Sunday was supposed to be the afternoon the 2026 IndyCar season stopped being interesting. Álex Palou put the No. 10 Chip Ganassi Racing Honda on pole for the series' first race in Washington, D.C., his seventh pole of the year, on a street circuit, the surface where he had already won three of his five starts this season. He is a four-time champion, in 2021, 2023, 2024 and 2025. He has won six races in 2026. Through the whole week before Washington his contract sat in the mid-90s on Kalshi, the CFTC-regulated exchange that lists a market on every full-season driver, and it traded as high as 99 cents on race day.
He finished 20th.
Kyle Kirkwood won the race and led the most laps in it. Christian Lundgaard finished second. When the points were counted, Palou banked 11 for the afternoon and Kirkwood banked 53, and a lead that stood at 133 points on Sunday morning stood at 91 by dinner. The market repriced him hard, from a 94-cent close on race day to a bid of 88 and an ask of 89 by Monday mid-morning, on the heaviest day of trading his contract has seen since late July. New to how those contracts work? Our four-step guide to Kalshi's sports markets explains why the prices are quoted in cents rather than as moneylines.
So we did what we do with every board like this: handed eight AI models the full season ledger with the market prices stripped out, asked each of them to price all 25 drivers in one pass, then let them read each other's reasoning and revise. The answer they reached is not the one the headline invites, and the reason is arithmetic that almost nobody has run.
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.
The Arithmetic That Just Closed The Door On 20 Drivers
IndyCar pays 50 points for a win, sliding down to 5 points for anyone classified 25th or worse. On top of that there is one point for pole at every race except the Indianapolis 500, one for leading a lap, and two more for leading the most laps. Add it up and the most any driver can score in a single one of the three remaining rounds is 54. The Indianapolis 500 runs its own qualifying-points scale on top of that, paying down from 12 points to the fastest of its Fast 12, but that race was run in May.
Three rounds remain. That is 162 points, and it is the entire budget every chaser has left.
Run it against the standings and the field collapses to five names:
- Álex Palou, 553 points, the leader.
- Kyle Kirkwood, 91 back.
- Christian Lundgaard, 110 back.
- David Malukas, 142 back.
- Pato O'Ward, 154 back.
Josef Newgarden is 177 back, which is more than 162, so he cannot reach Palou by scoring points no matter what happens at Milwaukee or Laguna Seca. Neither can any of the 19 drivers behind him.
The real line sits tighter still. A driver who takes the start and is classified last still banks 5 points, so Palou's floor across three starts is 15, and the largest swing a chaser can produce on track is 147 rather than 162. That puts O'Ward on the wrong side of it. To win the championship he does not need Palou to run badly. He needs Palou to miss races, and not merely one of them: sit Palou out of a single round and leave him last in the other two and his floor is still 10 points, a swing of 152 against a gap of 154. Two points short.
The points system does carry one clause that subtracts from a leader instead of requiring a chaser to out-score him. Listed in the driver-standings rules alongside the pole and lead-lap awards, it reads that each entrant-initiated engine change-out performed before the engine has reached its distance mandate "results in a loss of 10 championship points." That is the lever that reopens O'Ward's door without Palou missing anything, because it cuts his gap to 144 and back inside the 147 he can actually swing on track. It does nothing at all for Newgarden, whose 177 stays beyond the 162 still on the table even after the deduction. Three of the eight seats did leave a rounding-level fraction of a percent on him; none of them described a way for him to score his way there, and the panel's median on him is zero.
Which leaves the number that actually decides the season. Kirkwood's ceiling is 624 points. Palou is on 553. Scoring 72 more points across three races puts him beyond every chaser's maximum, and 72 across three races is an average finish of eighth. There is a second, earlier exit: if Palou leaves the Milwaukee doubleheader still more than 54 points clear of Kirkwood, the title is settled before the cars unload at Laguna Seca. He can afford to give back 36 of his 91 over two days and still be uncatchable in the finale.
