Removing Max Scherzer: The Right Call

In Game 2 of the ALCS, Max Scherzer was utterly dominating, striking out 13 of the 25 Red Sox batters he faced. Despite all the strikeouts, he managed to get through seven innings on just 108 pitches, mixing in some efficiency with his ability to avoid contact. The Tigers held a 5-1 lead when Scherzer was replaced to start the top of the 8th inning; by the time the inning was over, the game was tied, and the Red Sox won in the bottom of the 9th to tie the series at 1-1.

On Saturday night, Scherzer was again very good, but not quite as dominating as his first outing. He struck out 8 of the 27 batters he faced, but struggled with his command at times, issuing five walks in the process. His last walk was the last batter he would face in 2013, as a free pass to Xander Bogaerts put runners at 1st and 2nd with only one out, and clinging to a 2-1 lead, Jim Leyland opted to replace Scherzer with Drew Smyly to go after the left-handed hitting Jacoby Ellsbury. Ellsbury would reach on an error, and then Shane Victorino hit the grand slam, and the ALCS was over.

Twice, Max Scherzer was removed from the game with the Tigers having the lead. Twice, Tigers relievers coughed up the lead by giving up a grand slam. Had the Tigers bullpen protected those leads, they very likely would be playing St. Louis for the World Series title, but instead, they have to sit at home and wonder what could have been. And it’s easy to wonder whether Scherzer could have done better than his bullpen had he been entrusted to pitch just a little bit longer.

Fox broadcaster Tim McCarver certainly thinks removing Scherzer was the wrong call. Upon seeing Jim Leyland walk to the mound to give Scherzer the hook, McCarver began to extoll the virtues of letting starters pitch deeper into games than they do now. A transcript of his comments on the matter, for posterity’s sake, beginning after Scherzer is replaced:

“Well Jim Leyland told us that Scherzer told him, in game 2, he was finished after seven innings. He couldn’t have told him this time because he was on the mound. I don’t think he told him before the inning started, one wouldn’t think.

Lecture interrupted by ongoing baseball. Jacoby Ellsbury reaches on an Iglesias error that should have been an easy out and maybe an inning ending double play. Then there’s another commercial break as Leyland removes Smyly and brings in the right-handed Jose Veras to go after Shane Victorino. Play resumes, and three pitches later, Victorino does his thing. Tigers lose the lead, and the series, with Veras on the mound and Scherzer watching. And then we go back to the lecture.

“Six days ago, we saw a similar reaction from the crowd in the right field bleachers, as David Ortiz hit a grand slam. Now, it’s Victorino.”

Joe Buck: “The common denominator: Both games started by Max Scherzer, and grand slams allowed by pitchers not named Max Scherzer.”

McCarver, again: “Boy, that’s well put. How deflated can you be?”

More baseball play by play from Buck, as McCarver says nothing while Veras strikes out Dustin Pedroia. They proceed with play by play as normal, as the inning ends and the Red Sox set down the Tigers in order in the top of the 8th. We go to the bottom of the 8th, where Al Albuquerque is now pitching.

Buck: “It will be a field day for reporters and analysts after this game, wondering why Max Scherzer came out of this game in the bottom of the seventh. We’ll get that story after the game.”

McCarver: “Jim Leyland said that Max Scherzer told us he was through after seven innings in game 2, when the Tigers had the big lead 5-1. Well he didn’t tell him tonight, because Jim had to come out and come get him. You talk about pitch counts, pitch counts, pitch counts, and we’ve talked about it ad nauseum, you can pay attention — in my opinion — you can pay attention to pitch counts, but you can’t be a prisoner to them. And to me, the majority of baseball — the mangers and the people who run this game — are becoming prisoners to pitch counts. And in my view, that’s wrong. That doesn’t make me an old timer, a purist, or anything else. That’s just wrong. Pay attention to them, but don’t be a prisoner to them.”

Buck: “Scherzer had thrown 108 pitches in game 2, and threw 110 pitches tonight, and then Jim Leyland went and got him. And you saw this bullpen for the Tigers, that will be a big topic for conversation too.”

One could argue that, without the context of the situation being discussed, McCarver’s comments might have some merit. You shouldn’t be a slave to pitch counts. There’s no magic single number at which every pitcher is finished, and pitchers shouldn’t be lifted from the game as soon as they cross the 100 pitch threshold simply because it’s a round number. But there’s no way to ignore the context of the discussion, and this was a critique of Leyland’s decision to remove Scherzer in Game 6 of the ALCS.

Given the result, some second guessing is natural, and it’s easy to assume that Scherzer could have and would have done better than his relievers. But a look at the numbers strongly suggests that going to the bullpen was the right move. We’ll deal primarily with Game 6, since Scherzer essentially took himself out of Game 2, and it would be silly to suggest that a manager should force a starter to go back to the mound when he feels he has nothing left.

At the point at which Scherzer was removed in Game 6, he had gone through the Red Sox starting line-up exactly three times. Here’s how the Red Sox had fared against him each time through the order:

1st PA vs Scherzer: 1 for 7 (1B), 2 BB, 4 K
2nd PA vs Scherzer: 1 for 8 (2B), 1 BB, 2 K
3rd PA vs Scherzer: 2 for 6 (1B/2B), 2 BB, 1 HBP, 2 K

The first two times through the order, Scherzer allowed a grand total of five baserunners. The third time through the order, five of the nine hitters he had faced had reached base. Scherzer, at the point he was replaced, had not been recently dominating Red Sox hitters. And this shouldn’t come as any huge surprise.

I’ve shown this chart several times, but it bears repeating given the storyline here. Here is the league average batting line against a Major League starting pitcher in 2013, based on times through the order.

