As mentioned several times before, I’ve been playing with the Gameday data from MLB.com. I took a particular interest in Brandon McCarthy, possibly because he started quite well, slumped, then came good again. The variation in his starts could show some interesting information, perhaps reasoning why he slumped, what he was doing wrong, and what he did to correct it. The following is the initial analysis I have done on his data.
Through 5/20 McCarthy had 833 events that were logged in the Gameday system. 29 of these were pickoff attempts, leaving 804 pitches that were thrown. With the full Gameday system only having been installed in a few parks, fortunately one of which is the Ballpark in Arlington, I had 326 pitches in six games with the detailed stats that are becoming available. Of those, two starts had 19 pitches between them, which meant there were four starts which combined for 307 pitches. The charts that follow show all pitches, because even a little data is better than none, but there will be some explanatory notes with them.
This chart shows McCarthy’s pitch speeds on the various days he has pitched. The horizontal axis is the starting speed, and the vertical is the speed as it crosses the plate. As you can see, most of his pitches are in the 88-93 mph range, but he does go all the way down to 72 mph for his breaking balls. One thing that is clear to me from this is the bias in the different parks. His 4/9 (red) and 5/4 (orange) starts were at the Ballpark in Arlington, and they appear to show a significant increase in speed, of the order of a couple of miles per hour, over the 4/15 (green) start in Seattle and the 4/29 (blue) start in Toronto. The other two starts are buried in the noise, and don’t have enough data to really see much of anything.
This issue was discussed on a recent blog posting by John Beamer at the Hardball Times. He found effectively the same situation I did, where different parks show different biases for all of the stats in Gameday. In a future post I intend to follow up somehow, perhaps by comparing different teams in different parks. For example, as seen here, TBIA is a couple of miles faster than Seattle and Toronto. I will compare other Ranger pitchers at those parks, and also Seattle and Toronto pitchers, to see how they fare, and how it might be possible to actually get some numbers for the bias and use them to neutralize these stats.
Looking at McCarthy’s release points shows another interesting picture. Again, these studies have been done in other places, and it is worthwhile comparing to one of the studies on the Baseball Analysts website, by Joe P Sheehan. He shows a few different pitchers and their release points. The pitchers, John Lackey, Felix Hernandez and Jeff Weaver, show markedly different release points, with there appearing to be a much tighter zone for the better pitchers and a wider zone for Weaver. McCarthy compares quite favorably to Hernandez, which Sheehan describes as having a release zone of 11.3 by 7.1 inches. I haven’t gone that exact in my measurements, but looking at the large chunk of pitches, I would estimate it at about 13 inches wide by 9 inches high, just slightly larger than Hernandez. McCarthy is also throwing at a much higher angle, his pitches being from 0 to about -1 foot horizontally compare to Weaver and Hernandez starting at -2 and Lackey about -2.75.
You’re probably wondering about the pitches away from the central concentration. Although I can’t be sure, I believe the purple dots below the main group are an error, that is from one of the days where they only gathered about 10 pitches, so my guess is something was wrong. The group way out to the left is more interesting. They all came on one day, 5/4 in Arlington, and although a couple of them came early in the game, at one point in the 4th inning there was a gap of one pitch then every pitch after that came over in that section, and also a number of pitches were missed after that. I had vague recall that I had read somewhere that McCarthy had been told by the pitching coach to try throwing some sidearm, which would explain this, but I cannot find that reference and I’m guessing I imagined it. In reality, I suspect one of the cameras that measures these things was bumped, causing the pitches to be off for the rest of the game. At some point I will go back and look at that game and see if I can confirm it from the other pitchers that appeared.
Next up are the horizontal and vertical breaks. From this we see definite clumping of pitches, and articles have stated that you should be able to tell what pitch was thrown based on these numbers. I found another article online which talks about the pitches McCarthy throws, and it says he has four pitches: a 4-seam fastball (90-93 MPH), a 2-seam fastball (87-89MPH), a big breaking curveball(74-77MPH), and a changeup (77-81MPH). Looking at the chart, the two big clusters at the top are the fastballs, both the 4 and 2 seams, although there is not much to discern which is which. He has a clear cluster in the 71-73 mph zone, where the horizontal break (blue) and vertical break (red) reverse, this is presumably the curveball. The other group in the 75-78 mph range is probably then the changeup, although there is quite a bit of noise around it, especially in the 80-85 mph area.
The fourth chart may help clear it up a little though. With his speed along the bottom, we are comparing the break angle in blue to the break length in red. The break length clearly relates to the speed, the faster it goes the smaller the break length. Again we see the clusters shown in the previous graph, this time making the curveball pretty clear, being the area with the highest break length and the -10 break angle. A little further along we see the change, with both angle and length being about +10, then moving on up to the fastballs, again not discernable from each other. There is still a lot of noise in the break angle, and it is hard to say what it is. Are those pitches around 80 mph, with a break angle of -10, curveballs that didn’t curve? Perhaps they were thrown too hard to curve properly? And why are the numbers not the same as previously reported? The article mentioned above gave his fastballs from 87-93 mph, which I think we are seeing in the charts. But the curve is not 74-77, it is more like 72-74, and the change is not 77-81 but closer to 75-77. What does this all mean? Is there something wrong, or is he throwing differently? Or is it simply a problem with the Gameday data being measured differently in different parks?
Next up will be an analysis of what happened to each pitch. Is there a difference in result between a curveball and a fastball? Were more of one thrown for strikes? Was he highly successful in one area, that perhaps he can build on?
Actually, next up will be McCarthy’s start against the Red Sox on Friday, where we hope he can continue his recent success, and we hope we can get some additional Gameday data to add to the mix.