College Football Undefeated Odds: The AI Model Verdict Vs. The Kalshi Market

Every summer the "will they run the table" market is a bet on hope. No 2026 college football has been played, the season does not kick off until late August, and Kalshi has already split the question into 27 separate team contracts, each asking whether that program will finish the regular season without a single loss. Traders have bid Notre Dame up near 30 cents and stacked another dozen teams behind it. We asked five AI models to score the entire field without letting any of them see a single price, and they came back with a blunt read on the college football undefeated odds: almost every ticket on the board is too expensive.

The Quick Answer

Kalshi's market prices roughly 2.4 undefeated teams into this field. Recent seasons have finished with anywhere from one to four unbeaten teams, so that is a defensible number. Our price-blind, five-model panel is more skeptical, pricing about 1.3 — the low end of the historical range. The result is a board the models fade almost top to bottom.

• Market's Priciest Ticket: Notre Dame, ~30¢, then Texas Tech ~25¢ and Georgia ~20¢
• Model's Most Likely To Run The Table: Ohio State, 12.6% blend (no team clears 13%)
• Biggest Model Fade: Notre Dame, 11.6% blend vs a ~30¢ price
• Biggest Positive Edge (The Models' Only Real Underdog Value): Boise State, 6.2% blend vs a ~1¢ price
• The Honest Frame: the models don't single out one team as wrong; they think the market spread its ~2.4 undefeated tickets too thickly across the mid-priced names, and they are the contrarian, skeptical side of this bet

The three names that carry the story, the two spots where a rich price meets real model skepticism, the full 27-team board, and a plain-English guide to reading these prices are all below.

The Market At A Glance

Venue	Kalshi, a CFTC-regulated event-contract exchange (18+; availability varies by state, as of July 2026)
The Event	"College Football Undefeated Regular Season" for 2026, 27 separate yes/no team contracts (Kalshi event KXNCAAFUNDEFEATED-26)
The Contract	Each market resolves YES only if that team goes undefeated in the 2026 regular season, per Kalshi's settlement rules (zero losses across the roughly 12-game schedule; conference title games sit outside the regular season)
Settles	mid-December 2026, once the regular season is complete
Prices Below As Of	July 30, 2026 (daily-close candlestick snapshot)

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.

What Actually Makes A Team Go Undefeated

Before any team name, the shape of the question matters, and it explains why the models built the board they did. An undefeated regular season is not simply a "who's the best team" contest. It is a survival bet, and two forces decide it:

• Team Strength. A better team is a favorite in more games. As a strength anchor we gave every model ESPN's 2026 preseason SP+ ratings, which rank Ohio State first, Oregon second, Notre Dame third, and Georgia fourth.
• Schedule Difficulty. This is the force the market keeps underrating. A merely good team on a soft slate can be favored in all 12 games; an elite team in a brutal division only needs one off Saturday to lose. Every model landed on the same theme: the SEC and Big Ten heavyweights cannibalize each other, while independents and teams in thinner leagues get a cleaner path.

And one number frames the whole board: how many teams usually run the table. Recent regular seasons have finished with anywhere from one to four unbeaten teams — three in 2023 (Michigan, Washington and Florida State), then just one in 2024 (Oregon) and one in 2025 (Indiana). That puts the honest base rate around one to two a year. The market's 2.4 sits at the high end of that history; the models' 1.3 sits at the low end. That gap is the entire disagreement — and it means the models are the skeptical, contrarian side here, not the side history is backing. No team on their board clears 13%, and the panel's biggest single conclusion is not about one team at all.

Here is the full field, sorted by the model blend, with each model's raw number and the gap against the market price.

