At 10:33 p.m. Eastern on Friday, two days before the Miami Dolphins had played a snap, someone sold more than 5,300 contracts of "Miami wins at least five games" into Kalshi in a single second. The order ate through every bid in the book, from 36 cents down to 8, before it was filled. By Saturday morning the quote was back around 32 bid, 35 asked.
Then Sunday happened. Miami lost 27-13 at Las Vegas, Malik Willis was sacked five times, and by Monday morning that same contract was quoted at 25 cents bid and 32 asked. On Thursday it had been 40 and 43.
We gave Miami's entire season to seven AI models and hid every price from them. They saw the roster, the transactions, last year's game log, Sunday's box score and 15 seasons of what happens to teams in this spot. All seven called this a rebuild. Most of them still came back higher than the market.
Updated September 14, 2026 · 12 min read · by Jake Hari
How Miami's Ladder Sold Off In Four Days
Kalshi does not post one Dolphins win total. It lists a separate yes-or-no contract for every threshold, from one win up to a perfect season. Each pays $1 if Miami gets there and nothing if it does not, so a price of 28 cents reads as a 28% chance. The 17-win contract is already settled at zero, because Miami lost a game. Across the ladder, 248,509 contracts have traded and about 140,000 are still open.
The middle of that ladder is where the money sits, and it moved three times in four days:
- Thursday, Midday. The 5-win contract sat at 40 bid, 43 asked. The 4-win rung was 55 and 57.
- Friday Night. A 2,532-contract sale at 8:06 p.m. ET pushed the 5-win rung from 42 to 40. At 10:33 came the one-second sweep that cleared the book.
- Sunday, Before The 4:25 P.m. Kickoff. 5 wins: 33 bid, 37 asked. 4 wins: 50 and 55.
We found no Friday injury report or transaction that lines up with the selling. Whoever sold did it before the game, and Sunday gave them a result to point to.
The Week 1 loss did the rest. Miami trailed 17-3 and cut it to 17-13 early in the third quarter. A 50-yard return on the ensuing kickoff then set up the touchdown that put it away, per the AP recap. The Dolphins converted 2 of 10 third downs, scored one touchdown on three red-zone trips and gained 259 yards. Willis threw for 220 yards with an interception and ran for a touchdown. ESPN's Marcel Louis-Jacques counted 17 pressures on Willis and wrote that no Miami receiver besides rookie Caleb Douglas (five catches, 94 yards) got open with any regularity.
Stay ahead of the markets.
Daily insights and expert picks on Kalshi, Polymarket, and what's moving markets.
Free forever. Unsubscribe anytime.
The Team Miami Took Apart
The price makes more sense once you list what left. Last season's Dolphins went 7-10, scored 347 points and allowed 424. That differential of -77 fits a 6.5-win team, so 2025 was already slightly flattering.
Since then:
- Tua Tagovailoa Was Released In March with a post-June 1 designation. That left Miami with $99.2 million in dead money, an NFL record, according to UPI. The previous record was the $85 million Denver took on for Russell Wilson.
- Jaylen Waddle Was Traded To Denver for picks 30, 94 and 130. He led the 2025 team with 64 catches for 910 yards.
- Tyreek Hill Was Released. Bradley Chubb, the 2025 sack leader with 8.5, was cut. Minkah Fitzpatrick went to the Jets for a seventh-round pick. Linebacker Tyrel Dodson, who led the team with 129 tackles, was released.
- The Front Office And Coach Are New. Chris Grier is out; Jon-Eric Sullivan is the general manager and Jeff Hafley made his NFL head-coaching debut on Sunday.
- Willis Signed For Three Years And $67.5 Million, per Pro Football Rumors' offseason review. He appeared in four games for Green Bay last season and threw for 422 yards.
