AL Central Odds 2026: White Sox Surge Past 60¢ On Kalshi

The Quick Answer

Eight AI models scored Kalshi's AL Central odds 2026 board price-blind on July 27 and landed almost on top of the market's favorite: Chicago White Sox at a 45% blend against a 44-cent price, with Cleveland at 28% and Minnesota at 18%. The consensus number hides the actual story, though: the eight models disagree with each other by 17 points on Chicago, and all eight flipped the market's order on Detroit and Minnesota — a call the field has been stress-testing ever since, with the Tigers repriced from 14 cents to 31 by mid-August before easing back toward 18 as the White Sox pulled away past 60 cents. The full board, every model's number, the worked cents-to-probability math, and the August 18 standings-and-price update are below.

The Race Nobody Is Pricing Loudly

Every division market has a personality. The NL Central is a near-settled question, one runaway leader and a crowd deciding how much to pay for near-certainty. The AL Central is the opposite: four teams within 6 games of each other at the July 27 capture, a first-place team trading at just 44 cents, and a market that has quietly priced a real four-way argument while the baseball world writes about other races. That contrast is why I keep coming back to this board. When a race is actually open, the interesting question stops being "who leads" and becomes "how much is a small lead worth with two months left." That is the question our eight-model panel answered without seeing a single price, and one pair of teams on this board produced the panel's only unanimous disagreement with the market's ORDER. We will get to that team, but the argument starts at the top.

The Market At A Glance

The five displayed prices added up to roughly 106 cents at capture, not 100, because each contract is quoted separately with its own spread and stale prints. Hold that number; it matters when we convert cents to probabilities later.

The Board: Market Price Vs. AI Blend

Prices and model numbers below are from the July 27, 2026 capture, the night the panel ran and this article first published. The standings card every model saw: White Sox first, Cleveland 2.5 games back, Minnesota 3.5, Detroit 6.0, Kansas City 11.5, with roughly 55 to 58 games left depending on the team. Prices move daily; the August 18 board and standings update is further down.

The blend column sums to 97.5%, the price column to 106 cents, and the single most interesting row is not Chicago. It is the pair at the bottom of the contending tier: the market paid one more cent for Detroit than for Minnesota, while the panel had Minnesota three times as likely as Detroit. Chicago is the agreement; Detroit against Minnesota is the fight. Before we pick that fight, look at what the averages are smoothing over.

Every Model's Number

The blend is an average of eight independent reads, and the spread inside it is wider than the headline suggests. Reader-facing model names, every number as stored the night of the run:

Run your eye across the Chicago row: 38% at the low end (Claude Opus), 55% at the high end (DeepSeek V4). That 17-point spread on the favorite is the widest on the board, and it comes from the same standings card. All eight models worked from one shared snapshot of the race; they priced what a small lead is worth very differently.

The Disagreement Is The Story

A price-blind panel earns its keep when the models argue, because the argument shows you which assumption each number depends on. On Chicago, the argument is about what a small lead means in a weak division.

DeepSeek V4, the high read at 55%, was willing to pay the most for the lead: "White Sox lead by 2.5 games with ~58 left, a small but meaningful edge." Claude Opus, the low read at 38%, looked at the same line and discounted it: "White Sox lead by 2.5 with ~58 games left, but weaker underlying talent and three close chasers temper the edge." Claude Fable split the difference at 45%, calling Chicago a "modest favorite" in "a weak AL Central." In plain terms: DeepSeek priced the standings, Opus priced the roster, and the market's 44 cents sits in the low-middle of the panel's 38-to-55 range. One honesty note on that range: Claude Sonnet's 40% came with a rationale that misread the card as a "half a game" lead, so part of the spread is one model pricing a different race, which is itself a finding about price-blind panels. When a market lands inside the models' own range of disagreement, that is usually a sign the price is efficient rather than lazy.

Cleveland is the mirror image. Seven of the eight models bunched between 28% and 31% on the Guardians, the tightest cluster on any contested row of the board, and all of them landed below the market's 34 cents. The panel's read was consistent: Cleveland is the most credible chaser, but a 2.5-game deficit in a race with a third live team splits the comeback odds. When eight blind reads agree that closely, the number is probably about arithmetic, not opinion, and the arithmetic said the market was paying a small premium for Cleveland's reputation.