Team	Market (implied)	SP+	Opus	Sonnet	GLM	Kimi	DeepSeek	Model blend	Edge vs market
Ohio State	10¢	#1	17	12	11	13	10	12.6%	+2.6
Notre Dame	30¢	#3	15	11	9	12	11	11.6%	-18.4
Miami	14¢	#8	14	8	5	12	9	9.6%	-4.4
Oregon	14¢	#2	13	8	8	8	8	9.0%	-5.0
Texas Tech	25¢	#7	14	9	4	11	7	9.0%	-16.0
Georgia	20¢	#4	11	6	7	8	8	8.0%	-12.0
Indiana	15¢	#5	8	10	5	10	6	7.8%	-7.2
Texas	14¢	#6	9	7	6	4	6	6.4%	-7.6
Boise State	1¢	#39	9	9	6	5	2	6.2%	+5.2
Texas A&M	8¢	#9	7	5	4	4	5	5.0%	-3.0
LSU	6¢	#10	7	4	4	4	5	4.8%	-1.2
Alabama	2¢	#11	8	4	5	3	4	4.8%	+2.8
BYU	5¢	#18	9	6	4	2	3	4.8%	-0.2
USC	5¢	#13	6	3	3	2	3	3.4%	-1.6
Michigan	3¢	#14	6	3	3	2	3	3.4%	+0.4
Penn State	9¢	#17	6	2	4	2	3	3.4%	-5.6
Oklahoma	13¢	#12	4	3	3	2	4	3.2%	-9.8
Tennessee	1¢	#15	6	2	3	2	3	3.2%	+2.2
Ole Miss	4¢	#16	6	2	3	1	3	3.0%	-1.0
Utah	9¢	#25	6	3	3	1	2	3.0%	-6.0
Iowa	5¢	#22	5	2	3	1	2	2.6%	-2.4
Florida	1¢	#19	4	2	2	1	2	2.2%	+1.2
TCU	no trade	#34	4	2	2	1	1	2.0%	—
Navy	3¢	#63	4	2	3	0	1	2.0%	-1.0
Virginia	5¢	#40	2	3	1	1	1	1.6%	-3.4
Army	9¢	#83	3	1	3	0	0	1.4%	-7.6
South Florida	5¢	#80	2	1	2	0	0	1.0%	-4.0

These are model estimates, not predictions of fact and not financial advice. Market figures are raw YES quotes from Kalshi's daily-close candlesticks and include the exchange's bid/ask spread. Because these 27 contracts are independent (multiple teams can go undefeated, or none can), the prices do not need to sum to 100%.

A note on the panel. Our standard Model Verdict lineup runs eight seats. On this run three were unavailable and one was capped: the ChatGPT (codex) and Gemini (agy) seats were not installed on the machine that produced these verdicts, the Grok seat was likewise offline, and the Claude Fable seat hit its usage cap. This board therefore runs on five models: Claude Opus, Claude Sonnet, GLM, Kimi, and DeepSeek. All five scored the field on July 30, 2026, without seeing any prices. With a smaller panel, treat the low-single-digit numbers as noisy and lean on the pattern, not the decimal. Every model's number here also lands in our back-tested Model Verdict ledger, where past price-blind calls are graded against settled results so the method is measured, not just asserted.

The one number that matters most: add up the market's 26 priced tickets and you get about 236 cents — the crowd is paying for roughly 2.4 undefeated teams. The models' blends on those same teams add to about 133, or 1.3 teams. Recent history sits between them and nearer the market (one to four a year), so this one-team gap is the models betting the under, not a settled fact.

With the field in view, three teams carry the story: the model's favorite, the favorite it fades hardest, and the longshot it actually likes.

Ohio State: The Model's Favorite (And The Market Agrees)

If the panel has a "most likely" pick, it is Ohio State at a 12.6% blend, and unusually for this board, the market is right there with it at 10 cents, making this the rare ticket the models grade as fair-to-slightly-cheap (+2.6).

The logic is clean. Ohio State is the SP+ No. 1 team with the highest projected per-game floor in the country, so it will be favored in nearly every game it plays. What caps it is the same thing that caps every Big Ten power: the schedule. As Kimi put it, "the clear SP+ #1 at +31.8 has the highest per-game floor in the country, but a road trip to Texas plus a nine-game Big Ten grind with Oregon/Michigan-class opponents still caps a perfect run near 1-in-8." GLM landed in the same place, warning that "a Big Ten slate featuring Oregon, Penn State, and Michigan" leaves at least two or three real landmine games on the schedule.

