Rocket Classic 2026: 8 AI Models Say The Winner Is Probably Nobody's Pick
By Jake Hari
July 28, 2026
The Quick Answer
Golf's honest answer first: on a 149-man field in a birdie-fest, the panel's revised blend puts 51% on "someone outside the short list", every model's single biggest bucket. Among named golfers, Xander Schauffele leads the blend at 9% against a market price of just 4¢, with Hideki Matsuyama and Patrick Cantlay next. The market's own favorite, Cameron Young at 8¢, sits fourth on the blend. This one grades fast: the tournament finishes Sunday and every number below goes on the scoreboard that night.
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 Market
Kalshi's Rocket Classic winner market (event KXPGATOUR-ROC26) is the biggest sports market on the exchange without coverage — about $17.5M traded. Per-golfer rule, verbatim: "If [golfer] wins the Rocket Classic, then the market resolves to Yes." Markets close August 2. The tournament runs Thursday through Sunday at Detroit Golf Club.
Two timestamps: panel verdicts generated the morning of July 28, 2026, price-blind; prices fetched from exchange candlesticks the same morning.
What The Fetched Record Says
The card carried the full listed field and one fetched ledger: the final leaderboard of the tour's last completed event, the 3M Open, where Jackson Koivun won at -25, and Koivun is in the Detroit field this week, priced at 4¢. The card deliberately did NOT carry world rankings or strokes-gained data (no fetched source available at run time), and the models were told to use their own knowledge for those, flag rather than assume withdrawals, and never infer from the card's silence.
How The Panel Read It
All eight seats independently made "rest of field" their largest single bucket, and the reasoning was uniform: 72-hole golf is the highest-variance forecasting surface in sports, and this is a week where many top stars traditionally sit out.
"Even strong favorites rarely exceed 10-15% in a full-field PGA event, and this field is deep but star-light; the winner is more likely to come from the pack than from any single name I can list." (Claude Fable)
The named tier the panel settled on: Schauffele (9%), Matsuyama (6%), Cantlay (6%): class over form, three of the most decorated resumes on the entry list.
The Revision Round
Each model read the other seven's anonymized reasoning and could revise. Two arcs worth the record:
"One targeted change: moved 3 points from REST_OF_FIELD to Cameron Young, whom peers priced as the field's form horse." (Claude Fable)
"I revised because several peers surfaced Detroit Golf Club's specific historical record of first-time winners and low winning scores, which supports a flatter distribution than I originally drew." (DeepSeek V4)
Two seats (ChatGPT, Claude Sonnet) did not return revisions and are excluded from the round-2 blend; their round-1 numbers stand in the table below.
Where The Machines Split From The Money
Xander Schauffele: market 4¢, AI blend 9%. The panel's biggest gap over the market, and its preferred class name: a two-time major winner priced like one of the pack.
Cameron Young: market 8¢, AI blend 4%. The market's favorite; the panel respects the form but will not make a 8¢ favorite of anyone on this course.
Jackson Koivun: market 4¢, AI blend 3%. Fresh off the 3M Open win per the fetched leaderboard, and the panel actively faded the recency, pricing him below his market number. Back-to-back tour wins are historically rare.
Dark Horses The Panel Won't Dismiss
- Russell Henley (3¢, blend 4%): the revision round's quiet riser; multiple seats moved him up on course-fit reasoning.
- Tony Finau (1¢, blend 3%): several seats cited his Detroit history in their reasoning; at 1¢ the panel keeps him alive.
- Sungjae Im (1¢, blend 4%): volume-of-birdies profile the panel likes for this course at 1¢.
