ALCS Game 2 Desk Card: The Machine's Rays at 64.0%, the Desk Recipe's Guardians at 63.7%, and One Dinger You Can Only Buy at the Right Window

Cleveland was 1-29 all time in road playoff games when trailing by multiple runs from the sixth inning on. Then Chase DeLauter went deep, Patrick Bailey went deep, and Nathaniel Lowe flicked a 70-mph single off an 0-2 pitch into the eighth-inning night, and the Guardians left Tropicana Field up 1-0 in this series (ESPN). Seven runs on 14 hits, six of them over the final three innings, 48 hours after winning Game 5 in Chicago. Tonight is the only game on the board. The desk card has one machine leg and one dinger, and a longer argument about why the house is not carding its own side.

The game

The flags point opposite ways. Messick is a fastball-changeup workhorse — 30 of 32 regular-season starts of five-plus innings (Covers) — whose hard-hit and barrel rates both climbed through September while his innings count approached 200 after two straight sub-140 seasons (VSiN). His postseason xERA is 6.10 with six walks in 6.2 innings (Covers). Peralta is a reclamation with a fixed changeup — usage up from 20.8% as a Met to 29.7% as a Ray — and a fixed left-handed split, but his shutdown finish ran on a .187 BABIP and a 94.0% strand rate against a 4.44 xFIP, which is the market's way of saying some of it was luck (Covers). Both bullpens are rested enough. Cleveland's is better: 2.93 on the season with Cade Smith at 1.87, against Tampa Bay's 3.31 (Covers).

The market, split three ways

VSiN's number has Tampa Bay at -132 with a total of 7 (VSiN). Fox carries -137, with Cleveland at +113 (Fox). Covers' best-odds board ran Cleveland at +121 this morning, and 71% of the picks on that site ride the Rays (Covers). The public is on the home favorite and one sharp write-up is on the road underdog at a near-pick'em win probability — the market disagrees with itself, and we'd rather tell you than not.

Two reads, opposite walls

Every instrument the house owns produced a number for this game. They do not agree on the address.

The recipe, applied honestly with the season lines: the starter term is 8×(2.56 − 4.42) + 10×(1.06 − 1.33) + 0.8×(9.33 − 8.31) = −16.8 before its ±12 cap engages, which is the single biggest starter penalty the house table has printed all October. Bullpen is 4×(2.93 − 3.31) for Cleveland's pen, record is 1.5×(.525 − .605) against the 98-win top seed, and the Trop reads −3 on the park term — a 94 runs index on the three-year view, 98 this season, with a 104 home-run index that says the dome suppresses runs and not dingers (Statcast). Add it up and the mechanical answer is Cleveland by a landslide.

Where the machinery is wrong: Peralta's 4.42 is two-thirds Met. As a Ray he has run a 3.27, and his final six regular-season starts allowed three runs total (VSiN). Messick's 2.56, meanwhile, is a fatiguing workhorse's best work — September barrel creep, a 6.10 postseason xERA, six walks in 6.2 October innings (Covers). Re-run the starter term with the form lines — 3.92 against 3.27 — and the cap never engages: the recipe prints Tampa Bay at roughly 52%, which is the market's neighborhood with worse posture. The desk's honest adjusted read is that the raw number is a filing error, the form number is the market, and the loop is the loop. The house rides the machine's ticket and adds nothing of its own. When your two instruments point at different walls, the discipline is to not build a third one.

The card

The loop carded Tampa Bay this morning: 64.0% against a 54.2% no-vig market, +9.8 points of edge, sized 0.5u at 1.64 off its recorded price. The ledger behind that number stands at +13.59u on 97 positive-edge bets, a 0.2371 mean Brier across 206 graded predictions — the house's book of record, written down before first pitch and graded against the price on the page. The market opened this morning well short of Game 1's close of -163 — VSiN -132, Fox -137 (Fox) — which is the market grading Monday before the house does. Every number on that board pays better than the loop's recorded 1.64, so the fill should beat the carded price; the leg grades at the fill. One watch item prices itself: Junior Caminero collided with Jonny DeLuca in Game 1, grabbed at his pectoral, and was down long enough that the manager had to explain it (VSiN). If Caminero does not start, the loop's 64.0% is stale and the desk says so here, in advance.

The dinger rides against the lean. A Cleveland home run hurts a Tampa Bay ticket, and the house cards both anyway, at a combined 0.75u that says we know.

The dinger board

The recipe, applied to every regular in both lineups: season HR/PA, times 4.3 plate appearances, times the starter's home-run adjustment — Messick at 0.82 HR/9 against a 1.17 league rate for a ×0.71 drag, Peralta at 1.23 for a ×1.05 boost. League rate is the 2026 MLB average, 5,575 home runs over 43,079.2 innings, per the MLB Stats API.

Cleveland, against Peralta:

Tampa Bay, against Messick:

Caminero is the best bat and it is not close — but +329 prices him at a 23.3% breakeven against an 18.1% fair, the shortest ticket on the board is a tax, and the shoulder makes it a tax on a sore asset. The house passes, same as it passed Muncy at the wrong book in Game 1 of the other series. Bailey and Adell top the Cleveland raw board among expected starters at 15.9% and 15.5% — no verified price, no leg; the scrape-gap rule from the division series stands.

Why Ramírez at a 10.0% raw fair: the same discretionary form credit the house printed on William Contreras two nights ago, named again so the reader can discount it. Ramírez's 12-homer season is a hamate surgery on the mend — his own manager said the further from the surgery he gets, the better he will feel, and the postseason evidence agrees: 8-for-23, a .348 average and .739 slugging this October, two Game 1 singles clocked at 106.5-plus mph, and a .456 xwOBA against Peralta's stuff across a small sample (SI) (Covers). The desk weights that to roughly 14%, which clears the 12.5% breakeven at +700 and does not clear DraftKings' +543. Bet it where it clears or not at all. The house's carded dingers are 0-3 this October, and a form-credit quarter unit is how that ledger gets rebuilt — one priced edge at a time, never retroactively.

Counterweights

• The Rays' regular season against left-handed pitching: 27th in wOBA, 30th in ISO — Tampa Bay spent six months not hitting pitchers like Messick (Covers). The loop's 64.0% is an Elo number built on results, not a scouting report; if you think Monday rewired the lineup, that is your call to make, not ours.
• Messick against this specific team: a .196 xwOBA and .049 ISO allowed to current Rays, on a small sample, and Tampa Bay's contact profile — aggressive, low-walk, low-strikeout — is the profile that makes a fastball-changeup rookie pitch from behind in the count or get hit hard doing the opposite (Covers).
• The Cleveland bullpen is carrying mileage: 32 innings across six postseason games, and Vogt leans on Gaddis, Sabrowski, and Smith because the alternative is worse (Covers).
• The total is a coin flip wearing a price tag: VSiN likes Under 7 at -122 on the soft-contact starters; Covers' staff plays Over 6.5 at -118 on regression and RISP volume. The desk still owes the house a priced totals recipe after the division-series scrape gap, so none is carded tonight (VSiN) (Covers).

Grading

Every leg grades tonight against the prices on this page: the machine leg at 1.64 or the actual fill, whichever is worse; the dinger at the placed price, with the +650 floor a condition of the bet, not a guideline. If Caminero is scratched, the machine leg's grade carries an asterisk and the log will say so. Noon tax rules do not apply — this one starts at 6:00 p.m. MT under a dome, which is the only October venue where the weather cannot hurt you.

Bring your own pencil.