There was some controversy when Jake Peavy pitched on short rest in his last start in Arizona. As it turns out, Peavy got hammered. It was by far his worst start of the year giving up 8 runs in 4 innings. It’s certainly tough to tell how much the short rest affected the poor performance, but you have to figure it at least had some impact on Peavy’s night. I thought I’d take a peak into the gameday data to see if he did anything different on short rest.

Let me first say that these first few graphs are very messy. If you don’t see anything here, I don’t blame you. The picture is even worse when you transfer it into wordpress and there are a lot of colors and a lot of dots, etc. I figured since I did it, I’d share it, but I’m not sure if you’ll get anything out of it. Anyhow, I retrieved every Peavy start from August 2nd to September 5 (except his Aug 27th one … no data there) and attempted to see if there was any difference in what he brought on Wednesday night. The first one here is simply pitch speed on the y axis and pitch count on the x axis. I tried to make all of the other starts lighter colors and his last one black.

There are some missing pitches early in his last start. Again, not sure if there’s anything here. He definitely didn’t have his best velocity, so that may be something. Later in the game it appears his lost some speed on his fastball and it wasn’t quite coming as fast as usual, so I suppose that could be a sign of tiring.

The next one is movement, with horizontal break on the x axis and vertical on the y. One thing I noticed here was the lack of pitches in the area that I circled. His pitches, overall, seemed to have a lot of movement, but maybe a little less velocity as we saw above.

Alright, enough with that for now. Since I have all of these starts in a spreadsheet, I figured I’d take a look at all of his pitches. Here are a few more graphs with all of Peavy’s pitches (that were tracked) since August 2nd.

This one is just start speed vs end speed. You can see 4 different pitches here, I think. In the 90+ range you have a large cluster of fastballs. Next, in the 85-89 range you have a grouping of sliders. From 80-84 is another group of sliders (or slurves). And then finally in the low to mid 70′s range are the rare curve balls (and maybe a few changes).

The next two are velocity vs horizontal and vertical movement. On the second graph, the only negative vertical breaks are Peavy’s rare curveballs. In the data I have for Peavy, he only has 37 pitches with a negative vertical break value. 23 of those pitches were to lefties, so it appears Peavy like to unveil the curve against left handers — in the rare times when he chooses to use it at all.

That’s it for now. As I always say at the end of these … I’m just getting used to them … just messin’ around a little. There’s a lot of great (much more refined) work being done out there and you can find it all at Fastballs. I don’t know if you’ll find anything useful in what I just posted, but I figured I’d throw it up anyway. If you like it, don’t like it, or have any suggestions for me … let me know.

As reader (and fellow blogger) sky pointed out in my earlier post, you can attempt to classify pitches using break and velocity. So I tried to do that with Peavy in his last start at Shea …

The x axis is deviation in inches from a theoretical pitch with no spin. The pink data points are the vertical breaks of the pitch and the blue ones are the horizontal breaks. So there are two data points for each pitch and they are plotted by velocity on the y axis. I’ll let Doctor Nathan describe exactly what these breaks are:

pfx_x,pfx_z: The deviation (in inches) of the pitch trajectory from a straight-line in the x (horizontal) and z (vertical) directions between y=40 ft and the front edge of home plate, y=1.417 ft. It is important to note two things. First, the initial value is y=40 ft, regardless of the value of the initial value y0 (defined below). If the pitcher’s release point had been used (approximately y=55 ft), then the deviation would have been nearly twice as large. Second, the effect of gravity has been removed from pfx_z, so that both parameters are the “break” of the pitch due to the Magnus force on a spinning baseball. Note that the online Gameday reports the quantity pfx, which is presumably the square root of pfx_x²+pfx_z². Given our sign conventions, a positive value of pfx_x cooresponds a deviation to the catcher’s right and a negative value to the catcher’s left. Similarly, a postive value of of pfx_z is a pitch the drops less than it would from gravity alone (most pitches fall in this category), whereas a negative value is a pitch that drops more than from gravity alone (e.g., a “12-6″ curveball).

Yea, what he said. : )

You can see at least 3 pitches that Peavy worked with in this game. He threw a fastball (probably 2 and 4 seamers), an 85-88 mile per hour slider, a slower 79-83 mile per hour slider (I think), and a few change ups (again, I think …).

I believe there’s a better way to display this with a graph, but I’m still working on my “graphing skills”. Anyhow, I think it’s interesting to actually visualize this stuff. It’s a nice compliment to hearing somebody say what pitches someone throws or trying to remember what you saw on tv.

While, perhaps, I haven’t quite precisely classified what Peavy throws, at least we have a glimpse of what he tossed against the Mets on August 22nd.

Of course, that’s a relatively small sample. The next step is doing this with more than one game and diverging into different kinds of analysis. Again, some people are really doing great work with this data. Check out the links I provided in the first Peavy post and/or some of the ones on my sidebar. Me … I’m just getting started. But hopefully I can share a few interesting things as I explore the data.

I was fiddling around with the new gameday data … here’s what I came up with.

That is Peavy’s last start against the Mets. He struck out 11 in 6 innings that night, but also walked a season high 5 batters. The y-axis is velocity and the x-axis is the pitch number for that start. For example, the lonely data point in the botom left hand corner is an 83 mile per hour pitch … Peavy’s 14th of the game. The only real trend I can pick up is that he went to more off speed stuff as the game went on. In pitches 80-100, he only threw 5 balls over 90 mph.

Here’s his start before that against the Astros. Peavy was even more dominate this outing, striking out 11 and only walking just 2 in 7 innings of work. Again, it seems to me he goes away from the fastball a little bit late. Now, of course, this could just be pitching to situation or something (ie, as the game situation gets more crucial, he goes to the slider more often …). It may not be that he actually changes his pattern late, but it sure seems that way.

Finally, here are all of his pitches from his last two starts, ordered by velocity.

You can see that a lot of his pitches register at 92 mph+. In fact, 111 of his 212 tracked pitches were 92+.

From what I’ve read about the PITCHf/x system, the pitch speeds should be very accurate. I used the “start speed” which is the velocity of the pitch 50 feet from the plate. Of course, there could be some slight errors in what’s being recorded. I don’t think it’s perfect by any means (but it’s pretty damn accurate).

Anyhow, this is just the very tip of the iceberg with this kind of analysis. There is a lot of great work being done. If you’re interested in reading some great articles and learning more about PITCHf/x …

Anthony at Friar Watch has done a lot of stuff with it.

Dan Fox and Joe P. Sheehan have been doing some fascinating work.

Alan Nathan has some great info on his Physics of Baseball website.

Mike Fast is tracking it at Fast Balls.

I’m hoping to get to a point where I’ll be able to do some stuff like that. Uh … that might take a while. It’s fun to look into, though.