Two Ovals And A Road Course Are All That's Left
The remaining schedule is the case for the chasers, and it is a better case than the points table suggests. Milwaukee Mile hosts a doubleheader on August 29 and August 30, two races on the same flat one-mile oval on consecutive days. The season then ends at WeatherTech Raceway Laguna Seca, a road course, on September 6.
Ovals are the one surface where Palou has been ordinary. His four oval starts this year read 24th at Phoenix, 7th in the Indianapolis 500, 17th at Gateway and a win at Nashville. Strip out Nashville and the other three average 16th. On road courses he averages 3.4 and on street circuits 4.8. Two of his three worst results of the season came on ovals, and two of the three races left are ovals, run back to back, where damage on Saturday is still damage on Sunday. That is the shape of a non-finish, or a DNF in racing shorthand, and it is the thing that most plausibly threatens a lead this size.
That is the whole bear case, and every model in the panel found it. It is also why, once they had all read each other, not one of them put him above 95.
Kyle Kirkwood, Andretti Global: The Only Chaser With A Racing Path
Kirkwood is the one driver whose surface profile beats Palou's on the surface that hosts two of the last three rounds. He averages 8.5 on ovals with a second place at Phoenix, against Palou's 12.25. His two wins this season both came on street circuits, at Arlington in March and in Washington on Sunday, and his pole came on the Nashville oval.
Traders noticed. By Kalshi's own 24-hour volume figure, his contract was the busiest line on the board after Washington, turning over roughly 8,500 units against about 3,900 on Palou's.
What the models kept circling back to is that his path is still a compound one. Even a Palou non-finish at Milwaukee plus a Kirkwood win moves the gap by about 49 points, which leaves Kirkwood roughly 42 behind going to Palou's best surface. He needs two Palou disasters, or one disaster and a near-sweep of his own, and his last five results include a 21st at Markham. The panel put that whole package at a little over 6%.
Dark Horses The Panel Won't Dismiss
David Malukas is 142 points back in a Team Penske car and has the best oval record of anyone on this board: finishes of 3rd, 2nd, 7th and 3rd, an average of 3.75, plus pole and most laps led at Phoenix. Four of the eight models named him their dark horse. GLM was blunt about why he still prices as a fraction of a percent:
"A naive reading of his 142-point gap might price him near zero, but his oval average of 3.75 is the best among all chasers and two of three remaining races are ovals at Milwaukee." — GLM
Christian Lundgaard is third in points, only 19 behind Kirkwood, and the panel still prices him at a little over a third of Kirkwood's chance. That is a far wider split than 19 points suggests, and the reason is the order of the calendar rather than the order of the standings. His road-course average of 3.0 is the best figure on the board, and Laguna Seca is a road course. His oval average of 14.0 is the worst among the five drivers still alive, and both ovals come first. He reaches the surface that suits him best with the least left to play for.
Josef Newgarden is the cruel one. His oval finishes this season are a win at Phoenix, a win at Gateway, second at Nashville and a 28th in the Indianapolis 500. He has won two of this season's four oval races, and the arithmetic closed on him a week before an oval doubleheader. By the same 24-hour figure his contract still turned over about 5,100 units.
The Board: 2026 IndyCar Championship Odds
| Driver | Team | Kalshi price | AI panel (median) | Gap |
|---|---|---|---|---|
| Álex Palou | Chip Ganassi Racing | 88.5¢ | 90.1% | Panel +1.6 |
| Kyle Kirkwood | Andretti Global | 5.0¢ | 6.3% | Panel +1.3 |
| Christian Lundgaard | Arrow McLaren | 1.5¢ | 2.4% | Panel +0.9 |
| David Malukas | Team Penske | 1.0¢ | 0.8% | Market +0.2 |
| Pato O'Ward | Arrow McLaren | 0.5¢ | 0.5% | Level |
| Josef Newgarden | Team Penske | 1.5¢ | 0.0% | Market +1.5 |
These are model estimates, not predictions of fact and not financial advice.
The other 19 drivers on the board are quoted between 0.5 and 1.0 cents each, and the panel prices every one of them at zero. Prices are the midpoint of the displayed bid and ask, fetched August 24, 2026 at 13:19 UTC. The panel figure is the blend, and that blend is the seat median: the middle of the eight seats' numbers rather than their average, so a single outlying model cannot drag it.
