The most important decision a manager has to make in any individual postseason game usually involves when to pull his starting pitcher. There can be other important decisions -- whether to bench your best hitter, for example, or whether to bring in your sixth-best reliever in a tie game in the seventh inning -- but baseball games revolve around pitching, and it's the starter who has to carry the biggest workload.
The question, then: When should a manager remove his starter?
Obviously, there are myriad influencing factors in any game: how the starter feels, his pitch count, how many days of rest he's pitching on, the score of the game, how tired or rested the bullpen is, the quality of the relievers, the state of the series and so on.
So we're talking in broad terms here. One of the hot topics among sabermetric writers and analysts this offseason has been the idea that starters generally do worse the third time through a batting order. The batters have seen him twice by then, plus the starter is getting tired. It's certainly no coincidence that both times Clayton Kershaw blew up against the Cardinals came in his third time through the order as he approached and went beyond 100 pitches.
Here are the numbers we're talking about, all starting pitchers in 2014:
First time through the order: .246/.304/.377
Second time through the order: .256/.313/.395
Third time through the order: .268/.327/.421
The most hardcore sabermetricians will advocate for a quick hook; overall, relievers have lower ERAs than starters, so the theory is that going to your bullpen over a tiring starter is the way to go. Dave Cameron of FanGraphs wrote a piece the other day praising Buck Showalter for his quick hooks in the Orioles' series win over the Tigers. Buster Olney wrote a few days ago that there's no perfect time to remove a starter.
And it's hard to say that there should be a hard and fast rule. If managers always managed like that, we wouldn't have had Jack Morris pitching his 10-inning shutout in the 1991 World Series or Chris Carpenter beating Roy Halladay 1-0 in Game 5 of the 2011 NL Division Series or any number of great postseason performances. You have to allow for a manager to adjust to what's going on in the game.
Anyway, I thought it would be interesting to look back and see how World Series winners have managed their rotations in recent years.
2013 -- John Farrell, Red Sox (16 total games)
Average batters faced: 24.2
Long outings (28+ BF): 3
Short outings (20 or fewer BF): 5
Farrell did extend his starters a few times, but all were in games when the Red Sox had big leads: 6-1, 8-1 and 12-2 were the finals of those three games. The Red Sox won three of the five short outings, including Game 4 of the World Series when he pinch-hit for Clay Buchholz with the score tied 1-1 in the top of the fifth. There were some extenuating circumstances as Buchholz was pitching through a sore shoulder that was limiting his velocity. But Farrell also pulled Jake Peavy after 74 pitches in the sixth inning of Game 4 of the division series. The Red Sox were down 1-0 and Peavy hadn't walked a batter; they ended up winning 3-1. In Game 5 of the ALCS, he pulled Jon Lester after 24 batters in the sixth inning with a 4-1 lead. Lester was at 98 pitches and there were two runners on, but Farrell didn't wait.
2012 -- Bruce Bochy, Giants (16 total games)
Average batters faced: 23.7
Long outings: 3
Short outings: 4
Two of the long outings came with big leads. The one exception was Matt Cain in Game 4 of the World Series, when he faced 28 batters. He was at 102 pitches and had retired the side in order in the seventh, but Bochy pulled him with the game tied. The Giants would win in 10 innings.
The Giants won two of the short outings. In Game 3 of the division series (the Giants were down two games to none), Ryan Vogelsong was removed after 20 batters (and five innings). The game was tied 1-1 and Vogelsong led off the sixth; plus he was at 95 pitches, so that was strongly dictated by circumstances. In Game 4, Bochy removed Barry Zito in the third inning, after 20 batters faced. The Giants were ahead 3-2 at the time. Bochy's decision was certainly influenced by Zito's four walks, but he took him with two outs and a runner on first, not the most threatening of moments. In the same game, Dusty Baker left in Mike Leake to give up two more runs in the fifth inning and the next game he left in Mat Latos to give up six runs, including a grand slam to Buster Posey the third time through the order.
2011 -- Tony La Russa, Cardinals (18 total games)
Average batters faced: 22.0
Long outings: 3
Short outings: 6
La Russa had a very quick hook throughout this postseason, with five other outings of 23 or fewer batters. Two of the long outings were from Chris Carpenter, including that memorable duel with Halladay, when he faced 31 batters. In the ninth inning, La Russa left him in to face Chase Utley, Hunter Pence and Ryan Howard. He probably shouldn't have, but it worked out. Sometimes it does.
