Making predictions in baseball isn’t actually about predicting the future. We know from decades of experience and more than a century of history that you cannot predict baseball. Instead, predictions act as a jumping-off point for interesting discussions. Dave Cameron knows this. Before the season began he made a series of predictions that included James Shields as the AL Cy Young winner. We used this to create an interesting conversation on FanGraphs Audio, during which I expressed some confusion over the pick. Shields has always been a solid pitcher, but he never struck me as a Cy Young candidate. So what did Dave see in him?
Whatever he saw became evident to everyone earlier in the season. Shields mowed down opponents like he never had before. Through his first 10 starts he not only held a 2.99 ERA, but had struck out 71 in 69.1 innings, far exceeding his career average. It might have been a fluke, since we know that anything can happen in 10 starts. Watching Shields, though, something felt different. It was as if he had figured something out during the off-season and was putting it to use on American League hitters.
The difference was most noticeable with his changeup. He threw the pitch at nearly the same frequency as 2009, but the results were much better. While in 2009 he got batters to whiff on 17.4 percent of his changeups, during his first 10 starts in 2010 he got an astonishing 26.1 percent swings and misses. He had generated a few more swings in general, but the increase was not at all in line with the swings and misses. Combine that with a lower foul-ball rate, and it seems like the change really turned a corner.
After that 10th start, an eight-inning, two-run effort against Boston, Shields hit something of a rough patch. In his next eight starts he pitched just 47 innings, striking out 38. Worse, he had allowed a ton of runs in that period, a 7.66 ERA, bringing his season total to 4.87 after the Indians scored four runs in 6.1 innings against him on July 9. That stretch certainly changed the view of Shields’s season. For starters, Cameron stopped bragging about his awesome pick. But it also signaled that maybe the early season run was just that, a run of excellent starts. Pitchers have them all the time, so why should Shields be any different?
When I went to check Shields’s pitch selection for this period I expected to see that he reduced his changeup usage, instead dipping into the other pitches in his deep arsenal. Surprisingly, I saw an insignificant drop, down to 22.5 percent. He was still generating plenty of swings and misses, 20 percent, but not as many as earlier in the season. On the surface it might look as though he was just getting lucky with the changeup early in the season, and that hitters had finally figured out how to handle it — relatively, at least. Yet looking a bit deeper, this might not have been the case at all.
A fledgling yet intensely interesting aspect of baseball analysis is pitch sequencing. It’s interesting because so much of the batter-pitcher matchup relies on how the pitcher sequences his pitches. It’s intense because even though two pitchers may throw the same pitches in name, their pitches still have different effects. In other words, pitch sequence efficiency and effectiveness is going to vary pitcher-to-pitcher. I don’t have any analysis on Shields’s actual pitch selection, but rather an observation from his first 10 starts compared to his next eight.
In the first 10 it seemed as though he leaned on his fastballs and changeup more than the rest of his repertoire. To wit, he threw his four-seamer 20.1 percent of the time, his two-seamer 26.4 percent, and his changeup 23.5 percent, totaling 70 percent of his overall pitches. Yet during his rough patch he turned to the other pitches in his arsenal more frequently, going to those three pitches 61 percent of the time. He actually did increase his usage of another fastball, his cutter, though that is more complementary to his slider.
Could it be that moving away from the two-seamer changed his sequencing and therefore made his changeup more hittable? Without a strict sequencing mechanism in place it’s tough to say. Shields did improve in his first three starts after the All-Star break, though in this case we saw him use his changeup even less, 16.8 percent, and his four-seamer more, 32 percent. Yet in these starts he turned to his curveball more often and generated a 14.8 percent swinging strike rate, over 10.2 percent on his changeup. So was the pitch less effective in general, or just less effective because he’d been using it less frequently?
If yesterday’s start was any indication, it’s the latter. Sheilds attacked the Yankees with almost 70 percent four-seamers and changeups, using the changeup for 31 of of his 116 pitches. With it, he generated 35.5 swinging strikes, leading to 11 strikeouts. When he wasn’t getting hitters to swing and miss he was getting them to hit the ball on the ground, as only six out of 16 balls in play got some air under them. It was an all-around dominant start, made possible almost exclusively with the changeup.
Shields’s odd season makes it difficult to pinpoint what he’s doing well and what is luck. As we saw, he used his changeup about the same in his first 10 starts as he did his next eight, yet he saw drastically different results in those starts. Then he turned away from the change and saw positive results, and then had what might have been his best start of the season yesterday by leaning heavily on the change. So what works best for him overall?
It’s a fallacy to say that because his changeup is so good that he should throw it more often. If hitters know it’s coming it’s easier to lay off and therefore won’t be as effective. Yet with Shields’s case that might not be the case. As our own R.J. Anderson said in the River Ave. Blues Yanks-Rays series preview:
Shields is such a weird case. You’re talking about a guy with maybe the best changeup in the American League who says and does intelligent things all the time; I would not be shocked to see him become a pitching coach down the line because he helps teammates with mechanical issues and philosophical talk. He seems to understand game theory and he’s even said this season that he likes it when team’s ambush him and figure him out. Presumably so he can mix things up. You can see that attitude prevalent in his arsenal too. He could probably just throw the changeup all day and night with good results – ask Nick Swisher – but he goes to a cutter, he goes to a curve, and sometimes maybe he outthinks himself. Joe Magrane used to say he gave the hitter too much credit by not throwing his change more often and maybe he had a point.
It does seem that leaning heavily on the changeup works, especially when it goes along with a correspondingly heavy usage of his fastball. Shields does have other weapons, and by all means he shouldn’t let them become stale. But with an all-world changeup, he might not need those weapons unless the changeup fails him here and there. It doesn’t seem, though, like that happens too often.
Joe also writes about the Yankees at River Ave. Blues.