July CPI Report Odds: Will Inflation Rebound?
Kalshi's July CPI report odds turn one number, the July inflation print, into a ten-rung ladder of yes/no contracts, each asking whether the headline Consumer Price Index rose by more than a given amount last month. The crowd has settled on a tidy answer. It prices a modest rebound to about +0.1% after June's surprise decline, and it puts little weight on anything hotter. We asked an AI model panel to price the same ladder without letting any of it see a quote, working only from a fetched card of BLS data and the current economist consensus. The panel agrees on direction. It disagrees, hard, on how hot.
New to Kalshi's economic contracts? Start here: how Fed, CPI and jobs markets work.
The Quick Answer
The July CPI report lands Tuesday, August 12 at 8:30 AM ET, and the economist consensus is a +0.1% monthly headline print, a rebound from June's -0.4%. Kalshi's market and our price-blind panel both expect a positive number. Where they part ways is the size of it:
- Market's Single Most Likely Print: +0.1% (about 44% implied), with 0.0% the main downside at roughly 27%. The crowd prices a return to modestly positive and mostly stops there.
- Panel's Single Most Likely Print: also +0.1%, but with far more weight stacked above it. The blend puts about 23% on a +0.2% print and another 21% on +0.3% or hotter.
- The Biggest Gap On The Board: the "Above +0.1%" contract (a print of +0.2% or hotter) trades at 17¢ while the panel blends to 44%, a 27-point disagreement. The panel thinks the market is underpricing a hot July.
- Where They Agree: the floor. Both see roughly a 9-in-10 chance the print is 0.0% or better, so the deep-negative rungs are close to settled.
Free: The Weekly PM Market Brief — the 8-model panel's graded record, the week's biggest market-vs-model gaps, and what's spiking next. One email, Sundays. Sign up in the box at the end of this article.
The full ladder, every model's number, where the panel split from the crowd, and a plain-English guide to reading these prices are all below. Our free expert picks are refreshed daily in the meantime.
The Market At A Glance
| Venue | Kalshi, a CFTC-regulated event-contract exchange (18+; availability varies by state, as of August 2026) |
| The Event | "CPI in July" (Kalshi event KXCPI-26JUL), 10 yes/no contracts, one per threshold |
| The Contract | Each market resolves YES if the July 2026 headline CPI-U rose by more than the stated amount, per Kalshi's contract terms: the signed one-decimal, seasonally adjusted month-over-month change published by the Bureau of Labor Statistics. First print governs; later revisions do not count |
| Settles | Tuesday, August 12, 2026, on the 8:30 AM ET BLS release (the market closes 8:25 AM ET) |
| Prices Below As Of | August 10, 2026 (latest daily-close snapshot; the core rungs are liquid, with six-figure open interest, while the deep-positive tail is thin) |
The panel never saw any of these prices. It read a fetched summary of the BLS seasonally adjusted CPI series, the composition of June's decline, and the published July consensus, then priced all ten thresholds from one internal view of the month. We line its numbers up against the market only after the fact.
What June's -0.4% Was Really About
To read July, start with the print right before it. Headline CPI fell 0.4% in June, its sharpest monthly drop since April 2020, and the entire move was energy. The energy index fell 5.7% on the month, its largest one-month decline since April 2020, with gasoline down 9.7%, unwinding roughly a quarter of a spring run-up tied to a Middle East conflict. Strip energy out and the picture flips: core CPI was flat for June, and food and shelter kept advancing.
That composition is the whole ballgame for July. A one-month, energy-only plunge tends not to repeat. If gasoline simply stops falling, the drag that produced -0.4% fades, and the underlying trend, still-rising shelter plus tariff pass-through working through goods, reasserts itself. That is exactly the logic behind the consensus +0.1%. It is a rebound-toward-trend call, not a call for renewed acceleration.
