Recession In 2026 Odds: The AI Model Verdict Vs. The Kalshi Market
Few questions carry more standing search demand than a plain one: will there be a recession this year? Kalshi has run a contract on exactly that since spring, and traders have a firm answer. The market trades near 8 cents, treating a 2026 recession as a long shot. We put the question to six AI models without letting any of them see that price. They landed right about where the market is, and the interesting part is the one model that did not.
The Quick Answer
Kalshi's "Will there be a recession in 2026?" contract trades near 8¢ (about 8% implied). Our price-blind, six-model panel lands right alongside it, at an 8% blend. The reason is arithmetic: the U.S. economy had exactly one negative quarter in the window, Q1 2025, and it rebounded so hard the next quarter that it never strung two negatives together. With Q2 2026 already in at +1.5%, the only realistic path left to a YES is for both Q3 and Q4 2026 to come in negative. One model makes a real bear case at 15%, and it is below.
- Market Price: near 8¢ YES (~8% implied), as of August 3, 2026
- AI Blend: 8% YES (median about 6.5%) across six models
- The Key Fact: just one negative quarter since 2025 (Q1 2025, −0.5%), and it never became a pair
- The One Live Path: Q3 2026 and Q4 2026 both negative; the Q4 advance (~late January 2027) is the settlement print
- The Catch Worth Knowing: this market settles on the mechanical two-negative-quarters rule, not NBER's official recession call
Why the panel sits near the market, what the lone 15% dot is seeing that the others aren't, the full GDP trajectory, and a plain-English guide to what this contract actually pays out on are all below.
The Market At A Glance
| Market | Will there be a recession in 2026? |
| Venue | Kalshi, a CFTC-regulated event-contract exchange (18+; availability varies by state, as of August 2026) |
| Ticker | KXRECSSNBER-26 (single YES/NO contract) |
| Settles | YES if the Bureau of Economic Analysis reports two consecutive quarters of negative real GDP growth in 2025 or 2026. Resolution is tied to the Advance Estimate of Q4 2026 GDP, expected around late January 2027. |
| Market YES Price | near 8¢ (~8% implied) |
| Panel YES Blend | 8% (six models) |
| Prices As Of | August 3, 2026 |
These are model estimates, not predictions of fact and not financial advice. Kalshi event contracts trade on a CFTC-regulated exchange; you must be 18 or older and in an eligible state to participate. The price above is read from Kalshi's live quote and refreshes as the data develops; the "as of" date carries the timestamp.
What This Market Actually Counts
The ticker carries the letters NBER, and that is the single most important thing to get right before reading any number on this page, because it is misleading. The contract's own settlement rule is one sentence, and it says nothing about NBER:
"If there are two consecutive quarters of negative GDP growth in 2025 or 2026, according to the Bureau of Economic Analysis, then the market resolves to Yes."
The National Bureau of Economic Research is the body that officially dates U.S. recessions, and it does not use a two-quarter GDP rule. Its Business Cycle Dating Committee weighs the depth, diffusion, and duration of a downturn across a range of indicators (real income, employment, industrial production, and spending), and it announces its calls with a long lag, sometimes a year or more after a recession has begun or ended.
This contract ignores all of that. It keys on one thing only: two consecutive quarters of negative real GDP growth per the BEA — the popular "technical recession" shorthand. That can diverge from NBER's judgment in both directions: a technical two-quarter dip that NBER never labels a recession, or an NBER recession that never strings together two negative GDP quarters. When you trade or read this market, you are trading the BEA GDP sign, not the history books' verdict. The panel scored it on exactly that rule.
One Negative Quarter, Never Two In A Row
Here is the fetched ground truth the whole question turns on. These are BEA real GDP figures, quarter-over-quarter at an annualized rate, across the entire settlement window through the Q2 2026 advance estimate released July 30, 2026.
| Quarter | Real GDP (annualized) | Note |
|---|---|---|
| Q1 2025 | −0.5% | the window's only negative print |
| Q2 2025 | +3.8% | rebounded hard the very next quarter |
| Q3 2025 | +4.4% | strong |
| Q4 2025 | +0.5% | the softest positive print |
| Q1 2026 | +2.1% | third estimate |
| Q2 2026 | +1.5% | advance estimate |
The economy contracted exactly once in this window, in Q1 2025, and the BEA attributed that −0.5% to a surge in imports ahead of tariff changes (imports subtract from GDP) plus a drop in government spending, not a broad collapse in demand. Growth snapped back to +3.8% the next quarter. So the single most important thing about this market's history is that the economy has never put two negative quarters back to back, which is the only thing that pays this contract out.
