Jake Peavy is a topic that usually elicits a negative kneejerk reaction from White Sox fans. And with good reason. While it's a bit unfair to lump Peavy in with Alex Rios and Adam Dunn because he actually has done okay when he's pitched, that sort of highlights the issue: $17 million is supposed to buy you "very good" and "makes 30+ starts".
Over in a fanpost on the right rail, there was a discussion about whether Peavy has pitched better than the results show. Those who think he has cite to his FIP, which in 2011 was significantly better than his ERA (3.31 versus 4.92). Those who think he hasn't point to his unremarkable BABIP (suggesting no bad luck) and his terrible strand rate, which was coupled with hitters teeing off on him to the tune of .355/.400/.555 when runners are on base (as opposed to .216/.246/.297 with the bases empty).
One theory for his inability to pitch with runners on base is that he is mentally weak or, to put it more mildly, he is so pissed off that he let someone on base that he loses the ability to pitch. His animated nature when bad things happen lends credence to this argument.
On the other hand, one of the other things Peavy and every other pitcher does when there's a runner on base is pitch from the stretch instead of the wind-up. And some pitchers have mechanical issues pitching from the stretch. I suggested that perhaps this was part of the answer because I had noticed throughout the season the difficulty Peavy appeared to have with his mechanics generally. I also noted that Peavy was unable to throw side sessions for most of the season and that is pretty much the only time other than during games a pitcher can effectively work on mechanics.
All this naturally led me to Peavy's Pitch F/X data at the indispensable Brooks Baseball. After the jump, I'll show you what I found. And, for those who prefer to "enjoy the game for what it is", be warned that there are graphs.
This initial set is taken from 5/18/11-5/22/11. All games were at U.S. Cellular Field. While I'm not concerned with the Pitch F/X tool being calibrated the same for my purposes in this post, there are other things that one can take from this data that would benefit from that so, for those looking at those things, there shouldn't be any issues with different calibrations. It's really the size and shape of the cluster we're looking at, wherever it may be on the graph.
First, this is the release point of Gavin Floyd.
Now, Jake Peavy.
(If you want to see Phil Humber, John Danks and Mark Buehrle, go here). You probably noticed that Peavy's is a bit more spread out than Floyd's and the others but not remarkably so. While it's generally a good idea to have a consistent release point, even for different pitches, it's not necessarily an issue for a pitcher to vary their release point (and "release point" in the singular may be a misnomer, as pitchers have different "release points" for different pitches). And, at least for this game, Peavy didn't have much of an issue as that game was arguably Peavy's best all season: a complete game shutout of the Indians with 8 strikeouts, no walks and 3 hits.
So let's take a look at a bad Peavy outing. You can pretty much take your pick of games after his ill-advised relief appearance on June 25. Here's July 10 against the Twins, in which he took a beating: 4.1 IP, 10 hits, 2 walks, 2 strikeouts and 5 runs.
It's a bit more spread out. How about some from later in the season just before he was shut down?
A similar story, probably even a bit more spread out. Now, again, this isn't per se a problem for a pitcher. And there may be some other things going on, such as moving on the pitching rubber, that is causing the variance. But the same release point for various types of pitches can be useful for deception, as the trained eye of a major league hitter is capable of picking up even slight variations. And I would argue that it's good for control and command as the pitcher is doing the same thing from the same spot, over and over, no matter whether there is a runner on base, no matter who is batting, no matter what he's throwing. For example, when you read criticism of prospects like Jake Petricka for not repeating their mechanics, this is the sort of thing being referred to.
So let's get to what may be the point of all of this. I'm not going to resolve in this post the issue of why Peavy sucks with runners on base because I got sidetracked when I went back to one of the reasons Peavy is paid so much: his 2007 Cy Young award winning season.
May 11, 2007:
September 1, 2007:
Those both look a lot tighter.
[Now, since I don't want to be accused of too much selection bias, there were indeed times that 2007 Peavy wasn't as sharp as these two examples. And 2011 Peavy did tighten things up on occasion (7/26 is one). But what you see here is representative of the release point clusters for these two seasons.]
I then wanted to take a look at the cluster for his entire 2007 season. But my go to site for that, Texas Leaguers, still doesn't have the 2007 season data up. So I did the next best thing and looked at his 2008 data. Then I compared it to his 2011 data. Unfortunately, the 2008 data seemed a bit messed up and it wasn't making sense to me.
So I went to 2009, which appeared to make more sense when compared to 2011.
2009 appears to be a tighter cluster than 2011. But, again, how much can the data be trusted? I'm not sure but I continued to plow on and I commend those of you who are still with me. Looking for confirmation, I compared his slider (the classifications of his fastballs are not as accurate in 2009 as they are in 2011, so I picked a pitch that should be correctly identified in both years). I also checked 2010.
And then I compared his curveball.
In both, 2011 looks a lot messier than 2009. And 2010 also looks better than 2011.
So what's the takeaway? I'm not totally sure at this point, in part because of the gaps in the data and how much to trust the data. I think it is safe to say that his release point is less consistent during games now than in 2007. The data from the full seasons of 2009, 2010 and 2011 also appear to show a gradually less consistent release point. It looks like he's having trouble repeating his mechanics, resulting in reduced command. It merits further and more granular analysis. Unfortunately, it's past 4am so I'm going to have to stop for now.
Whatever it is, the results for Peavy haven't been what they were in 2007. Getting back to whatever it was that he was doing mechanically in 2007 (or even 2009) may be the key to getting back to more like the results he was producing last decade.