Did Manny Parra change his approach in the last start, or did his command just improve?

I'm going to compare Parra's last start to his June 7th start against the Braves, because he only threw 56 pitches in his start against the White Sox before his demotion. I want the pitch trends, and 56 pitches isn't really enough to see what he was doing with his pitch selection.

Maximum fastball velocity:

v Braves: 92.5

v Cardinals: 93.9

Pitch f/x data can be skewed a little bit, so it's possible that Miller Park or Turner Field has a faster camera. His average fastball was almost a mile per hour faster in the most recent start. So maybe it was a mechanical change in AAA that increased his velocity or it was just a little rest.

How about pitch selection? Back in June, I suggested that Parra throw his curveball more and his changeup less-- he had thrown the change almost 20% of the time so far this year, as opposed to about 10% in the past, and his run values allowed with that pitch were quite bad. I said this:

The team has stressed that Manny is throwing his changeup more, but maybe it's time to go back to switching out some changes for curves again. We can only hope that would make the changeup more effective.

So here we'll get to the relevant portion of the data-- pitch selection. First I'll establish the trend I was concerned about. Here's Parra's pitch selection of a line as a four pitch slash-- Fastball, Changeup, Curveball, Slider. He seems to throw a few splitters yet, but they're pretty indistinguishable from the change. If I couldn't determine what a pitch was and it wasn't a split or change, I just left it out. The splitters and changeups are in their own category.

Here's Parra's pitch selection lines from the past two years overall:

2008: 58%/ 17%/ 20%/ 5%

2009: 56%/ 24%/ 14%/ 6%

Pretty easy to see the difference there. Less fastballs, a lot more changeups, and fewer curveballs. And I think he's thrown fewer splitters as well, so we have to add up that 7% increase in changes/splitters to even more changeups than it would appear. 

So lets check out these two particular games:

v Atlanta, 6/7/09: 44%/ 30%/ 17%/ 8%

v St. Louis, 7/9/09: 65%/ 10%/ 18%/ 8%

And there it is. More fastballs, better velocity, and far fewer changeups and splitters. I'd imagine those 10% were much more effective because he worked them off of his fastball than if he had thrown almost as many changes as actual fastballs, as he nearly did in Atlanta.

So it's been one start, but in that one Parra did make significant changes that were right in line with what I suggested soon before his demotion using his Fangraphs player page and pitch f/x data to look at his effectiveness and pitch selection. If he sticks with the new approach, there's no reason to think he won't continue to be effective.