Which brings us to the promised quarrel.

The Detroit-Minnesota Flip

At capture, Kalshi had Detroit at 14 cents and Minnesota at 13. Every single model had that order backwards, and not by a little: the blend put Minnesota at 18% and Detroit at 6%. Not one of the eight models rated the Tigers above the Twins.

The one unanimous order call on the board: all eight models rated Minnesota above Detroit while the market paid a cent more for the Tigers. The field answered back: Detroit passed Minnesota in the standings, and the market repriced the Tigers from 14 cents to 31 by August 12 (Kalshi's daily candle for that date shows Detroit trading between 22 and 35) before easing to 17-18 by August 18, still roughly double the Twins' 8-9. The panel's most confident call is now the one the market disputes hardest.

The models' reasoning was standings math, and not always tidy math: the same card produced games-left counts anywhere from 53 to 60 across the rationales. GPT-5.5 on Minnesota: "Minnesota is only 3.5 games back with roughly 55 games left, but trails two teams and needs meaningful outperformance." Gemini 3.1 went furthest, calling the Twins "a very viable contender in a tight divisional race" at 22%. On Detroit the same models saw a harder problem, six games back with three teams to leapfrog; DeepSeek V4 put it at 1.5% and called a division title "very unlikely." A team 3.5 back with two teams to pass simply has a shorter path than a team 6 back with three to pass, so the panel treated the market's extra cent on Detroit as a mispricing. Here is what the card could not show them and the market apparently knew: run differential. At capture Detroit was plus-28, second-best in the division behind only Chicago, while Minnesota was minus-27 and Cleveland minus-18. A standings-only card structurally blinds a panel to the one number that said the fourth-place team was the second-best club. (Our Tigers-specific verdict page, from a separate July run, was built on a fuller card and made the run differential its whole thesis.)

Here is the honest August 18 update, and it does more than complicate the panel's cleanest call — it inverts it, for now. Detroit sits second, 4.5 games back, while Minnesota sits third at 5.0 back; on the board, the Tigers sit at 17-18 cents against the Twins' 8-9. The order the models called has flipped on the field, and the market that once paid one extra cent for Detroit now pays about 9. Detroit's run differential has swelled to plus-86, the best in the division by 41 runs, while Minnesota has sunk to minus-46. It is still too early to grade this board; what the weeks since capture have already graded is the panel's certainty, and that is worth saying out loud. This is why every verdict we publish is graded in public on the Sims vs. Kalshi ledger: a panel that only reports its wins is a marketing asset, not a measurement.

And Kansas City? Both sides agree there is nothing to argue about. The market's 1 cent and the blend's 0.5% describe the same team. Kimi K3 had the Royals at 1%, citing the 11.5-game deficit and adding an assumption its card never showed, that Kansas City were deadline sellers, needing "a historic surge plus a White Sox collapse."

From Cents To Probability: A Worked Example

If you are newer to event contracts, here is the full math on one row of the board, using Cleveland at 34 cents.

A YES contract on Cleveland cost $0.34 at capture and pays exactly $1.00 if Cleveland wins the division, $0 if it does not. Buy 100 contracts and you have risked $34 to win $100, a profit of $66. The break-even, before fees and spread, is simple: you need Cleveland to win the division more than 34% of the time for that trade to profit over the long run; the exact break-even sits a little higher once execution costs are in. So a 34-cent price is a 34% implied break-even, not a de-vigged true probability.

Remember the five AL Central prices summed to about 106 cents. Since exactly one team must win the division, five true probabilities would sum to 100, so displayed prices across separately quoted contracts overstate the field a little; part of that is spread, part is stale prints. A rough re-scale divides each price by the total: Cleveland's 34 cents becomes 34/106, about 32%, before you compare it to the panel's 28%. The 6-point gap you see in the board table shrinks to about 4 under that re-scale. Treat it as an apples-to-apples comparison, not a true no-vig probability; a real cross-market cost needs simultaneous asks on all five plus fees. (At the August 18 refresh, the five bid-ask midpoints summed to about 101.5 cents.)

