August 8, 2012
Zach McAllister's Secretly Not-So-Good Season
On Monday, Zach McAllister was pitching at home for the Indians. It was the second inning, and he wasn’t having his best game: two solo home runs to lead off the second inning had him in an early hole. Then there was a double, and a walk, but with two outs he got the grounder to second base he needed to get out of the inning. Jason Kipnis threw poorly, Carlos Santana stretched poorly.
McAllister is having a fine season. He entered the game Monday with a 3.42 ERA; he left it with a 3.60 ERA. His FIP is 3.71, and he is striking out more batters now than he has since he was a 19-year-old in short-season ball. He has nine game scores higher than 50 in 12 starts. PECOTA didn’t expect much from McAllister this year—a 4.69 ERA—but he has been a nice surprise. No Indian who has started even one game is within a run of his team-leading ERA. Those are all things people talk about when they talk about pitching.
But then we poke around and find, hidden in McAllister’s basement, the deformed Olsen triplet of pitching statistics: unearned runs. Because of a fluke of baseball history, most pitchers get to ignore these blemishes. The analytical set quite reasonably refuses this logic, but for the most part the detente holds. Unearned runs are, mostly, not a big deal.
McAllister could have been part of this sweet arrangement. But folks only look the other way up to a point, and McAllister—who allowed his 12th unearned run in 12 starts on the Kipnis error—went too far. People who go too far have to be dealt with.
The question is, is this true? Sure, McAllister has given up a lot of unearned runs, and I'm comfortable blaming him for those runs, but there are all sorts of ways to give up a lot of unearned runs without caving in. He could be allowing too many runners just before the errors, and, of course, the errors themselves set up run-allowing opportunities. Twelve unearned runs—now 13, after Revere's single—is a lot for 12 starts, but it’s not so much that it necessarily supports broad conclusions. So here is a quick accounting of how McAllister has done after each of the errors Cleveland defenders have committed with him on the mound this year.
Error no. 1: McAllister himself makes a throwing error on a pickoff attempt. Subsequent batters in the inning: 0-for-3. Run scores on a sacrifice fly.
Hey, so far it looks like McAllister hasn't been doing anything wrong at all. All this talk about him caving in is just some blinking fertilizer. Meanwhile, in Monday's game:
Manning: “He had a chance to get out of it and the floodgates opened after that error.”
OK, let's keep going.
Error no. 4: Third baseman makes a fielding error with two on, one out. An earned run scores; two on for Mike Trout. Subsequent batters: 3-for-5, two home runs. McAllister had been pitching a shutout.
Back in Cleveland:
Manning: "Look at that. Fifteen unearned runs now. He leads the league."
Overall, the batters McAllister had faced after errors (before Monday) were 6-for-19 with two walks, 14 total bases, and seven strikeouts. The size of those numbers puts them in perspective, as 21 plate appearances are the equivalent of about three-fourths of one start. There’s no lesson worth drawing from 21 plate appearances, probably; 21-plate appearance samples exist for people to manipulate to make any dumb point they want. Watch me do it: after an error behind him, Zach McAllister has struck out one third of the batters he has faced in the same inning, so must be pretty great at buckling down!
Here’s another way of looking at the same information. On the left are the run expectancies of the situations that McAllister’s defenders put him in. On the right are the actual runs he allowed (earned AND unearned) in the rest of those innings. (In one case, he left the game with runners on, so I credit him with also allowing the “runs” of the run expectancy at which he departed.)
Three runs is three runs, and pennants have turned on three runs. But most of the damage credited to McAllister after errors was actually in place before the errors, or as a direct result of the errors.
In Cleveland, things get worse:
Manning: "He could not recover. He made 21 pitches since that error and he just never got over it."
McAllister on Monday made it hard to continue defending him, even if reasonableness suggests we keep defending him. After the error, he gave up three hits and a walk. He left two runners on base, and both scored when his reliever allowed a Ryan Doumit home run before getting out of the inning. That means McAllister has now allowed 18 unearned runs, out of 46 total runs. He has allowed 18 unearned runs in 70 innings; Mike Fiers has allowed 18 total runs in 80 innings. So McAllister has visibly been pretty good this year, but secretly been fairly lousy.
To put this all into perspective, 39 percent of the runs he has allowed came as the result of scorer-declared miscues. Since 2000, nobody who has allowed at least 45 runs in a season has had even 30 percent of those runs be declared excusable. Only five other pitchers have had their unearned runs reach even 25 percent. (As a group, they reverted to a more typical 11 percent the following year.) So McAllister is definitely doing something. Whether he’s, you know, doing something is still debatable.