The backdrop raises the stakes on the number. The 2026 spring ran hot: CPI printed +0.9% in March, +0.6% in April and +0.5% in May before June's reversal. And the Fed's next move is very much live. Kalshi's September FOMC market leans to a hold (59¢) over a 25-basis-point hike (41¢), with a cut priced near zero. In that environment, a hot July CPI is not a footnote. It is the input traders will read straight into the odds of a September hike.
Why The Panel Reads July Hotter Than The Crowd
Hand the panel the same June autopsy and the same consensus, and it lands in a different place than the market. Five of the eight seats returned this cycle, and every one of them prices a positive July print, but they collectively push more probability into the +0.2% and +0.3% buckets than the crowd will.
The reasoning is consistent across the seats. They treat June's -0.4% as almost entirely a gasoline event, so they expect a cleaner snap-back once that specific drag is gone. On top of that they weight two forces the June headline masked: shelter, which never stopped rising, and tariff pass-through into goods prices, which the spring's +0.5% to +0.9% run showed is real and still in the pipeline. Put a full energy reversal on top of firm core, and the models' upper tail fattens well beyond the market's. GLM is the clearest expression of that view, pricing a print of +0.2% or hotter as better than a coin flip; even the more conservative seats leave far more mass above +0.1% than the crowd does.
This is the honest way to read the gap: the crowd is pricing the consensus tightly, and the panel is pricing a fatter right tail around it. The panel is not smarter about August 12 than the market; it has no information the market lacks, and the market has real money behind its view. What the split tells you is where the disagreement actually lives. It is not about whether inflation turned positive again. It is about whether July merely climbs back to +0.1% or overshoots it.
How To Read One Contract
Each rung is a yes/no contract that pays out at 100¢ if it resolves YES, so the price is roughly the market's probability. Take the "Above 0.0%" contract at 61¢: the crowd is saying there is about a 61% chance July CPI prints +0.1% or higher (a print of exactly 0.0% or below loses). Our panel reads that same contract at 69%, an 8-point lean toward YES. Now step up one rung to "Above +0.1%" at 17¢: the market prices only a 17% chance of a +0.2%-or-hotter print, while the panel blends to 44%. That single gap, the price of a hot July, is the widest disagreement on the board and the reason to read the whole ladder rather than one headline number. Because the rungs are cumulative, the difference between two neighboring prices is the odds of landing exactly in that bucket: 61¢ minus 17¢ means the market puts about 44% on a print of exactly +0.1%.
The Board
The narrow view first, built to fit a phone. Every rung is cumulative: a higher threshold is strictly harder to clear.
| Threshold (Monthly) | YES needs a print of | Market | AI blend |
|---|---|---|---|
| Above -0.4% | -0.3% or higher | 99¢ | 99% |
| Above -0.3% | -0.2% or higher | 98¢ | 96% |
| Above -0.2% | -0.1% or higher | 96¢ | 92% |
| Above -0.1% | 0.0% or higher | 88¢ | 85% |
| Above 0.0% | +0.1% or higher | 61¢ | 69% |
| Above +0.1% | +0.2% or higher | 17¢ | 44% |
| Above +0.2% | +0.3% or higher | 4¢ | 21% |
| Above +0.3% | +0.4% or higher | 1¢ | 9% |
| Above +0.4% | +0.5% or higher | 1¢ | 3% |
| Above +0.5% | +0.6% or higher | 1¢ | 1% |
Every Model's Number
The full board, every returning seat pricing the whole ladder from one distribution, after a revision round in which each model read the others' anonymized reasoning and could adjust.