Where The Race Stands Now (August 18 Update)

The panel's numbers are frozen at the July 27 capture; the standings and the prices are not. Here is the live AL Central table as of the afternoon of August 18, 2026 (before that night's games), per MLB's official stats feed, alongside each contract's Kalshi bid-ask midpoint at 3:07 p.m. CT:

The weeks since capture redrew this race top to bottom. The White Sox went from a 2.5-game lead at capture to 4.5 games clear, and the market followed: 44 cents the night the panel ran, 46 on August 12 (inside that day's 44-to-51 trading range), 62-63 on August 18. That last number is the one to sit with. The panel's blend was 45%, and its single most bullish read on Chicago was DeepSeek's 55%; the market has now moved past even that. The market moved beyond the panel's entire Chicago range as the standings opened up, an early mark against the panel's July 27 ceiling. Everything below Chicago moved too. Detroit jumped both Minnesota and Cleveland in the standings and got repriced from 14 cents to 31 by mid-August, then gave some of it back to 17-18 as its own record sagged to 61-64 — still about 11 points above the panel's 6%. Look at the run differential column, though, and the market's side of that argument stops looking like a hot-streak bet: the Tigers outscore opponents by 41 more runs than the division leader. Cleveland, the row where seven models clustered most tightly, has slid to 5.5 back and watched its 34-cent contract trade down to 12-13; the small reputation premium the panel flagged is gone and then some. We do not restate model numbers to fit new standings; the July 27 verdicts stay frozen. What you can do with the frozen numbers is watch which side of each argument the market keeps paying for as the gaps move.

The DFS Lens On A Division Race

The panel's Minnesota-over-Detroit call was path math, 3.5 back passing two teams against 6 back passing three, and the market's August answer, Tigers at 31 cents and then 17-18, is a bet that underlying performance beats the path. Run differential is where the two views meet, because a plus-86 is nothing more than roughly 125 nightly game environments added up, and DFS players price those environments one slate at a time. That is the practical read for lineup builders: the standings say Detroit is a fourth-place-turned-second-place team, the run column says the Tigers have outscored everyone in the division since the day the panel ran (plus-58 in that span, while Chicago is minus-1), and a projection tool that prices matchups game by game sees the second thing. If you build MLB lineups, MLB DFS projections, ownership and stacks in the DataHub price all five of these offenses that way every day, a much faster feedback loop than waiting two months for a division contract to settle.

For the daily version of this exercise across Kalshi MLB markets, our Kalshi MLB picks board runs the same sims-versus-market comparison every slate. And if you are coming at event contracts from a pick'em background, we broke down what actually transfers from DFS to prediction markets.

More on this: NHL Eastern Conference Odds 2027: Panthers Overtake Carolina · NL Central Odds 2026: Brewers 88%, Cubs The Only Question · NBA Eastern Conference Odds 2027: LeBron's 76ers Lead At 23\u00a2 · NHL Western Conference Odds 2027: Avalanche Favored At 21\u00a2 · Fantasy RB1 Odds 2026: Gibbs Leads, Bijan Is The Value

Graded In Public

Numbers from a black box deserve zero trust, so every model's graded record against real settlements is public on the Sims vs. Kalshi ledger. Accuracy on mostly lopsided markets should be high, which is why the ledger grades every call rather than a highlight reel, and why a wide-open board like this one is worth more to the record than ten near-settled ones.

So, resolving the thesis this article opened with: the AL Central really is the race nobody is writing about, and the market has it priced more carefully than the silence suggests. On the favorite, the price sat inside the models' own argument at capture and has since run past the top of it, 62-63 cents against a 55% ceiling. On the chaser, eight blind reads found a small reputation premium, and the market has since taken that premium back and then some. And on the one row where the panel unanimously called the market wrong, the field flipped the order the models bet their unanimity on: Detroit passed Minnesota on the diamond and by about 9 cents on the board, and the run differential column says that was never just a hot month. A quiet board, it turns out, is not the same as a sleepy one.