BBMI Division Series Preview: One Clear Favorite, Three Near Toss-Ups
The Dodgers are 67% favorites over a Braves club that beat them five times in six tries this season. The other three series are close enough to turn on a start: Yankees 55% as road favorites at the top-seeded Rays, Brewers 55%, Guardians 54%.
Four more Game 1s today, starting at noon Central in Cleveland and ending under the lights in Milwaukee. The byes are over: the top two seeds in each league enter, and BBMI simulated each best-of-five 60,000 times, game by game on its real calendar, with the actual rosters, starters and ballparks. A bullpen rest rule is part of the method, detailed in the methodology; it moves nothing here by more than about a third of a point.
at
White Sox at Guardians
at
Braves at Dodgers
at
Yankees at Rays
at
Padres at Brewers
These are pre-series odds, built only on what was known before each Game 1, with player stats through the Wild Card round. Once a Game 1 is played, that series’ number is out of date, and the Cleveland opener’s already has an asterisk: it was computed just after the noon first pitch, on inputs frozen before it. Each series reaches a fourth game about three times in four and a fifth a little more than a third of the time. Sixteen of the forty possible starts are the model’s projections rather than announcements, so every series gets re-priced as rotations firm up.
Series by series, in order of first pitch.
White Sox at Guardians: the closest series of the round
Chicago won the season series with Cleveland, seven games to six, and seven of the thirteen meetings were one-run games. That is about as even as two division rivals get, and the model agrees: the Guardians are 54% favorites, the smallest edge on the board, with no game in the series further than 56-44 either way.
Hagen Smith (L)vs. Parker Messick (L)Sean Burke (R, reported)vs. Gavin Williams (R)Foster Griffin (L, proj.)vs. Davis Martin (R, proj.)Parker Messick (L, media)vs. Anthony Kay (L, proj.)Hagen Smith (L, proj.)vs. Gavin Williams (R, media)Cleveland’s plan, per MLB.com’s rotation analysis, is to lean on its two best arms twice each: Parker Messick, who rates better than 76% of qualified starters on BBMI’s pitching scale, in Games 1 and 4, and Gavin Williams, at the 60th percentile, in Games 2 and 5. (The scale weighs home runs, walks and strikeouts per batter faced by what each costs; lower is better, the median qualified starter sits at 0.382, and every percentile in this piece is against that 140-starter pool, with Wild Card games counting.) Those four starts are the Guardians’ entire case, because projected Game 3 starter Foster Griffin is the only other Cleveland arm in the series above the median, and Chicago’s three non-Smith starters all rate in the 30s.
GuardiansWhite SoxChicago is running it back with Hagen Smith, the opener who beat Houston, on three days’ rest, and if the series goes five he does it twice. Both of his starts are bullpen games: the simulation sends him about two innings, which makes Chicago’s 4.8 relief innings a night the heaviest bullpen load in the round. One honest caveat: Smith’s length is our assumption, and MLB.com’s Game 1 preview puts his limit at three or four efficient innings. A longer Smith outing would help Chicago; on an earlier version of these inputs it was worth one and a half to two points of series probability.
Game 4 is where the bats concentrate. Jo Adell against Anthony Kay is the sharpest single matchup of the entire round, a .447 BBMI matchup value against a Cleveland lineup average of .321 in that game, and José Ramírez is right behind him at .422 against the same pitcher. BBMI matchup value is the run value of everything a hitter produces, weighted the way wOBA weights it, per plate appearance against that specific starter; it reads like wOBA but is neither wOBA nor batting average, so compare the values only with each other. Miguel Vargas, at .348 to .365 across the series, is Chicago’s loudest bat.
Braves at Dodgers: the champs, against the one team that handled them
Atlanta won five of six regular-season meetings with Los Angeles. BBMI makes the two-time defending champion Dodgers 67% favorites anyway, favored in all five games, because the model prices the five games in front of these clubs, and those games run through the best rotation on the board.