| Threshold | Market | Opus | Sonnet | GLM | Kimi | DeepSeek | AI blend | Edge |
|---|---|---|---|---|---|---|---|---|
| Above -0.4% | 99¢ | 98 | 98 | 99 | 99 | 99 | 99% | 0 |
| Above -0.3% | 98¢ | 95 | 96 | 96 | 98 | 97 | 96% | -2 |
| Above -0.2% | 96¢ | 90 | 90 | 93 | 94 | 94 | 92% | -4 |
| Above -0.1% | 88¢ | 80 | 79 | 89 | 83 | 92 | 85% | -3 |
| Above 0.0% | 61¢ | 66 | 61 | 80 | 65 | 75 | 69% | +8 |
| Above +0.1% | 17¢ | 42 | 38 | 57 | 42 | 40 | 44% | +27 |
| Above +0.2% | 4¢ | 20 | 19 | 29 | 21 | 15 | 21% | +17 |
| Above +0.3% | 1¢ | 9 | 8 | 13 | 8 | 6 | 9% | +8 |
| Above +0.4% | 1¢ | 4 | 3 | 5 | 2 | 2 | 3% | +2 |
| Above +0.5% | 1¢ | 2 | 1 | 2 | 1 | 1 | 1% | 0 |
Model estimates generated August 10, 2026, price-blind, each model pricing the full ladder from one distribution. The panel is eight models; five returned this cycle (Claude Opus, Claude Sonnet, GLM, Kimi, DeepSeek), and three were unavailable for this run (Claude Fable, ChatGPT, Gemini). These are model estimates, not predictions of fact and not financial or trading advice. Models are frequently wrong; the market price reflects real traders' money.
The One-Number Read
Collapse each side to the print where its probability crosses 50%, and the surprise is how much the two agree on the center:
- The Market's Implied Median Is +0.1%. It clears the +0.1% floor (61¢ at "Above 0.0%") but falls off a cliff at the next rung (17¢ at "Above +0.1%").
- The Panel's Implied Median Is Also +0.1%. It just refuses to fall off the same cliff: 69% clears the floor and 44% keeps going to +0.2% or better.
So the disagreement is not about the middle. Both call +0.1% the single most likely outcome. The disagreement is entirely in the shape of the upside. The market treats +0.1% as a near-ceiling; the panel treats it as a midpoint with a real chance of an overshoot. The "Above +0.1%" contract at 17¢ is the whole story of this board. If tariff pass-through and a full energy reversal show up together on August 12, the panel's 44% looks right and that contract is cheap. If July is the clean, contained +0.1% the consensus expects, the crowd's 17¢ is correct and the panel is paying for a tail that never arrives.
More live boards from the same panel: the closest neighbor is the July core CPI ladder, which settles on the same August 12 report and where the market prices core at about +0.2% ("Above +0.1%" trading near 70¢); and the Fed's September decision, split between a hold (59¢) and a 25-basis-point hike (41¢) that this CPI print feeds directly. Prices fetched August 10, 2026.
Where The Panel Changed Its Mind
The models priced the ladder once, then read each other's anonymized reasoning and could revise. Most held, and the one real move was toward caution, not conviction:
- Kimi Cooled Off. It cut its "Above 0.0%" number from 71% to 65% after peers pushed back on how much of June actually reverses: "the heavier negative tails flagged what I underweighted: June unwound only about a quarter of the spring run-up." It is the clearest revision on the board, and it moved the panel's center slightly down, not up.
- GLM Held Its Hot Call. Presented with more conservative peers, it did not blink: "peers used the same drivers I already weighed but assigned more conservative upside." GLM remains the seat most willing to price a +0.2% print as the base case.
- DeepSeek Held Too, And Said Why: "peers' reasoning echoes mine, and no new evidence suggests a stronger rebound than I already priced."
The takeaway from the revision round is that the panel's hotter-than-market read is robust. Reading each other's work barely moved the blend, and the only meaningful edit pulled toward the crowd, not away from it.
What Would Change The Panel's Mind
This board turns on a small number of concrete, datable inputs. Each of these pushes the number a specific direction:
- Gasoline Prices In July. A second straight monthly leg down in gasoline pulls the print back toward 0.0% or negative and validates the market's tight +0.1% ceiling. A flat-to-higher pump price is what the panel's hot tail is built on.