The Q2 2026 internals actually argue against a downturn even as the headline cooled. The BEA reported that growth came from increases in consumer spending, investment, and exports, partly offset by a decrease in government spending. More tellingly, real final sales to private domestic purchasers, the cleanest read on underlying private demand once you strip out inventories and trade, rose +3.9%, up from +1.7% in Q1. The headline slowed; the private-sector engine underneath it sped up.
The Only Path Left To YES
This is where the market's low price stops being a mood and becomes a calculation. The contract needs two consecutive negative quarters somewhere in 2025 or 2026. Walk the calendar:
- The lone Q1 2025 negative (−0.5%) is bracketed by positives (Q2 2025 was +3.8%), so it can never be half of a two-in-a-row pair. Dead.
- Every quarter from Q2 2025 through Q2 2026 is positive, so no pair inside that stretch can be all-negative. Dead.
- Q2 2026 (+1.5%) is positive, so the Q2+Q3 pair cannot be all-negative either.
- That leaves exactly one live pair: Q3 2026 and Q4 2026, both negative.
The Q3 2026 advance estimate lands around late October 2026; the Q4 2026 advance, due around late January 2027, is the print this market settles on. One caveat keeps this honest: Q2 2026 is an advance estimate, and the BEA revises those, so a later downgrade of Q2 below zero is the only thing that could technically reopen a second path. That is a remote tail, not the base case. Strip it away and the entire remaining question is narrow and knowable: will an economy that just grew +1.5%, with private demand accelerating, reverse into two straight negative quarters in the second half of the year? At 8 cents, the market is calling that a roughly one-in-twelve long shot.
The Case For YES: What Would Have To Break
The bear case is not empty, and one model on the panel leaned into it. The June 2026 jobs report, the latest available with July's due August 7, showed nonfarm payrolls up just +57,000, roughly half the ~113,000 economists expected, with leisure and hospitality shedding 61,000 jobs even as the unemployment rate held at 4.2%. Hiring is clearly slowing, and the labor market is often the first domino in a consumer-led downturn.
Layered on top is a real policy squeeze. The FOMC held its target range at 3.50–3.75% on July 29, 2026, a fifth straight hold, on a 9–3 vote in which all three dissents wanted to raise rates, not cut them. The reason is inflation: the Q2 PCE price index rose +5.1%, up from +4.6% the prior quarter. That combination, decelerating growth alongside re-accelerating prices, is the textbook stagflation bind, and it is exactly the setup in which a central bank tightening into weakness can tip an economy over. A Fed that cannot cut because inflation is climbing has less room to catch a stumble. That is the argument for a number above the market's 8 cents, and it is why the panel is not at zero.
The Case For NO: Why The Panel Sits Near The Market
Against all of that sits the arithmetic and the base rate. Economies growing near +1.5% with strengthening underlying private demand almost never flip to two consecutive negative quarters within the very next two prints. When that has happened historically, it has clustered around a clear, sudden shock (the 2008 credit freeze, the 2020 pandemic stop), not a gentle slowdown from a positive baseline. Absent such a shock, the joint probability of two straight negatives from here is a low single-digit event, and the +3.9% jump in real final sales to private domestic purchasers is the opposite of an economy buckling. That is why five of the six models landed between 5% and 10%, essentially confirming the crowd's 8-cent read. The market is not being complacent; it is being arithmetically correct about a narrow remaining path.
Model Verdicts
Market: about 8% · AI blend: 8% (median ~6.5%). The panel confirms the crowd, with one bearish dissenter.
| Model | YES probability | Why |
|---|---|---|
| Claude Fable | 5% | The lone Q1 2025 negative is bracketed by +3.8%, and with Q2 2026 at +1.5% YES needs both Q3 and Q4 negative, a rare abrupt collapse. |
| Claude Opus | 5% | An economy growing +1.5% with core private demand at +3.9% almost never flips to two straight negatives without a sharp exogenous shock. |
| Claude Sonnet | 10% | Only Q3+Q4 2026 both-negative remains, and against a growing base that stays the tail case, not the base case. |
| GLM | 15% | The two-quarter path is a high bar, but a cracking labor market and a Fed that can't ease into a slowdown fatten the left tail. |
| Kimi | 6% | With +3.9% core demand and 4.2% unemployment, a flip to two straight contractions in two quarters needs a major shock. |
| DeepSeek | 7% | Real final sales surged to +3.9%, a strong internal buffer making a sudden two-quarter flip exceptionally unlikely. |
| Blended Verdict | 8% | equal-weight mean of six models (method always disclosed); median ~6.5% |
Model estimates generated August 3, 2026, price-blind, then passed through a second round where each model read the others' anonymized reasoning and could revise. 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 revision round did real work here. Two seats that started higher pulled toward the pack after reading the others' reasoning: GLM came down from an initial 28% and Sonnet from 14% to 10%, both after peers stressed that the +3.9% jump in core private demand is a real buffer against an imminent collapse. It is worth being honest about one limit of that convergence: these are large language models drawing on overlapping training data, so five of them clustering near the same number is partly shared priors, not six fully independent reads. Treat the agreement as a soft confirmation of the arithmetic, not as proof.