Dylan Dodd (L, opener) + Grant Holmes (bulk, reported)vs. Tarik Skubal (L)Chris Sale (L, reported)vs. Blake Snell (L)Yoshinobu Yamamoto (R, media)vs. Tyler Mahle (R, proj.)Tyler Glasnow (R, media)vs. Grant Holmes (R, proj., short rest)Chris Sale (L, proj.)vs. Tarik Skubal (L, proj.)Tarik Skubal rates better than 99% of qualified starters, the best arm anywhere in this round, and every Dodger starter sits in the top quarter: Snell at the 95th percentile, Glasnow at the 80th, Yamamoto at the 77th. Against Skubal, Atlanta’s lineup projects at .240 BBMI matchup value, the lowest any lineup shows in any game of the round, with Matt Olson’s .267 the best of a quiet bunch.
Atlanta’s Game 1 answer is to not really start anyone. MLB’s listed starter is Dylan Dodd, opening for about an inning, with Grant Holmes carrying the bulk behind him, the same shape Philadelphia and Chicago have already tried this October. The structural cost comes later: with three true starters on the roster, the model brings Holmes back to start Game 4 on three days’ rest, the only projected short-rest start in the round, and he rates better than only 14% of qualified starters even rested, the lowest-rated start on the board. Worth knowing as you weigh that: Holmes and Sale both pitched in relief in Thursday’s Wild Card clincher, and the model prices every starter, short rest included, at his normal rates and workload.
Chris Sale at the 95th percentile is the one Brave who belongs in this rotation fight, and he gets Games 2 and 5. Those are among the games where Atlanta comes closest. Shohei Ohtani is the bat: .401 against Holmes in Game 4, the fourth-sharpest matchup of the round against a Dodger lineup average of .335 in that game, and .371 against the opener Dodd in Game 1. The model keeps the Dodgers’ lineup built for the left-handed Dodd on the field when the right-handed Holmes takes over, because it does not substitute mid-game.
Yankees at Rays: the road team is favored, again
In August, BBMI published a hypothetical version of this series and made the Yankees about 57% favorites despite Tampa Bay holding home field. Two months later it is real, the Rays won 98 games and the AL’s top seed, and the model still takes New York: 55%, the only series in the round where the lower seed is favored.
Gerrit Cole (R)vs. Drew Rasmussen (R)Cam Schlittler (R)vs. Freddy Peralta (R, media)Nick Martinez (R, media)vs. Max Fried (L, reported, order proj.)Drew Rasmussen (R, proj.)vs. Carlos Rodón (L, reported, order proj.)Gerrit Cole (R, proj.)vs. Freddy Peralta (R, proj.)The mechanism is the same one the August piece found: the back of Tampa Bay’s rotation. Drew Rasmussen, at the 79th percentile, is announced for Game 1 and is the Rays’ only above-median starter, so the model has him back on regular rest for Game 4 and the rotation turning to Freddy Peralta (38th percentile) and Nick Martinez (36th) in between. New York counters with Cam Schlittler at the 84th, Max Fried at the 74th and Carlos Rodón at the 64th; Boone has named Fried and Rodón for Games 3 and 4 without giving an order, so the order is projected. The result is a series where Tampa Bay is favored only in Rasmussen’s Game 1, the Yankees take Games 2 through 4, peaking at 61% behind Fried in the Bronx, and a potential Cole-Peralta Game 5 at the Trop is a dead coin flip.
The honest counterpoint belongs to Peralta: his rating sees the whole season, and he closed it with a 0.79 ERA over his last six starts. If October Peralta is that pitcher rather than the 38th-percentile one, the middle of this series looks different.
The season lines push New York’s way regardless: the Yankees outscored opponents by 0.86 runs a game to Tampa Bay’s 0.53, so the better run-margin team is the road team here. Ben Rice is the bat the Rays’ plan walks into, at .382 against Peralta and .371 against Martinez, against Yankee lineup averages of .298 and .307 in those games. Junior Caminero, at .315 and .321 against Schlittler and Rodón, leads Tampa Bay’s side. The rest rule registers slightly toward the Rays here, about a quarter of a raw point.
