our read, at least part of the gap between 4% and the 1.8% pump average is the seasonal adjustment, which expects gasoline to ease in late summer and reads a flat month as a rise. Last August the adjusted gasoline index rose 1.9% while the unadjusted price rose 0.3%. The contract settles on the unadjusted side of that gap, which is the side the pump average describes.
One more number from the same file belongs to next month, not this one. The reading for September 7 jumped to $4.157, and diesel went from $5.599 to $5.967 in a week. August's index was measured before any of that happened.
The Wholesale Print That Moved The Fed Board, Not This One
Thursday's other release settled one question about how the market reads inflation news. The producer price index for August rose 0.4%, seasonally adjusted, after 0.1% in July, and 5.4% over twelve months, per the BLS release. Final-demand goods rose 1.1%, with over three-fourths of that from energy, up 4.2%, and over a third of it from diesel fuel alone, which jumped 24.1%. CNN's write-up said the report "could strengthen the case for the Federal Reserve to raise interest rates at its policy meeting next week" (KTVZ/CNN), and the exchange took three minutes to decide, then repriced the decision by nine cents over the next half hour.
The quarter-point hike market inside Kalshi's September 16 decision event, KXFEDDECISION-26SEP, traded at 55 and 56 cents in the minutes before 8:30. It printed 56 at 8:31, 58 by 8:35, 60 at 8:38, 64 at 8:42, 65 at 8:59 and 66 at 9:08, where a single 15,885-contract buy went through. Between 8:30 and 9:15, 139,088 hike contracts traded, and 89% of the 120,358 contracts that changed hands between 8:00 and 9:00 were bought at the ask rather than sold at the bid. The hold contract did the mirror image on twice the size: 298,334 contracts in the same 45 minutes, from 46 to 35, including a 30,000-contract sale at 45 at 8:31 and a 57,794-contract minute at 8:45 with the largest order, 20,852 contracts, sold at 36. One buyer did take 20,000 hold contracts at 46 at 8:30:25, four seconds before the hike contract's first post-release print, and is down twelve cents on them at the bid. As of 9:07 the hike is 64 bid, 65 ask, the hold 34 bid, 35 ask, and a cut of any size has no bid and a penny offered.
The two busiest rungs on the inflation ladder watched all of that and barely twitched. "Above 3.3%" printed 61 at 8:29:56, thirty seconds before the number, printed 61 again at 8:32:40, and every one of the 878 contracts that traded on it through 9:15 went through at 61. "Above 3.4%" traded between 20 and 23 on 877 contracts, and its last print in the window, at 8:58, was 20, three cents under its last trade before the release. A wholesale print hot enough to move the hike nine cents and the hold eleven moved "Above 3.3%" by nothing after the release and "Above 3.4%" by three cents, because the two boards are not the same question. PPI's energy jump is diesel and wholesale margins in the adjusted producer index; the contract on this page is the retail price of a fixed basket, unadjusted, and the part of it that matters for August was measured at pumps that were flat to lower in the second half of the month. The market read the PPI as evidence about the committee, not about Friday's decimal, and the price history says it read it correctly: crude is the input that can still move the Fed, and it cannot touch August. Brent traded above $105 on Thursday, its highest since May 19 (Trading Economics). Siebert Financial's Mark Malek, in the Kiplinger roundup: "Oil sitting at $100 a barrel, arriving two days before a closely watched inflation print, is not going to make Warsh's job any easier." True for the meeting. Irrelevant to the print.
What A 3.3 Versus A 3.4 Does To The Hike Price
There is no contract that settles on both numbers at once, so the answer has to be built from the two boards. The hike is 64 bid, 65 ask, or 64.5 on the midpoint. The chance of a print of 3.4 or higher, which is "Above 3.3%" on this page, is 60 on the midpoint. If the hike were purely a bet on a hot print, those two numbers would match. The four and a half cents between them is the market's price on the committee hiking anyway on a 3.3, or on the diesel and crude arguments carrying a room that already had three dissents for a hike in July, and it is a price that did not exist on Wednesday, when the hike closed at 54 and "Above 3.3%" at 63.