The Board
| Golfer | Kalshi | AI Blend (after revision) |
|---|---|---|
| Any Other Golfer In The 149-Man Field | – | 51% |
| Xander Schauffele | 4¢ | 9% |
| Hideki Matsuyama | 2¢ | 6% |
| Patrick Cantlay | 3¢ | 6% |
| Cameron Young | 8¢ | 4% |
| Russell Henley | 3¢ | 4% |
| Sungjae Im | 1¢ | 4% |
| Jackson Koivun | 4¢ | 3% |
| Tony Finau | 1¢ | 3% |
| Cameron Davis | 0¢ | 1.9% |
| Ben Griffin | 3¢ | 1.6% |
| Aldrich Potgieter | 1¢ | 1.3% |
| Akshay Bhatia | 2¢ | 1.0% |
| Max Greyserman | 1¢ | 1.0% |
| Davis Thompson | 1¢ | 0.7% |
| Keegan Bradley | 1¢ | 0.7% |
| J.J. Spaun | 3¢ | 0.7% |
Every Seat's Round-1 Number
The per-model split before the revision round. Named golfers below the top 16 are omitted for space, so columns sum slightly under 100%. On a phone this table scrolls sideways; the board above carries the headline numbers, which average the six seats that returned revised round-2 estimates.
| Golfer | ChatGPT (GPT-5.5) | Claude Fable | Claude Opus | Claude Sonnet | Gemini 3.1 Pro | GLM 5.2 | Kimi K3 | DeepSeek V4 |
|---|---|---|---|---|---|---|---|---|
| Rest of field | 38% | 46% | 48% | 36% | 49% | 32% | 59% | 30% |
| Xander Schauffele | 6% | 8% | 8% | 8% | 8% | 8% | 6% | 10% |
| Hideki Matsuyama | 5% | 5% | 4% | 6% | 6% | 7% | 6% | 8% |
| Patrick Cantlay | 6% | 6% | 4% | 4% | 8% | 7% | 3% | 5% |
| Cameron Young | 8% | 4% | 5% | 4% | 3% | – | 3% | – |
| Russell Henley | 5% | 4% | 4% | – | 2% | – | 3% | – |
| Sungjae Im | 3% | 3% | 3% | 4% | 4% | 6% | 1.8% | 6% |
| Jackson Koivun | 2% | 4% | 2% | 7% | 2% | 4% | 3% | 7% |
| Tony Finau | – | 2% | – | 4% | 4% | 5% | – | 5% |
| Cameron Davis | – | 2% | – | – | – | 2% | 1.3% | 4% |
| Ben Griffin | 3% | 4% | 4% | 3% | – | – | 3% | – |
| Aldrich Potgieter | – | – | 3% | – | – | – | 3% | – |
| Akshay Bhatia | 4% | 2% | 3% | – | 1.5% | 4% | 1.5% | 3% |
| Max Greyserman | – | – | – | – | – | – | 1.7% | – |
| Davis Thompson | 2% | 2% | – | – | 1.5% | 3% | – | 4% |
| Keegan Bradley | – | 3% | 2% | – | – | – | – | 3% |
| J.J. Spaun | 3% | 2% | – | 5% | – | – | 2% | – |
Every seat on this panel is graded against real market settlements — records to date: GPT 83% on 1,693 graded calls · Claude Fable 85% on 331 graded calls · Claude Opus 86% on 382 graded calls · Claude Sonnet 84% on 363 graded calls · Gemini 85% on 1,718 graded calls · GLM 80% on 1,874 graded calls · Kimi 84% on 1,863 graded calls · DeepSeek 80% on 1,877 graded calls. Recomputed daily; the full scoreboard is public.
Model estimates generated July 28, 2026, price-blind from live fetched data. 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. Kalshi is a CFTC-regulated exchange; 18+, availability varies by state.
The Shape Of The Distribution
Half the probability on "the field" is not a cop-out; it is what golf actually is. The panel names its tier (Schauffele, Matsuyama, Cantlay) and concedes the rest to variance. Unlike the season-long boards on this shelf (RB1, QB1), this one settles Sunday night: every number above hits the public scoreboard the moment the trophy is handed out, wins and losses both.