League	BA	OBP	SLG
1st PA vs SP	0.250	0.310	0.390
2nd PA vs SP	0.259	0.319	0.411
3rd PA vs SP	0.270	0.331	0.429

At the beginning of a game, starting pitchers are mostly quite good, but they begin to falter as they face the same hitters repeatedly on the same day. There are multiple factors at play here, including the pitcher himself tiring, as one’s pitch count is naturally much higher the third time through the order, but the fact remains that starting pitchers lose effectiveness as the game goes along. Not every pitcher decays at exactly the same rate, but even Scherzer is not immune to this. His career numbers, based on times facing a hitter within the same game:

Scherzer	BA	OBP	SLG
1st PA vs SP	0.226	0.296	0.372
2nd PA vs SP	0.253	0.308	0.425
3rd PA vs SP	0.258	0.325	0.404

Scherzer’s OBP and SLG allowed are both 30 points worse the third time through the order than they are the first time he faces a hitter that day. The league average, this year, was 20 points of OBP and 40 points of SLG. Scherzer, when he’s tiring, appears to lose his command more than his stuff. That held true the third time against the Red Sox line-up, as he walked two and hit a batter, and seemed to be losing effectiveness at a pretty rapid pace.

The ~100ish pitch limit might seem arbitrary and silly, but it actually corresponds very well to a pitcher getting through an opponents batting order three times. In the AL this year, the average PA lasted 3.86 pitches. Multiply that by 27 batters, and you’re at 104 pitches, on average, after going through the line-up three times. It isn’t so much that a pitcher’s 105th pitch is guaranteed to be a meatball as it is that 100 pitches corresponds to the point in the game at which the hitter begins to have a significant advantage over the starting pitcher, due to both pitcher fatigue and hitter familiarity.

It is entirely incorrect to look at Max Scherzer, at the point he was removed in the 6th inning on Saturday night, and expect him to pitch at his own personal averages against the next few batters. At that point, given that he would be facing each hitter for a fourth time, and was reaching the end of his own endurance, you would have to expect Scherzer to perform worse than his overall averages that include early game performances.

How much you expect him to decline is tricky, but you have to decay his expected performance by some amount. Maybe it’s not as much as his career numbers suggest, given that he’s a better pitcher now than he was when it began, but given what we know about the results of the batter/pitcher confrontation as the game progresses, it has to be a significant figure. For sake of argument, let’s use Scherzer’s career 30 point decay in both OBP and SLG.

Left-handed hitters, this year, hit .218/.278/.367 against Scherzer this year. Had he stayed in to face one more batter, that batter would have been the left-handed hitting Ellsbury. If we add 15 points of OBP/SLG to that line — half the 30 point gap, since we are estimating difference between 1st and 3rd time through the order — that would go up to .293/.382, or a .685 OPS. Drew Smyly, the pitcher who actually got to face Ellsbury, has held left-handed hitters to a .204/.254/.315 career line, and was even better this year, giving up just a .476 OPS against left-handers. And this doesn’t even factor in Ellsbury’s own platoon splits, which see him fare much better against right-handed pitchers.

There is simply no way to think that Scherzer vs Ellsbury, for the fourth time in a game in which Ellsbury is 1-2 with a walk against him, is a better match-up for Detroit than Smyly versus Ellsbury. Given a one run lead, and the tying run in scoring position, leaving Scherzer in to face Ellsbury in that situation would have been borderline managerial malpractice. That at-bat called for a pitching change so clearly that every manager with any understanding of the probabilities would have gone to his bullpen. It was, quite clearly, a huge improvement in expected outcome for the Tigers.

And the results even back this up. Smyly got a potential double play ball, leading Iglesias to a spot where he could flip the ball to second for a force and there would be a play at first base to try and double Ellsbury up and end the inning. At the worst, that should have resulted in a 1st-and-3rd, 2 out situation, with the Red Sox needing a hit to tie the game, and some percentage of the time, the Tigers turn that double play and the inning ends. The move made sense, and it worked, up until the point that Jose Iglesias tried to field a routine ground ball.

But maybe, you argue, that by removing Scherzer to get the advantage against Ellsbury, you’re setting yourself up to have a less good right-hander go after Victorino and Pedroia. Yes, you get the advantage in that one at-bat, but wouldn’t you rather have Scherzer up there against the two righties?

No, no you would not. Right-handers, in his career, against Jose Veras: .212/.306/.336. Right-handers against Veras in 2013: .165/.254/.266. Veras is an excellent right-on-right reliever, and his .520 OPS allowed to RHBs this year is not that much higher than Scherzer’s average .494 OPS against right-handers. And that’s Scherzer against all-handers, including the ones he gets to face while at full strength early in the game. Add in the 15 point decay in both OBP and SLG, and Scherzer versus RHBs is now, at best, equal to Veras against right-handed hitters. And you had to give up a significant advantage against Ellsbury to get to that same place.

It is very easy, in retrospect, to think that Jim Leyland put too much faith in a bad bullpen. In reality, though, Jim Leyland put the right amount of trust in a pretty good bullpen. Smyly against a lefty was a great match-up for Detroit. Veras against a righty was a very good match-up for Detroit. Both of these were better match-ups than Scherzer versus either batter.

It didn’t work, in large part because a great defensive shortstop botched a play that he makes almost every time. That it took a Jose Iglesias error to keep the inning alive should be a poignant reminder that players who are good at a thing still fail at that thing sometimes. Jose Veras failed at getting Shane Victorino out on Saturday, but that doesn’t make the decision to remove Scherzer from the game a poor one.

Tim McCarver might hate pitch counts, and he might think that Leyland should have stayed with Scherzer. But, to quote the man, that doesn’t make him an old timer, a purist, or anything else — it just makes him wrong.