The monthly board says the same thing in adjusted terms. Kalshi's sibling market on the headline monthly change, KXCPI-26AUG, has "Above 0.3%" at 58 bid, 61 ask and "Above 0.4%" at 12 bid, 14 ask, so it puts about 47 points on a printed 0.4, 31 on a 0.3, 13 on 0.5 or hotter and 9 on 0.2 or cooler. Its coin-flip rung is priced at 60, the same as the twelve-month board's, on a different basis that arrives at the same place: the consensus number, with a slightly better than even chance of getting there. That market closes at 8:25 AM ET, four minutes before this one.
The forecasters who published a case for one side or the other split the way the price history does. BlackRock's weekly commentary, quoted by Kiplinger, had "a hot print likely to tip the balance toward a hike" while still not seeing one as "a foregone conclusion." Glenmede's note from Jason Pride and Michael Reynolds has headline inflation "expected to ease slightly to 3.3% year-over-year from 3.4%," one paragraph below Kiplinger's own 3.4% consensus, which is its own small illustration of how close this sits to the line, and it adds that investors "should resist treating a rate hike this month as a settled outcome, as the case for one continues to rest on an inflation impulse that has been energy-driven and has yet to show convincing evidence of broadening." That is the 35-cent hold in a sentence, and it is a forecast of the print that the board gives 40 cents. The one I keep coming back to is the twelve-month arithmetic: on a consensus month the print clears the 3.35 line by 0.015 points, and a print that clears it by that little is a print that could just as easily not have. Read the 64.5 as the market saying the committee has already made up its mind, and the 60 as the market saying the number that is supposed to confirm it is a coin flip with a slight lean. Both can be right on Friday, and then Wednesday is a separate question, which this site's Fed page has been tracking since August 20.
Where The Panel Disagrees With The Price
The headline gap on "Above 3.3%" is now inside the spread on one side. The quote is 58 bid and 61 offered against the panel's 52%, so the panel's number sits 6 points under the bid and 9 under the ask. The only position its number implies is selling into the 58 for about 6 points, and Kalshi's trading fee, which works out to about 1.7 cents a contract at this price, takes a bite from that. Buying the contract at 61 is the wrong side of the panel by 9. In August the panel's number sat 8 under the bid and 16 under the ask. The market did most of the closing; the panel's number has not changed.
The deeper split is still about the shoulders rather than the center, and it has widened on one of them. The market puts about 12% on a print of 3.2 or lower, the other side of "Above 3.2%" at 86 bid, 89 ask, up from 9% in August. The panel put 23.5% there and still does. On the hot side the two now agree on one number and not the next: the market gives a printed 3.5 about 14 points and the panel 14, but the panel still carries 11.5 on 3.6 or higher against the market's 8. Where the crowd sees a narrow question with one answer, the models see a wide one with a soft floor, and the crowd has spent three weeks moving weight off the hot tail.
The so-what, a day before settlement, is which of the panel's August assumptions survived. Its 23.5% on a cool print was built on gasoline adding 0.08 points to the month; that input is now 0.05, which pushes toward the panel by about three hundredths on the twelve-month rate. The nowcast moved 0.016 the other way. Net, the two roughly cancel, and the panel is riding one argument only: the rounding line, and that argument is intact.
What The Panel Knew On August 22
BLS had not published a single August price when the panel sat, but plenty of August was already observable, and the panel leaned on it. This section is the record of what it had, kept as written.
Gasoline was the live variable. The Energy Information Administration posts the national average for regular gasoline every week, and by August 22 three of the month's five readings were in.
| Week Ending | U.S. average, regular |
|---|---|
| July 2026 (Full Month) | $4.064 |
| August 3 | $4.211 |
| August 10 | $4.141 |
| August 17 | $4.182 |
| August So Far | $4.178 |
That was a rise of 2.8% on July's full-month average. Gasoline is 2.895% of the basket, so the arithmetic added about 0.08 points to the monthly change on its own, which was most of the difference between a typical August and this one. The rise had also flattened across the three prints, and several seats cited that flattening as the reason not to push higher. The two readings still to come, for the weeks ending August 24 and August 31, are covered above.
