Wild Card, Decided: Three Sweeps, One Extra-Inning Heist, and the Noon Tax

Six games across two days decided the wild-card round: three sweeps, one series that needed all six of its games plus an extra inning, and a desk ledger that grades every leg on the terms its cards published — wins, losses, and untracked games logged separately, nothing retroactively claimed. The short version: the Wednesday card went 2-0 on its leans and 1-1 on the owner's deviation tickets, the Thursday card went 1-0 on the round's biggest number, and the one bet the desk never priced is the one that cost money. The longer version is why the desk writes these.

The scoreboard

Road teams went 1-4 across the two days, and the one road winner was the biggest price on the board. The three sweeps were decided by starting pitching, which is the least surprising sentence of the postseason. The series that went the distance was decided by whose relievers cracked second.

Chicago (AL) 7, Houston 3 — the lean, and the same Astros story twice

Hunter Brown recorded five outs and was charged with four runs, and the game was effectively over before Houston came to bat in a second inning (box). The Wednesday card leaned White Sox +127 on exactly this shape: the Astros' market number was built for a settled staff, and the Astros' actual staff — four relievers down from Tuesday, starter chased with two outs in the first — was anything but. Chicago scored four before the inning ended, then let Houston's solo home runs (Jeremy Peña, Cam Smith, and Jose Altuve, one apiece off Sean Burke) die where they landed. Chase Meidroth's three-run homer in the sixth off Bryan Abreu turned a save situation into a box-check.

What the card got right, in its own words: the rebuild said the wrong team was favored, priced Houston 53.6 percent against Chicago's 46.4, and the game played to the rebuild's script — the Astros' season ended with the White Sox outscoring them 13-6 across two games, both lost by the team the market liked. Josh Hader, the counterweight the card named in public, threw 1.1 scoreless innings from the losing side. White Sox +127: win.

New York (AL) 9, Boston 2 — the pass that aged wrong, beside the ticket that cashed

Max Fried threw six innings of one-run ball, Luis García Jr. homered in the fourth, and the sixth inning turned a pitcher's duel into a rout: six runs, four charged to Garrett Crochet, the last three on Cody Bellinger's homer off Garrett Whitlock (box). Boston's only answers were solo shots from Willson Contreras — upheld on review — and Nick Sogard. The Red Sox bullpen gave up eight runs in Game 1 and seven more in Game 2; the relievers were the series, and the series was not close.

The grade the page owes: the desk's rebuild passed this game at +0.6 percent, a rounding error with a commission attached, and said so in public. The result disagrees, and the ledger says so plainly — the pass grades wrong in hindsight. The model priced Sonny Gray against Fried and read the two bullpens as a wash (3.15 against 3.27 on the season); the game was decided by the bullpens, and the season numbers lied on the day. Next to that pass, in the same section of the same card, sat the owner's Yankees −139 — a bet the model could not grade and the house played anyway. It cashes. The ledger shows a win where the card shows a pass because the desk logs deviations in the open rather than absorbing them silently, and because a coin-flip pass and a 9-2 result can both be true at once. The scorecard keeps both.

San Diego 4, Chicago (NL) 1 — the elbow held

Nick Pivetta made the Wednesday card the old-fashioned way: the lean hung on 34.2 regular-season innings and an elbow that cost him five months, and the rebuild's mechanical number pinned the starter term at its cap on that sample. Wednesday he allowed one run in 3.1 innings, Kevin Gausman — the shoulder the card flagged, the 4.51 ERA the card printed — gave back three in his own 3.1, and a pinch-hitting Gavin Sheets hit a two-run homer off Daniel Palencia in the fourth that held up as the deciding swing (box). The Padres' bullpen then did what it has done all series: 5.2 scoreless behind Pivetta, Mason Miller's second save of the round, and a Chicago offense that scored 850 runs in the regular season left fifteen men on base for the second consecutive game (Game 1 box).

Both versions of the model beat the market: the mechanical 69.1 percent and the regressed 59.1 both cleared a 56.1 percent price, which is the argument for publishing sensitivity ranges — the truth was inside the interval, and so was the money. The desk played it small, as published, and the small play was the right size for the wrong reason: Pivetta was fine, but the game was won off Gausman's shoulder, not off Pivetta's elbow. Padres −141: win.