Padres at Brewers: 103 wins buys a 55% edge
Milwaukee had the best record and run margin in baseball, 103-59 and +1.32 a game. San Diego went 18-6 in September, and so did Milwaukee. BBMI makes the Brewers 55% favorites, which says less about Milwaukee than about where this series gets decided.
The Padres also lost a starter on the eve of it: Walker Buehler went on the paternity list, with Randy Vásquez added to the roster. Buehler can be away up to three days and could return mid-series; for now the projected Game 4 is Vásquez’s, and the series gets re-priced if that changes.
Robbie Ray (L)vs. Jacob Misiorowski (R)Michael King (R)vs. Logan Henderson (R)Robert Gasser (L, proj.)vs. Nick Pivetta (R, media)Dustin May (R, proj.)vs. Randy Vásquez (R, proj.)Robbie Ray (L, proj.)vs. Jacob Misiorowski (R, proj.)The front of Milwaukee’s rotation is the case for the favorite. Jacob Misiorowski, at the 94th percentile, gets Games 1 and 5, and Logan Henderson at the 80th gets Game 2, against a San Diego front of Robbie Ray (27th) and Michael King (21st). Then the series crosses the country and inverts. Milwaukee’s Games 3 and 4 fall to Robert Gasser and Dustin May, both just under the median, while San Diego’s best start of the series, Nick Pivetta at the 80th percentile, waits in Game 3, the one game the Padres are clearly favored to win, at 58%. Game 4 is nearly even, Vásquez at the 23rd percentile against May at the 47th.
BrewersPadresMilwaukee’s bats carry the other half of the argument. Jackson Chourio owns the second-sharpest matchup of the round, .428 against Robbie Ray, against a Milwaukee lineup average of .306 in that game, and gets it twice if the series goes five. Brice Turang’s .395 against Vásquez in Game 4, against a .323 average, is the fifth-sharpest. Fernando Tatis Jr. at .347 against Gasser and Jackson Merrill at .381 against May carry San Diego’s side. The rest rule registers here too, about a third of a raw point toward the Padres, because the NLDS calendar’s two back-to-backs sit roughly twice as many Milwaukee relievers as San Diego’s, though San Diego’s own pen is thin enough that the rule’s five-arm floor forces tired Padres arms back in about three in ten of its post-back-to-back games.
The season at a glance
Final 2026 regular season, MLB Stats API. The tension of the round sits in the AL: Cleveland, with the thinnest margin of the eight, is favored over a White Sox club that outscored opponents by more and won the season series, and the 98-win Rays are home underdogs to a Yankee team whose margin was a third of a run better.
How BBMI built these numbers
Every figure comes from BBMI’s production MLB simulator: each best-of-five played out game by game on its real dated calendar, 20,000 times on each of three random seeds, with the real parks, the Division Series rosters, and lineups built from each club’s recent starts against each starter’s hand. Each club runs a four-man rotation by default, adjusted where clubs have said otherwise; every rostered pitcher not starting in the series pitches out of its bullpen, and the deciding game adds every rotation arm except that night’s starter and the Game 4 starter. Player stats run through the Wild Card round.
Bullpen rest is in these odds. A reliever who records more than three outs sits the next calendar day, nobody pitches three days running, and every club keeps at least five arms. The rule passed all five of its pre-registered validation checks in all four series before it was allowed into a published number. Its effect is small everywhere, never more than about a third of a raw point of series probability, and where it clears its own measurement noise it leans toward the Guardians (+0.2), the Rays (+0.3) and the Padres (+0.4 against the Brewers).
Only pitchers relieve, and two are held out by assumption. Shohei Ohtani is kept out of the Dodgers’ bullpen, and Joe Musgrove, who threw no 2026 innings, out of San Diego’s. The Padres’ thin modeled pen is partly that omission.
The odds are calibrated. Each game’s raw probability is adjusted by a map fit on 2,421 real 2025 games, then the five games are combined, so series and game odds reconcile. The map moves in steps, so nearby games can share a value, and the three close series should be read as close: within the resolution of those steps, the Yankees’ 55% could fairly read 54 to 56, the Guardians’ 54% could read 52 to 55, and the Brewers’ 55% could read 54 to 56. The Dodgers’ 67% does not move. Raw simulator figures are in the audit table below. Simulation noise on a series is 0.3 points or less; real uncertainty, from rosters, health and the model itself, is larger.