It's possible La Russa adapted after losing Game 3 of that division series. Jaime Garcia took a 0-0 tie into the seventh inning but gave up a single, intentional walk and a two-out, three-run homer to Ben Francisco (pinch-hitting for Cole Hamels, so he was the 27th batter Garcia had faced). After that, La Russa was determined not to let his starters lose a game late.
2010 -- Bruce Bochy, Giants (15 total games)
Average batters faced: 25.7
Long outings: 6
Short outings: 2
Bochy rode his starters longer this postseason, as he also had four starts with 27 batters faced. In Game 1 of the division series, he let Tim Lincecum finish off a 1-0, 14-strikeout gem with 119 pitches and 30 BF. In Game 5 of the World Series, leading 3-1, he let Lincecum face the 9-1-2 batters in the eighth inning, but Lincecum retired the side.
Bochy also had two interesting quick hooks, however. In Game 4 of the NLCS against the Phillies, he removed rookie lefty Madison Bumgarner in the fifth inning after 20 BF, a 2-1 lead and two runners on. The move backfired at first, as Santiago Casilla allowed the two inherited runners to score plus one of his own, but the Giants rallied to win 6-5. The critical one came in the clinching Game 6 when he removed Jonathan Sanchez in the third inning of a 2-2 game. Sanchez had walked a batter and hit a batter. Jeremy Affeldt got out of the jam and Bumgarner would pitch two scoreless innings, Lincecum would retire a batter and Brian Wilson got a five-out save.
That was some unconventional managing and it helped the Giants win the series. But to manage like that, Bochy had to have a plan of attack ready in place in case Sanchez faltered.
2009 -- Joe Girardi, Yankees (15 total games)
Average batters faced: 25.9
Long outings: 4
Short outings: 1
Girardi was pretty much by the book. All four long outings came from CC Sabathia, and the short one was a blow-up A.J. Burnett start in the World Series (two innings, six runs). He did have a quicker hook on Andy Pettitte, but that was in part because Pettitte made some starts on three days' rest.
* * *
Is there anything to learn from this? In the five postseasons from 2009 to 2013, there were 175 postseason games (so 350 total team games). There were 72 "long outings" of 28 or more batters faced -- 21 percent of all games. You'd think the team with the long outing would win most of those games, right? After all, you're usually leaving in a pitcher that long only if he's been pitching well or has a big lead. The long-outing teams were 47-25 (.652), but the starting pitcher lost 17 of those 25 games. Not all of those were bad losses -- Halladay faced 32 batters in losing to Carpenter, for example.
But some were bad decisions. In the 2011 division series, Charlie Manuel left in Cliff Lee to face the Cardinals' 2-3-4 hitters for a fourth time in a 4-4 game. Allen Craig tripled and Albert Pujols singled and the Cardinals won 5-4. (Meanwhile, La Russa yanked Carpenter after 16 BF and the bullpen threw six shutout innings.) In Game 1 of the 2011 division series, the Brewers led the Diamondbacks 2-0. Kirk Gibson let Ian Kennedy face Prince Fielder a fourth time and Fielder hit a two-run homer, cementing the game for Milwaukee.
By the way, in the Oakland-Kansas City wild-card game, holding a 7-3 lead in the eighth, A's manager Bob Melvin let Jon Lester face the first four batters a fourth time. Three of them reached base.
Looking ahead, we know Buck Showalter will have quick hooks and Ned Yost will go to his back-end trio if he's leading in the seventh inning. (The fifth and sixth innings will be Yost's test.) Bochy isn't afraid to pull a starter quickly -- Vogelsong and Peavy faced 21 in their starts against the Nationals, both leaving with leads -- although he'll go longer with Bumgarner.
That leaves Mike Matheny as the key guy in this area. For the most part, he's pretty by the book. In the 34 postseason games he's managed, only twice (Adam Wainwright both times) has a starter gone beyond 27 BF. But one of those was Game 5 of last year's World Series, when the Red Sox scored twice in the seventh to take a 3-1 lead -- with the 26th and 28th batters Wainwright had faced knocking in the runs. He also left in Michael Wacha in Game 6 to give up six runs when a quicker hook in a must-win game was necessary.
Of course, none of this touches on that gray area around 24 batters faced -- that crucial sixth- or seventh-inning time when a starter is tiring and managers are loath to use their setup guys too early. But that's another article.