Philadelphia 4, Atlanta 3 (10 innings) — the noon tax, itemized

The desk has a term for what happened in this game, and it is now in the glossary: the noon tax is what the desk pays for betting games it never priced. The Wednesday card could not track this one — a noon MT start ahead of the pricing tick, the second straight day of it — and the owner played Braves −106 anyway. The anatomy of the bill:

Tyler Mahle threw seven innings of one-run ball and left with a 3-1 lead the Braves built in the seventh on Drake Baldwin's RBI double and Matt Olson's sacrifice fly (box). Ronald Acuña Jr. had manufactured the tying run himself an inning into the game — a walk, two stolen bases, and a dead sprint home on Olson's single. Then Dylan Lee faced four batters in the eighth and gave up back-to-back solo homers to Kyle Schwarber and Bryce Harper, and a two-run lead became a tie. Alec Bohm led the tenth with a homer off Didier Fuentes, and Atlanta's evening of stranding runners — twenty-two left on base, which is not a typo — ran out of outs one inning short.

What the tax bought: not a model failure — there was no model number to fail. That is the point of the term. The noon tax is the price of betting a game the desk's own process never saw, and Wednesday's bill was house money plus a paragraph the desk would rather print than have you find. The mercy is the next entry on this page: the same team, a better price, a game the desk could actually see.

Atlanta 6, Philadelphia 2 — the flexed game grades the card's biggest number

The rebuild's largest published number of the postseason — Atlanta 62.8 percent mechanical, 56.2 committee-adjusted, against a 48.5 percent market — had its thesis in ink by the middle of the first inning: Aaron Nola recorded five outs and was charged with three runs, Michael Harris II hit the three-run homer, and the Thursday card never looked back (box).

The asterisk scenario played out on schedule. Zack Wheeler — the rested ace the card said the recipe could not see, the single biggest piece of information the mechanical number ignored — entered in the third on short rest and threw four innings of two-run ball. He dealt. The Braves already had five. Ozzie Albies took him deep for two in the fourth, Matt Olson added a solo homer off Jonathan Bowlan in the seventh, and the Atlanta bullpen made it stand: Ray Kerr opened with 3.1 scoreless, Grant Holmes got the win, and Chris Sale struck out Harper to end the eighth and fanned two more in the ninth. The market leaned Philadelphia — reverse line movement the card logged and declined to obey — and the market paid for it. Braves −102, small, as published: win.

The card, graded

Wednesday's card, at its published prices:

• Lean White Sox +127: win. Pays $127.00 per $100.
• Lean Padres −141: win. Pays $70.92 per $100.
• Pass on Yankees −139: wrong in hindsight. Graded on the published terms — no edge at +0.6 percent — and the result disagrees. Not retroactively a lean.
• Owner ticket Yankees −139: win. Pays $71.94 per $100. The deviation the card disclosed that morning cashed that night.
• Owner ticket Braves −106: loss. An untracked game bet anyway; itemized above as the noon tax.

Thursday's card:

• Lean Braves −102: win. Pays $98.04 per $100, on the round's smallest ticket.

For the round, at the prices each card published: card leans 5-0 (2-0 Tuesday per night one, 2-0 Wednesday, 1-0 Thursday), passes 1-1, owner deviation tickets 2-1. The Wednesday card's optional +288 parlay cashed on paper and was not a logged ticket; the ledger does not claim it retroactively, because that is the whole rule. The caveats travel with the number: five is a sample of five, the model was built on 162-game seasons, two of the five legs were graded at prices pulled from published odds articles rather than the app, and the desk's standing position is that a 5-0 week is a headline, not an identity.

What the round actually teaches

Every series the desk priced correctly turned on the same hinge: a starting-pitching number the market could not update fast enough. Houston lost both games its improvised staff started. San Diego's whole case was an elbow the market could not price and the desk could barely price itself. The one game the desk refused to price is the one that cost it money. The division series opens Saturday — Chicago (AL) at Cleveland, New York (AL) at Tampa Bay, Atlanta at Los Angeles, San Diego at Milwaukee — and the board is already up. The model reads season lines; October is being decided by bullpen availability on zero days' rest. The desk will keep saying which numbers it can see and which it cannot.

Results verified against final box scores; every leg graded at the price published on its desk card. Series status as of the afternoon of October 2.