ESPN reports the total dead cap at $180 million and calls this the youngest roster in the NFL. All 13 draft picks made the initial 53. We checked each current Dolphin against the team he actually played for last season: 16 came from other clubs, and 17 more did not play an NFL regular-season game in 2025. ESPN's own matchup predictor made Miami the underdog in all 17 games before the season, with its best single-game chance at 33% against the Jets in Week 12.
That is the team the market is pricing, and the live question on the ladder is how many games it can still take.
What 38 Similar Teams Did Next
Base rates help here, but only if the comparison is honest. We pulled every NFL team-season from 2011 through 2025, all 32 teams, and counted how often teams reached a given win total, adjusted to a 17-game schedule.
| Group | Teams | 4+ wins | 5+ wins | 6+ wins | 7+ wins |
|---|---|---|---|---|---|
| Every Team | 480 | 92.3% | 84.6% | 77.3% | 69.6% |
| Lost Week 1, Coming Off A Losing Season | 123 | 82.1% | 69.1% | 56.9% | 45.5% |
| Lost Week 1 By 14 Or More, Off A Losing Season | 38 | 78.9% | 63.2% | 57.9% | 39.5% |
The last row is Miami's exact shape on paper. Those 38 teams reached five wins almost two times in three. Kalshi's 5-win midpoint sits well below that rate.
The rate is accurate for the teams it describes, and most of them were ordinary losing teams that still had the core of a six- or seven-win roster. A win-loss table cannot pick out a team that replaced its quarterback, its top two receivers, its sack leader, its head coach and its general manager in one offseason. Every model on the panel started from that table and then marked Miami down from it, which is where the argument actually happens.
The Board
We handed the ladder to seven AI models: Claude Fable, Claude Opus, Claude Sonnet, ChatGPT, Gemini, Grok and Kimi. None saw a price. Each built one forecast of Miami's final win total and read every rung off it, then read the other six models' reasoning with the names hidden and chose whether to revise. The standing panel has eight seats, and seven of them ran this board. The table below lists each rung through 8 wins with Kalshi's live quote, the quote just before Sunday's kickoff, and the middle answer of the seven models.
| Wins | Kalshi midpoint now | Bid / ask now | Before kickoff | Panel |
|---|---|---|---|---|
| 1+ | 92.5¢ | 87 / 98 | 90 / 98 | 96.5% |
| 2+ | 72.5¢ | 62 / 83 | 76 / 90 | 88% |
| 3+ | 61.5¢ | 59 / 64 | 70 / 72 | 73% |
| 4+ | 46¢ | 45 / 47 | 50 / 55 | 53% |
| 5+ | 28.5¢ | 25 / 32 | 33 / 37 | 35% |
| 6+ | 15.5¢ | 12 / 19 | 25 / 29 | 21% |
| 7+ | 12¢ | 11 / 13 | 15 / 23 | 12% |
| 8+ | 13¢ | 1 / 25 | 10 / 15 | 6.5% |
Kalshi quotes as of Monday, September 14, 2026, 8:57 a.m. ET; "before kickoff" is the last hourly quote before 4:25 p.m. ET Sunday. The panel figure is the seat median of the seven models after the revision round. These are market prices and model estimates, not predictions of fact and not financial advice.
Read the 4-win and 5-win rows first, because those two contracts hold 82,000 of the ladder's open contracts. The panel is higher than the market's midpoint on both: 53% against 46 cents for four wins, 35% against 28.5 cents for five. At seven wins the two sides meet at 12. The 8-win row is the odd one: a 1-cent bid and a 25-cent ask, so its 13-cent midpoint is too wide a quote to read as a price. The rungs from 9 wins up have the same problem, which is why they are left off.
Add the rungs together and you get an expected win total. Using the midpoint through seven wins and the real bids above that, Kalshi's ladder comes to about 3.3 wins. The panel's comes to 3.9.
Every number in this piece gets graded in public once the season settles, on the full graded scoreboard.