Guardians
Dodgers
Yankees
Brewers
Starter labels matter this round. Thirteen of the forty possible starts are announced, four are reported by a club or manager, seven come from press reports of clubs’ likely plans, and sixteen are the production rotation selector’s projections. Every non-announced start is labeled in the tables, and where a manager named two starters without an order, the order is marked projected. Each series is re-priced as rotations are announced. Starter reporting: Sean Burke and Cleveland’s Games 4 and 5 plan per MLB.com’s series preview; the Dodd-Holmes opener plan and Chris Sale’s Game 2 per Bolavip; Fried and Rodón per NBC Sports; Yamamoto and Glasnow per NBC Sports; Peralta and Martinez per MLB.com; Pivetta per MLB.com. Dodd’s one-inning opener length is our assumption, his 2026 average per relief outing; no length was announced.
The Dodgers’ roster is provisional. It had not been filed with the MLB Stats API when these runs were made, so Los Angeles is priced on its regular-season roster plus injured-list returns, which gives its modeled bullpen more arms than the seven clubs with filed rosters. That likely flatters the Dodgers, and the series gets re-run when the roster files.
The Cleveland opener is priced retrospectively. Its Game 1 began at noon, before these runs finished; the figure uses only inputs frozen before first pitch, the same pre-series convention as the Wild Card brief.
Six starter ratings are soft, resting mostly on a league-average anchor or on 2025: Hagen Smith, Blake Snell, Dylan Dodd, Tyler Glasnow, Carlos Rodón and Nick Pivetta. Their percentiles above carry more uncertainty than the rest.
Hagen SmithBlake SnellDylan DoddTyler GlasnowCarlos RodónNick Pivetta
no in-game substitutions or pinch-hitters, so rostered bench bats like Giancarlo Stanton are not in the modeled lineups; switch hitters are dropped from pitchers’ platoon splits, worth roughly 0.8 points in our August Yankees-Rays study; defense isn’t priced; short-rest and heavy-recent-workload effects aren’t modeled, so Grant Holmes’s Game 1 bulk outing, two days after a relief outing, and his Game 4 on three days’ rest are priced at his normal rates; and no individual reliever is named, because the simulator’s per-reliever usage hasn’t been validated against real postseason usage.
These aren’t the Playoff Pulse numbers. The Pulse prices the whole bracket nightly; these are standalone series runs, and the two won’t match exactly.
White Sox at GuardiansBraves at DodgersYankees at RaysPadres at BrewersRelated research
Yankees vs. Rays: The Model Likes the Team in Second Place
Tampa Bay leads the American League East and is likely to host. BBMI has the clubs meeting in 41% of simulated postseasons, and in the five-game series it considers likeliest, New York wins 57.1% of the time — without Aaron Judge — because the Rays’ third and fourth starters rate near the bottom of the league.
Meet 41% · Yankees 57.1% · 4.12 games · Rays host in 73%MLB · Postseason SimulationDodgers vs. Brewers: Where an October Series Would Turn
Milwaukee and Los Angeles have the National League's two best records, and BBMI has them meeting in 30.7% of simulated postseasons. If they do, the Dodgers win about 62% of the time — but the edge comes from the middle of the rotation, not the Game 1 aces.
Meet 30.7% · Dodgers 62% · 5.78 games · Milwaukee's opening is Game 5Awards · Cy YoungSchlittler Pulls Clear; Misiorowski and Yamamoto Lead the NL Race
September 16: Cam Schlittler leads the AL runs-saved model at 99.8%. Jacob Misiorowski leads the NL at 57.4%, with Yoshinobu Yamamoto at 37.2% and Chris Sale at 5.4%. Updated inputs, 500 simulated finishes.
Cam Schlittler 99.8% AL · Jacob Misiorowski 57.4% NL · 500 simulated seasonsSee it live
This is the “we show our work” side. The Prediction Record is the other half — every call locked before the game and graded in public.