Where The Panel Splits
The one number worth staring at is GLM's 15%, roughly double the rest of the panel and nearly double the market's 8 cents. It is not a different reading of the arithmetic; GLM accepted the same one-live-pair math as everyone else, and in fact revised down from 28% once peers pushed back. It simply keeps more weight on the leading indicators pointing down.
"The path to YES requires both Q3 and Q4 2026 negative from a +1.5% base with +3.9% core private demand, historically rare absent a major shock. But the stagflationary combination of accelerating PCE (+5.1%), a cracking labor market (+57k payrolls), and a hawkish-on-inflation Fed is a plausible shock mechanism, so I hold above the 5-to-7 cluster." — GLM
Read GLM as the panel's honest tail-risk voice: the arithmetic makes a 2026 recession unlikely, but "unlikely" is not "impossible," and a labor market cooling this fast while the Fed cannot ease is precisely the kind of setup that, once in a while, snowballs. The other five seats heard the same case and judged most of it already priced. That gap, 15% versus 5%, is the entire live debate on this market, compressed into two numbers.
How To Read This Market
- A Kalshi Price Is A Probability, Roughly. A YES contract at 8 cents pays out a dollar if it hits, so the market is treating a 2026 technical recession as about an 8% chance. Here is a fuller primer on what a Kalshi price actually means.
- The Clock Is The Story. With Q2 already positive, this market can only be decided by the Q3 (late October) and Q4 (late January) GDP prints. A single negative Q3 would make it live; a positive Q3 would settle it NO before Q4 is even released.
- This Is Not The NBER Call. If you care about whether economists officially label 2026 a recession, that is a separate, slower judgment. This contract is strictly the BEA two-quarter GDP rule.
- Every Number Here Gets Graded In Public. Once this market settles, each estimate goes on the full graded scoreboard so you can see whether the panel or the market called it better. These are model estimates, not predictions of fact and not financial advice.
- Eligibility. Kalshi is a CFTC-regulated event-contract exchange, not a sportsbook. You must be 18 or older and located in an eligible state, and availability varies by state and can change.
This market sits in a cluster of macro contracts our panel tracks: the Fed rate-cut market and the next Fed decision price the policy squeeze described above, the monthly jobs report prices the labor slowdown that GLM leaned on, and the tariff-and-stimulus market prices the fiscal side. Read together, they are a probabilistic dashboard of the same economy this contract asks one blunt question about.
New to Kalshi event markets? Start with our plain-English guide to how prediction-market prices work.
FAQ
What counts as a recession for this Kalshi market? Despite the ticker's NBER name, this contract does not use the National Bureau of Economic Research's official recession call. It resolves YES only if the BEA reports two consecutive quarters of negative real GDP growth in 2025 or 2026, the mechanical "technical recession" definition. NBER's committee weighs depth, breadth and duration across many indicators and dates cycles with a long lag; none of that matters here. Only the sign of the BEA GDP prints does.
Has a recession happened in 2026 yet? No. The window has one negative quarter, Q1 2025 at −0.5%, but real GDP rebounded to +3.8% the very next quarter, so it never chained two negatives in a row. Every print since has been positive: Q2 2025 +3.8%, Q3 2025 +4.4%, Q4 2025 +0.5%, Q1 2026 +2.1% and Q2 2026 +1.5% (advance). Because the market needs two consecutive negative quarters, it cannot yet have settled YES.
What would have to happen for this market to resolve YES? Because Q2 2026 already printed positive, the only realistic path is for both Q3 2026 and Q4 2026 to come in negative on the BEA advance estimates. The Q3 2026 advance lands around late October 2026; the Q4 2026 advance, expected around late January 2027, is the settlement print. An economy currently growing about 1.5% would have to swing to two straight negative quarters in the back half of the year.
How were these model estimates produced? Six AI models estimated the probability without seeing any market price, working from a fetched summary of the full BEA GDP series through the Q2 2026 advance, the June 2026 jobs report and the July 29, 2026 Fed statement. Each number was then run through a second round where every model read the others' anonymized reasoning and could revise. The results were blended into one estimate. These are model estimates, not predictions of fact.
To be explicit: nothing here is financial or trading advice, and these are model estimates rather than predictions of fact. Kalshi event contracts are CFTC-regulated; you must be 18 or older and in an eligible state to participate.
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.