Two things about that table are worth more than the top line. The first is that only two of the 25 contracts have both a buyer's bid and a seller's ask behind them: Palou's and Kirkwood's. The other 23 show an ask with no bid behind it, which is a listed price rather than a real quote. The second follows from it. Add up all 25 midpoints and the board sums to 109 cents rather than 100, because each of those one-sided listings still carries a price. Every model's board sums to exactly 100 by construction. Of the market's 109 cents, 12.5 sit on the 20 drivers who cannot reach Palou even on the full 162 points still available, and the panel's median is zero on every one of them.
Every Seat's Number
Each model priced all 25 drivers in a single pass without ever seeing the market. Round 1 is the first independent answer. Round 2 is after each model read the other seven's reasoning, anonymized, and was free to revise or hold. The figures below are the probability each seat assigned to Palou.
| Model | Round 1 | Round 2 | Moved |
|---|---|---|---|
| Opus | 94.2% | 94.8% | +0.6 |
| DeepSeek | 78.0% | 91.5% | +13.5 |
| Gemini | 97.0% | 91.5% | −5.5 |
| Kimi | 85.0% | 90.2% | +5.2 |
| ChatGPT | 91.0% | 90.0% | −1.0 |
| GLM | 89.8% | 89.0% | −0.8 |
| Fable | 93.4% | 88.5% | −4.9 |
| Sonnet | 79.8% | 88.1% | +8.3 |
| Panel Median | 90.4% | 90.1% | −0.3 |
Every seat on this panel is graded against real market settlements — records to date: Gemini 86% on 4,746 graded calls · GLM 81% on 2,504 graded calls · Kimi 83% on 2,494 graded calls · DeepSeek 80% on 2,542 graded calls. Recomputed daily; the full scoreboard is public.
These are model estimates, not predictions of fact and not financial advice. Every number here gets graded in public once the market settles, on the full graded scoreboard.
More live boards from the same panel: F1 drivers championship, Antonelli 75.5¢ · F1 constructors championship, Mercedes 89.5¢ · today's Kalshi board. Prices fetched August 24, 2026 at 15:07 UTC.
Where The Panel Changed Its Mind
The first round produced a 19-point argument. DeepSeek opened at 78% on Palou, Gemini at 97%, and the rest scattered in between. After one revision round the spread had shrunk to under seven points, and the movement went both ways.
DeepSeek moved furthest, from 78% to 91.5%, and it moved because it went back and did the clinching arithmetic properly:
"Palou leads by 91 and needs just 72 points over three rounds to clinch regardless of others — an average of 8th place. His season average finish is 6.33, with 11 top-fives and zero observed DNFs, making a collapse to sub-72 points very unlikely." — DeepSeek
[Editor's note: the season ledger the panel was given recorded finishing positions, not retirements, so "observed" is doing real work in that sentence. Palou's 24th at Phoenix was his worst result of the year.]
Gemini came the other way, down from 97%, and the reason it gave was that its own worst case for Palou still was not bad enough:
"A realistic worst-case scenario for Palou is a DNF in one of the Milwaukee races due to the correlated risk of a doubleheader (where damage on Saturday compromises Sunday), netting 5 points, and a midfield finish in the other (~15 points). Even in this scenario, Palou heads to his strongest track type (Laguna Seca road course, 3.4 average) where he can easily defend a diminished lead." — Gemini
Opus barely moved at all, from 94.2% to 94.8%, and its reason for holding is the cleanest statement of the whole board:
"A 91-point lead over 147 points of real swing means every chaser needs Palou to fail twice, not to be outdriven once." — Opus
One exchange in that round is worth reporting on its own. A seat asserted in round one that Palou had finished 20th at Nashville. Three others checked it against the season ledger in round two, found that he won that race, and said so before pricing. The record beat the recollection, which is the entire point of showing the models the ledger.
What Would Change The Panel's Mind
These are the events the models named, each with the direction it pushes the number.