Underneath the fuel, inflation was cooling. Prices excluding food and energy rose 2.5% over the twelve months through July against 3.4% for the headline, and every bit of that gap is food and energy. Shelter, over a third of the basket, rose just 0.1% in July as it did in June. The panel's read: the core has no upside impulse this month and the energy line is the only thing that can move the number.
An independent model said the same thing. The Cleveland Fed's daily inflation nowcast put the August twelve-month rate at 3.364% as of August 20, which rounds to 3.4 with 0.014 points to spare. Graded against 16 past releases at the same point in the month, the model's median miss is 0.063 points and its worst was 0.478 in June 2026. The uncertainty around the estimate is four times wider than the distance to the line, and that fact is the panel's whole case.
How The Panel Got To A Coin Flip
The revision round moved the panel more than any input did. In the August 22 run the seats ranged from 30% to 56% on "Above 3.3%" in round one with three different modal answers. After reading each other, all eight landed between 51% and 54% on the same one. That collapse is the finding, and it deserves one honest caveat. All eight converged because all eight were reading the same two anchors, the Cleveland Fed's nowcast and the weekly gasoline file, so the blend is tight without being any more likely to be right. July's miss was a different failure, a misread of an energy move rather than shared anchors, but it is the same reason for humility. What the convergence does show is that no seat found an argument for a wide miss in either direction, and three weeks of trading have not found one either.
The biggest move came from DeepSeek, which started with 3.2% as its most likely print and finished with 3.4% on top, citing the decelerating gasoline prints and a nowcast that had been flat for eight updates. Gemini named the mechanism:
… the 3.364% point estimate sits a mere 1.4 basis points above the 3.35% rounding line, while the model's median absolute miss is 0.063 points. This mathematically forces a near-tie between 3.3 and 3.4. — Gemini
Every seat is graded in public against real settlements on the full model verdict scoreboard.
How The Panel Did Last Month
This page carried a panel on the July report, and July has settled, so here is the scorecard rather than a quiet edit.
The July board was the month-over-month version of this market. The panel read June's −0.4% collapse as a one-off gasoline move and put 44% on a July print of 0.2% or hotter against a market price of 17¢. July came in at 0.1% and the contract settled at zero. The market was right and the panel was wrong in a specific way: it treated an energy drop as noise around a hotter trend when the energy move was the trend.
This month the panel's argument is different in kind. It says the crowd is overpricing its own precision on a number that sits on a rounding line. That gets graded on September 11 like everything else.
The Bottom Line
Three weeks ago the market's 64 cents on "Above 3.3%" treated the August print as a narrow question, and the panel said it was a wide one. The market has since taken four cents off that contract, nine off "Above 3.4%", and moved five points off a hot 3.5 onto the 3.4 everyone is forecasting, while the two soft gasoline weeks are already inside the Cleveland Fed's 3.38, which rose 0.016 from August 20 anyway, so the model found something to offset them and its estimate still sits within a few hundredths of the line. The producer prices that lit up the Fed board on Thursday morning did not touch this one, and that is the right read: crude at $105 is September's problem and Wednesday's, and Friday's number was set at pumps in August. The panel's case is the same one it made on August 22, that everyone's estimate sits closer to the rounding line than anyone's error, and the market at 58 to 61 is now saying nearly the same thing with a slightly straighter face.
Our models are graded against real settlements and are wrong regularly, including on this exact market in July. We will publish the grade the morning after the board settles.
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.
More on this: CPI Report Odds: Where Inflation Is Priced · Kalshi Jobs Report Markets: Closed One Minute Before The Number Drops · Trading The CPI Print On Kalshi: How Kalshi Inflation Markets Work · Core CPI Odds: Did July Inflation Cool? The Print Vs. The Market