Every Seat's Number
The per-model ladders, after the revision round, show how wide the disagreement runs underneath that middle answer.
| Model | 3+ | 4+ | 5+ | 6+ | 8+ | Median wins | Teardown cost it charged |
|---|---|---|---|---|---|---|---|
| Fable | 75% | 57% | 39% | 25% | 8.4% | 4 | -3.5 |
| Opus | 77% | 60% | 42% | 27% | 9% | 4 | -3 |
| Sonnet | 79% | 62% | 44% | 28% | 8% | 4 | -3.5 |
| ChatGPT | 62% | 39% | 21% | 10% | 2% | 3 | -2.5 |
| Gemini | 71% | 49% | 31% | 17% | 3.5% | 3 | -3.5 |
| Grok | 73% | 53% | 35% | 21% | 6.5% | 4 | -4 |
| Kimi | 62% | 41% | 26% | 13.5% | 3.5% | 3 | -3.5 |
| Panel Median | 73% | 53% | 35% | 21% | 6.5% | 4 | — |
| Kalshi Midpoint | 61.5¢ | 46¢ | 28.5¢ | 15.5¢ | 13¢ | — | — |
Every seat on this panel is graded against real market settlements — records to date: Gemini 86% on 9,496 graded calls · Grok 84% on 7,920 graded calls · Kimi 86% on 4,894 graded calls. Recomputed daily; the full scoreboard is public.
The split is clean. The three Claude models and Grok put Miami's median season at four wins and the 5-win rung between 35% and 44%. ChatGPT, Gemini and Kimi put it at three, and ChatGPT's 21% on five wins is the only seat below Kalshi's bid on that rung. Asked directly, every seat put the offseason's cost between 2.5 and 4 wins against last year's team; the seats started from different baselines (some from the 7-10 record, some from the 6.5-win point differential), so that column does not subtract cleanly from seven. They agree on what Miami lost and disagree about how many soft games are left on the schedule.
More live boards from the same panel: NFL worst record odds, where Miami became the favorite on Sunday and trades at 29¢ bid, 40¢ asked. NFL win totals 2026, all 32 ladders after Week 1. Giants win total 2026, where 8+ wins is 57¢ bid, 65¢ asked. Prices fetched September 14, 2026.
Hottest Prediction Markets Right Now
- 2028 Democratic presidential nominee · $217M traded
- 2027 Pro Football Champion · $106M traded
- Pro Baseball Champion · $72M traded
- 2028 U.S. Presidential Election winner? · $66M traded
- Fed decision in Sep 2026? · $66M traded
- 2028 Republican presidential nominee · $64M 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.
Where The Panel Changed Its Mind
The revision round moved three of the seven models, and the biggest move was a model catching its own arithmetic.
Opus opened at a median of five wins, with 59.5% on the 5-win rung, the highest number on the panel. After reading the others, it came down to 42%:
"A per-game win chance near 0.19 is consistent with ESPN's preview, where Miami's best single game is 33%. My round-1 median of 5 needed about 31% per game, which is higher than that best game. That was an internal inconsistency, not a matter of conformity. I did not go all the way to B's 3." — Opus
Gemini went the other way. It opened with 22% on five wins, a point above ChatGPT at the bottom of the panel, and moved up to 31%:
"I realized my initial drop of the 5+ rung to 22% effectively double-counted the offseason teardown penalty." — Gemini
Kimi made the smallest move, nudging its middle rungs up by a point or less after another seat pointed out that Kansas City and Cincinnati were 6-11 teams last season.
Fable held every number, and its reason is the cleanest statement of the panel's view:
"It is a smaller teardown charge than the record $180M dead-cap purge and simultaneous QB/HC/GM/WR1/WR2 churn support, and I already price Willis's rushing inside 'replacement, not worse.' No peer surfaced a fact outside the card, so I keep my ladder unchanged." — Fable
ChatGPT held too. It is the most pessimistic seat on the panel and the only one below Kalshi's bid on five wins. It argued that the 33% Jets game was Miami's best chance all season, not a typical one. It also said Sunday confirmed the pass-protection and receiver problems more than it revealed anything hopeful.