- The Shelter Line In The August 12 Report. Shelter is the largest core component. If its monthly pace re-accelerates, +0.2% becomes the base case; if it finally cools, the consensus +0.1% holds.
- Tariff Pass-Through Into Core Goods. The spring's +0.5% to +0.9% prints showed tariffs feeding goods inflation. Fresh evidence of that in the July core is the single most direct route to the panel's upside.
- Any BLS Data-Collection Disruption. June's series already carries an October 2025 gap. A delayed or curtailed release would change how, and when, this settles far more than any single component.
The Tails Worth Watching
The middle of this board gets the attention, but two low-probability outcomes are worth naming because at least one seat gave each of them a real number:
- A Second Negative Print. The panel puts roughly 15% on July going negative again (a print of -0.1% or lower) and about 8% on -0.2% or worse. That is the "gasoline keeps falling" world, and it is not zero. Sonnet and Opus carried the fattest downside on the board, each leaving about a fifth of their distribution below zero, while DeepSeek was the most confident the print stays flat or positive.
- A Real Hot Shock, +0.3% Or More. The blend puts about 21% on a print of +0.3% or higher (the "Above +0.2%" rung), and GLM alone prices it near 29%. That is the world where energy fully snaps back off a low base while tariffs and shelter bite at once, the same collision that produced the +0.9% March print earlier this year.
How These Prices Settle
| Settles On | Tuesday, August 12, 2026, 8:30 AM ET, when BLS publishes the July CPI report (market closes 8:25 AM ET) |
| Next Catalyst | The August CPI report, released September 11, 2026 |
| Then | The Fed's September FOMC decision, September 16, 2026, which this print feeds directly |
| Re-Scored | Whenever the story moves; this board is stamped as of August 10, 2026 |
Every number in this piece gets graded in public once the market settles. You can see the panel's full track record on the model verdict scoreboard, which is where this board's July call will land the morning after the report.
FAQ
When is the July 2026 CPI report released?
Tuesday, August 12, 2026, at 8:30 AM ET, when the Bureau of Labor Statistics publishes the July Consumer Price Index. The Kalshi market closes five minutes earlier, at 8:25 AM ET, so trading stops just before the number drops.
How is the Kalshi CPI market settled?
On the headline all-items CPI-U, seasonally adjusted, month-over-month, rounded to one decimal, exactly as BLS first publishes it. The contract terms specify that later revisions do not count: whatever number appears in the first release is what settles every rung.
What is the difference between this and the core CPI market?
This board is headline CPI, which includes food and energy, the two most volatile components and the reason June printed -0.4% on a gasoline plunge. Core CPI strips both out and moved very differently in June (it was flat). If you want the underlying-trend read, see the separate July core CPI ladder.
Why does the AI panel disagree with the market?
Both expect a positive July print. The panel simply puts more weight on a hotter outcome (+0.2% or above) because it treats June's drop as a one-off gasoline move and expects shelter and tariff pass-through to reassert once that drag fades. The market prices a tighter rebound to +0.1%. Neither has information the other lacks; the gap is a difference in how the same facts are weighed.
To be explicit: these are model estimates for information and entertainment, not financial advice and not a recommendation to trade any contract. Our models are graded against real settlements and are wrong regularly; on this board the panel is openly taking the other side of the crowd on the upside, and one of them will be right. Event contracts are offered on Kalshi, a CFTC-regulated exchange, for adults 18 and older where legal, and availability varies by state.
Hottest Prediction Markets Right Now
- 2028 Democratic presidential nominee · $166M traded
- Rocket Classic Winner · $124M traded
- 2027 Pro Football Champion · $51M traded
- Pro Baseball Champion · $40M traded
- Bitcoin price at the end of 2026 · $30M traded
- 2027 Pro Basketball Champion · $19M 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.