The read: Ohio State is the best team on the board and the likeliest to run the table, but "likeliest" here still means a roughly one-in-eight shot. This is the model-and-market overlap, which usually means the least edge and the most defensible number.

Notre Dame: The Favorite The Models Fade Hardest

Here is the sharpest disagreement on the board. Kalshi traders have made Notre Dame the single priciest undefeated ticket at roughly 30 cents. The models blend it to 11.6% — an 18-point fade, by far the widest on the board.

The nuance is what makes this honest: the models like Notre Dame's structural path. It is their No. 2 team by blend, and the reason is exactly what you would guess. As an independent, the Irish control their own schedule and dodge a conference title-game gauntlet. Kimi called it "the cleanest structural path on the board, provided it avoids its annual flat spot (see NIU 2024)" — a nod to the Northern Illinois upset that actually cost Notre Dame a loss in 2024. DeepSeek agreed it is the "highest-rated independent with a manageable schedule lacking a conference gauntlet."

So why fade it? Because agreeing that Notre Dame has the best path is not the same as paying 30 cents for it. A clean path still runs through 12 games, still includes marquee tests (Opus flagged "Miami and USC are real landmines"), and still has to survive the flat Saturday every long unbeaten run eventually meets. The market has priced Notre Dame as if a soft schedule all but assured the run. The models are saying: best path, yes; 30% likely, no.

The case for the market: if you believe any team's schedule is soft enough to make undefeated a real bet, Notre Dame's is the one. The models simply think 30 cents already assumes the run instead of pricing the risk of it.

Boise State: The Only Ticket The Models Buy

If Notre Dame is the fade, Boise State is the mirror image. The market has left the Broncos for dead at about 1 cent. The models blend them to 6.2%, the largest positive gap on the board.

This is a schedule bet, not a talent bet. Boise State is only SP+ No. 39, well behind the blue bloods, but it moved into the rebuilt Pac-12 for 2026 — a conference stripped of Power Four heavyweights. Kimi framed the whole appeal in one line: "Boise State punching above its rating in the new-look Pac-12." A good-not-great team with no elite opponent on the calendar is a live longshot to run the table, and a 1-cent price treats it as impossible.

The case against: SP+ No. 39 means Boise will not be a huge favorite every week, and a soft schedule still has trap games. This is a "cheap lottery ticket the models think is mispriced," not a play they are confident in — note DeepSeek sat at just 2% while Opus and Sonnet were at 9%.

The Traps: Big Fades That Deserve A Second Look

Transparency matters more than a clean story, so here is where the table's edges are real but need context:

• Texas Tech (25¢, blend 9%, -16) is the second-biggest fade, and the models' reasons are bullish. Every seat flagged the "strong team, soft Big 12" profile — Opus called it "one of the softest paths among elite squads," Kimi "the highest schedule-adjusted ceiling outside South Bend." The fade is not skepticism about the path; it is skepticism about paying 25 cents when the panel still tops out at 9%.
• Army (9¢, Blend 1.4%, -7.6) Is The Opposite Error. The market has bid the Black Knights up on their famously soft American Conference slate, but SP+ has them 83rd. The models' message: a soft schedule only helps if you are good enough to win those games, and a bottom-third team is not a favorite just because its opponents are weak.
• Only One Positive Edge Is Real. Boise State (+5.2) is the single contract the models would actively buy over the market. Ohio State (+2.6) is best read as fair value — model and market agree — not an underdog play. And Alabama (+2.8), Tennessee (+2.2), and Michigan (+0.4) sit a hair above their prices, but on a five-model panel those gaps are well inside the margin of error. Treat them as noise, not signal.
• TCU Has No Tradable Price (the contract showed no recent volume), so it carries no edge column. The models' 2.0% blend is there for completeness only.

How To Read A Market Like This One

The models can tell you where they disagree with the crowd, but the number in front of you still has to be worth trading. An undefeated field is a different animal from a "who wins the title" market, and reading it wrong is the easiest way to lose money on a correct opinion.