- An Entrant-Initiated Engine Change-Out On The No. 10 Before Milwaukee. A straight 10-point deduction with no chaser effort required. It pulls O'Ward back inside the scoring line and lifts Kirkwood and Lundgaard by a point or two each.
- Palou Failing To Finish Saturday's Milwaukee Race With Damage That Compromises Sunday. This is the correlated-risk scenario every seat priced separately from a single bad result. Opus put Kirkwood at 15% to 20% if it happens; Kimi put Palou down toward 70%.
- Kirkwood Or Malukas Sweeping Both Milwaukee Rounds. A sweep with bonuses is 108 points, more than the 91 that separates Kirkwood from Palou, but it only puts Kirkwood in front if Palou scores 16 or fewer across the same two days. Malukas is the likelier of the two to produce one on oval form, and the less likely to survive Laguna Seca afterwards.
- Rain Or A Cancelled Qualifying Session At Either Milwaukee Round. No qualifying means no pole point, which removes a small edge Palou has banked seven times this year, and wet or hot-track running raises pack-incident risk for everyone.
Settlement Timeline
| What | When |
|---|---|
| Round 16, Milwaukee Mile, Oval | August 29, 2026 |
| Round 17, Milwaukee Mile, Oval | August 30, 2026 |
| Round 18, Laguna Seca, Road Course, Season Finale | September 6, 2026 |
| Listed Market Close And Expiration, If Not Settled Sooner | September 13, 2026 |
| Latest Possible Expiration | September 27, 2026 |
The gap between the September 6 finale and the September 13 close is not an error. A champion is normally declared on track, and the exchange's own listing says the contract will close and expire once a winner is declared. The later dates are the contract's protection against a finale that does not run on schedule. Kalshi's racing contract terms say that if the deciding event is postponed, the market "will remain open and will resolve after the sooner of (1) the results ... being reported or (2) two weeks following the originally scheduled date of the final event." Two weeks past September 6 is September 20; the September 27 date is the outside expiration the exchange separately lists on this board, and it sits a week beyond that clause rather than being derived from it. Settlement runs on the official result reported by the governing series and the exchange's other listed source agencies, which the same document names as the Associated Press, ESPN, The Wall Street Journal and Fox Sports. Revisions made to the result after expiration, those terms state, "will not be accounted for."
This page is re-scored when the story moves. Prices and panel numbers are as of August 24, 2026, 13:19 UTC.
One structural note about these prices. Twenty-three of the 25 contracts have no bid behind them, so the round trip on a longshot here costs a large share of the position before anything is decided on track. Kalshi is regulated by the CFTC as an exchange for event contracts, and trading is 18+. The arithmetic above changes with whatever bid and ask are actually displayed.
Hottest Prediction Markets Right Now
- 2028 Democratic presidential nominee · $191M traded
- 2027 Pro Football Champion · $71M traded
- 2028 U.S. Presidential Election winner? · $62M traded
- 2028 Republican presidential nominee · $60M traded
- Pro Baseball Champion · $59M traded
- Bitcoin price at the end of 2026 · $33M 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.
- 2030 World Cup Odds: Why The USA Is Priced Above Norway
- Liga MX Apertura Champion Odds: The Panel Vs. Kalshi
- Liga MX Clausura 2027 Odds: América Surges To 29¢ Favorite
- Champions League Winner Odds 2027: Kalshi's 29-Club Board
- Super Bowl 2027 Odds: Eight AIs Won't Give Up On New England
The Bottom Line
The market and the panel agree on the shape of this and disagree on the details. Both have Palou around nine chances in ten. The panel is a point or two above the market on him, a point above on Kirkwood, and it takes the 12.5 cents the board is paying for drivers who are already out down to nothing.
Washington left the favorite in place and changed what he has to do to finish the job. A week ago the requirement was vague. It is now a number: 72 points across three races, or a 55-point cushion when the cars leave Milwaukee. Two of those three races are on the surface where his season has been at its worst, run on back-to-back days, starting Saturday. That is a smaller door than a 91-point lead sounds like, and it is still a very small door.
Hero illustration: OddsShopper, in the house collage style. Álex Palou photo by the City of Detroit, public domain; cropped, toned, and composited.
