Who's that? (Hey, we put his name in the title!) It's James McDonald, given a grand total of five starts by the Dodgers in 2009-10 before being summarily exiled to Pittsburgh for two months' worth of Octavio Dotel. This is one of those type of trades that all manner of sabe types (classic, neo, post-neo, meta-post-neo...) universally despise, even in the context of a "trade deadine" stretch run, and particularly when the pitcher has gotten so few opportunities in the role he'd been envisioned in by the organization while they were developing him.
(OK...yes, Clayton Kershaw supporters, the 2012 CYA winner is still tops. But imagine what the Dodgers might be like had they kept McDonald in the place of, say, Aaron Harang.)
Armed with extra battalions of spare relievers and starter reclamation projects in '10, Torre simply took one more look at McDonald--a mid-season start after a stint in AAA--and when that one didn't turn out so well, they walked him off the gangplank and into the Pirates' leaky little dinghy.
McDonald worked out some kinks in the Pittsburgh rotation last season, pushing past long-ball and control issues. Smoothing out his delivery, and developing a baffling new pitch (an off-speed slider), James has started 2012 with a bang: his QMAX score of 5.20 (2.5 S, 2.7 C)is nearly two points lower than what he averaged last year. A 90% success square ranking, while clearly not sustainable, is pretty electrifying.
Let's spend the last portion of our time here looking a little more into what's behind the current uptick in McDonald's performance. Craig Brown over at Fake Stats weighs in regarding the new pitch and how it has changed the landscape for James: we went a bit further and broke out the Brooks data into a chart that shows the distribution of pitches for 2011 and 2012 by count.
Legend: FA-fastball; SI-sinker; SL-slider; CU-curve; CH-changeup; A-Ahead; B-Behind; E-Even (1-1, 2-2), 2K-with 2 strikes, 3B-with 3 balls; 1st-first pitch (0-0 count)
What we are still missing in the data coming to us from the PitchF/X suite (so far as we can tell, at any rate...) is the type of pitch thrown on the precise count that decides the outcome of the plate appearance. So the best we can do with the above data is show the distribution of the pitches thrown on each count, as captured in the Brooks data (but not broken out in the format we're showing here).
When we do that, we see that the marked increase in slider usage by McDonald is probably (emphasis mine) helping him on the first pitch (of course, there's no way he's going to hold opposing hitters to a .481 OPS on the 0-0 count all year--that's where some of the regression is going to come from). More subtly but possibly more sustainable, however, are the gains show at 0-1, 1-0 and 2-1.
James is throwing many more sliders on the first pitch, more sinkers, less fastballs, and half as many curves. He's cut back on the curve ball on two-strike counts, but this doesn't seem to be making a lot of difference--he was already effective with two strikes, as are most pitchers. By throwing the slider more on the 1-0 count, James seems to be able to keep the ball in the park: the 1-0 count was a bit of a disaster for McDonald in '11.
More cross-mapping of this data just could tell us some things, and might be the place where we'll be able to verify how pitcher performance alters, for both good and bad. So far, these indicators have been applied in a manner that's mostly global: it's probably time to start seeing how well they can work when applied with greater levels of granularity.
In the meantime, keep an eye on J-Mac. He's wearing Don Drysdale's number (53) and right now he's doing a pretty good imitation of ol' Big D.