The de-vig here is not what you think. On a mutually exclusive market (one Heisman winner, one national champion) you strip the vig by summing every contract and rescaling to 100%. Do not do that here. These 27 markets are independent yes/no bets, so the only overround to strip is within a single contract: its YES ask versus its YES bid. Take the midpoint of that spread and you have a fair single-team read. Summing the field, as we did above, does not give you a probability at all. It gives you the market's implied count of undefeated teams (about 2.4), which is exactly the number the models think is too high.

Quick rule of thumb: on a mutually exclusive market, sum-and-rescale to strip the vig. On an independent field like this one, never sum for a probability, only for the expected count. Confusing the two is the fastest way to talk yourself into a bad bet on a correct read.

Where OddsShopper fits, honestly. Line shopping, meaning never taking the first number without checking a fair, vig-free version of it, is the habit that separates sharp bettors from square ones. OddsShopper's tools cover college football game lines, not season-long undefeated futures, so our live odds screen and free college football expert picks do the price-comparison work for you on the weekly slate. The undefeated futures themselves live on Kalshi and other prediction-market platforms, where an undefeated season is simply the top of a team's win-total distribution.

A few honest points before anyone acts on a number in the table above:

• 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.
• Liquidity And Spread. Preseason futures are thin. Many of these contracts show a wide bid/ask gap and light volume (TCU showed none at all), so the "price" is softer than a single number suggests.
• Time And Capital. This market does not settle until December, so any position ties up capital for months with no interim payout.
• Not Advice. Everything here is analysis for information, not a recommendation to trade. These are model estimates, not predictions of fact and not financial advice.

FAQ

What are the 2026 college football undefeated odds on Kalshi? Kalshi lists 27 separate yes/no contracts, one per team, each resolving YES if that team finishes the 2026 regular season without a loss. As of July 30, 2026 the priciest tickets are Notre Dame near 30 cents, Texas Tech near 25 cents, and Georgia near 20 cents. Our five-model AI panel thinks every one of those is too high.

Which team is most likely to go undefeated in 2026, according to the models? The panel's highest blend is Ohio State at about 12.6%, just ahead of Notre Dame at 11.6% and Miami at 9.6%. No team clears 13%. Recent seasons have produced one to four unbeaten teams a year (three in 2023, one each in 2024 and 2025), so the models' implied ~1.3 total is a skeptical read that sits at the low end of the historical range.

Where do the models disagree most with the market? The biggest fade is Notre Dame: a roughly 30-cent price against an 11.6% model blend. Texas Tech (25 cents vs 9%) and Georgia (20 cents vs 8%) are next. The only contract the models clearly prefer to the market is Boise State, priced near 1 cent but blended to about 6% (Ohio State grades roughly fair; every other team is a fade).

Can I trade the Kalshi undefeated market? Kalshi is a CFTC-regulated event-contract exchange open to users who are 18 or older and located in an eligible state. Availability varies by state and changes over time, so check Kalshi's current rules before trading. Nothing here is financial advice.

How were these model estimates produced? Five AI models scored all 27 teams at once without seeing any market prices, working from each team's projected strength and known 2026 schedule. Their per-team numbers were blended into one estimate per team. These are model estimates, not predictions of fact.

The Bottom Line

The market and the models agree the 2026 undefeated race is wide open. They disagree on how many longshots to pay for. The crowd has priced roughly 2.4 undefeated teams into this board; the panel sees about 1.3, the low end of a history that runs one to four a year. Ohio State is the models' pick to run the table and the one price they call fair; Notre Dame is the favorite they fade hardest, best path on the board but not a 30-cent lock; and Boise State is the only ticket they would rather own than sell. If this method is your thing, our college football national champion model verdict and 2026 Heisman Trophy model verdict score every contender the same price-blind way. In a July futures market, respecting the base rate is the sharpest position of all.

Verdicts generated July 30, 2026. Prices as of July 30, 2026 (daily-close candlesticks). Panel: Claude Opus, Claude Sonnet, GLM, Kimi, and DeepSeek, price-blind. These are model estimates, not predictions of fact and not financial advice.