Where The Four Wins Come From
Every model named nearly the same short list of games, and that list is the whole case for the panel's three- and four-win medians.
Eight of Miami's 16 remaining games are against teams that won 10 or more last season: San Francisco, New England twice, Buffalo twice, Denver, Chicago and the Chargers. The other eight are the softer half of the schedule on paper. Kansas City (September 27) and Cincinnati (October 11) both come to Miami after 6-11 seasons. The Jets, who went 3-14, get Miami twice. Minnesota and Detroit went 9-8 last year, Indianapolis 8-9 and Green Bay 9-7-1.
Traders price those Jets very differently from last season's record. Kalshi's contract on the Jets reaching five wins trades at 70 bid, 72 asked, more than twice Miami's price on the same rung. That is the market saying the two Jets games are not the gift they look like on a 2025 schedule.
Week 2 does not help either. The contract on Miami beating San Francisco on Sunday trades at 10 bid, 12 asked.
New to season-long contracts like these? Our guide to betting NFL futures on Kalshi walks through how they pay and when. The sportsbook side of the same number lives on our Miami Dolphins win total over/under page.
What Would Change The Panel's Mind
Four things, each one checkable, each with the direction it pushes.
- The Pressure Count On Willis Over The Next Two Weeks. If Sunday's 17 pressures and five sacks repeat at San Francisco on September 20 and against Kansas City on September 27, the four-win seats lose their main argument. Pushes down.
- The Two Home Games Against 6-11 Teams. Kansas City and Cincinnati are the games every seat counted. A win in either one before the Week 6 bye supports the panel's upper seats. Losing both pulls the panel toward ChatGPT's three-win median. Pushes both ways, hardest.
- A Quarterback Change. Willis is the starter on a three-year deal. Behind him on the roster are Kyle McCord and Brady Cook, and ESPN lists Cook as out. An injury to Willis, or a benching in favor of either backup, removes the running threat several seats credited. Pushes down.
- October 25 At The Jets. It is the first of the two Jets games, which every seat counted among Miami's likeliest wins. A Miami win there after a 1-4 or 2-3 start is the clearest sign the models read the Jets correctly. Pushes up.
Settlement Timeline
The contract settles on regular-season wins alone, and the dates below are the ones that move it.
| Settles | January 18, 2027, after the 17-game regular season (ties are not wins; playoff games do not count) |
| Next Data Point | Sunday, September 20, 4:25 p.m. ET at San Francisco |
| The Two Home Games Every Seat Counted | Kansas City, September 27 · Cincinnati, October 11 |
| Bye | Week 6 |
| The Jets, Twice | October 25 at New York · November 29 at home |
| Season Finale | Week 18 at New England, time to be set |
This page is re-scored when the story moves. Prices last read September 14, 2026, 8:57 a.m. ET.
More on this: The Board Is on Barkley, The Trade Is Goedert. · Can Five AI Models Beat The Market On Kalshi Player Props? · 49ers vs Rams Kalshi Player Props: Who Gets Fed In Melbourne · NFL Week 1 Kalshi Player Props: Six Contracts That Stick Out · Patriots vs Seahawks Kalshi Player Props: Seattle Gives Up Catches, Not Yards
The Bottom Line
The Dolphins cut their quarterback, traded or released their best receivers and pass rusher, replaced the people making the decisions and are now carrying more dead money than any team in NFL history. Seven models that never saw a price agreed on all of it and put Miami's season at three or four wins.
Traders went lower. The selling started on a Friday night with nothing in the news, and the loss at Las Vegas gave it a reason on Sunday. At a 28.5-cent midpoint, five wins is priced below every Claude seat, Grok and Gemini, and above only ChatGPT and Kimi.
September 27 against Kansas City is the first home game on every seat's short list, and the first chance to see which ladder read Miami better.
