This is part one of what I expect to be a series of studies of the Rangers rotation, based on the Enhanced Gameday data provided by MLB.com. This data is proving to be a treasure trove for stats nuts, of which I am proud to say I am one, and I’ve been digging through it for a long time. I’m trying to get some of my thoughts about the data down before I get too lost, because every time I start looking at something I end up finding some other little piece of info and sidetracking all over the place.
In this study, I’m going to start at the start, which is usually the best place to begin. In this case, the start in the data is the release point of the ball. The Gameday data provides three fields, called x0, y0 and z0, which are able to define a three dimensional position where the ball is released by the pitcher. y0 is always 55, which means it is being measured at 55 feet from home plate, in other words right about where the pitcher lets go of the ball as he comes off the mound. x0 and z0 are the horizontal and vertical positions of the ball at that distance, with z0 being feet off the ground, and x0 being a distance from a central point, presumably somewhere centered on the pitching rubber, also in feet. Since it is measured as from the catcher’s view, an x0 that is negative is a release point that is to the left of the mound, in other words that has been thrown by a right-hand pitcher, and a positive x0 is to the right of the mound or a left-hand pitcher.
So, on to the pictures:
The first graph shows the five members of the Rangers rotation who have pitched the most games this season. All five are right-handers, so all have negative numbers in the x (horizontal) direction, and the graph shows they all throw in a range of about five to eight feet from the ground.
Before I continue, I should note that I have found some inconsistencies in the data, most noticeably between ballparks but also some within ballparks. The between ballparks issue appears to be an alignment problem, where all parks are not calibrated the same. Within ballparks the numbers match up fairly well for individual pitchers (meaning if they threw it at the six feet mark one day, they will also be about the six feet mark in every start, not some at five and some at seven). The exceptions within ballparks appear to be problems with the cameras taking the measurements, I have blogged before that it appears some cameras will be bumped during a game and lose their readings, or start measuring slightly off. An example of this appears to be in the Tejeda data, where he has most of his pitches around the 5.5 to 6.5 height, but you can see a small section of the light blue dots down and to the right, at between the 4.5 and 5.0 height. I have not investigated his starts to determine definitively if this is the case (I am leaving that to a later study where I want to show some of the problems in the data). Thus, all the values you see in the charts today are the raw data, not in any way adjusted. I should also point out that there are a very few outliers outside these graphs, for example in this graph there were a couple of points over at the +3 feet mark, which, unless a Ranger pitcher suddenly decided to throw a couple of pitches left-handed, are clearly erroneous and can be ignored.
What you see in this graph is five pitchers who are among the worst in the league this season. Having looked at some more of the release point data, I can immediately tell that these are pretty widely scattered release points for all these pitchers. I had a theory that the more experienced a pitcher, the tighter their release point, but that proved not to be the case. As it turns out, and you will see below, the more closely bunched together the release points are, the better the pitcher’s ERA.
Looking at the Rangers, you can see Brandon McCarthy up at the top right, very straight and almost vertical in his release. The spread of his pitches is about a foot in each direction. The other four pitchers are releasing from similar areas. Loe is also about a foot square, Millwood about 18 inches horizontally and a foot vertically, Padilla about 18 inches by 15 inches, and Tejeda is probably the best of the lot at about a foot by about 10 inches. All in all, fairly spread out, and given that their ERA+’s range from 58 (Millwood) to 77 (McCarthy), this could be a good indicator that this is where their problems begin.
This belongs to the Anaheim Angels. The scale is the same in both directions. Their two best pitchers are Lackey (ERA+ of 163) and Escobar (146). Weaver is league average at 102, while Santana (80) and Colon (74) are very Rangers-rotation like in their quality this year.
Now, what I said earlier about different ballparks appears to be in play here. Look at Escobar, in bright red the easiest to read. He has a large cluster of points vertically between 6.5 and 7, and a smaller cluster between about 6.2 and 6.5. By looking in the data, and sorting by the z0 value, I can tell you that every single pitch below 6.5 was thrown in Chicago on 4/29. There are a few pitches above 6.5 thrown on that date, but the highest was at 6.6, which is the bottom of the main cluster of pitches. It is obvious that the measurements in the White Sox ballpark are showing the pitches about 5 inches below everywhere else. For the Angels, Escobar, Santana and Weaver each pitched in that stadium during that series. Although not so easily detected, there does appear to be a slight shift down for the other two pitchers as well.
So, throw out that start, and I see Escobar as being about 9 inches wide by 6 inches tall. Lackey, the other outstanding pitcher for the Angels this year, is about 11 inches by 8 inches. Average Weaver is about a foot by 10 inches (allowing for a Chicago start). The two bottom guys are both about a foot by a foot. Using these numbers, and matching them to the Rangers, all the pitchers with a below average ERA+ are throwing in an area about a foot by a foot, or more, whereas the better pitchers are much smaller areas, half or less the size of the bad pitchers.
This is the graph of the Oakland A’s. Only four pitchers this time, because they have not had a consistent fifth starter like the other two teams. Note that one of these guys, Joe Kennedy, is a left-hander, and his release is way over to the right. Note also that the scales are slightly different, to allow for Kennedy.
I chose the Rangers because they are pitching bad (and because this blog is about the Rangers), the Angels because they are leading the division with decent pitching, and the A’s because they have the best pitching in the league. All four of these guys have a much better than average ERA+, with Blanton and Kennedy both around 120, Gaudin at 175, and Haren an incredible 270. If you weren’t entirely sure what ERA+ is, this tells you that Haren is about 2.7 times better than the average American League pitcher, and his 1.58 ERA reflects that. He is having an incredible season so far.
But Haren upsets the applecart here. Look at his release points, in the green. They are about 15 inches wide and 18 inches high, much wider than even the bad pitchers on the other teams. What’s going on? Without investigating deeply, it does appear that most of the starts down and to the left were made in Oakland, with the up and to the right starts being around the league. I would suspect this to be the problem except that it does not manifest itself for either Kennedy or Gaudin, who are showing smaller ball-shaped release points. This is probably something to look into further as well. If you actually take Haren, rotate him so he is horizontal, and remove the Oakland bias, he is probably about a foot by 6 inches, which would put him back in the top of the bunch.
Gaudin is very good this year, and has a small release area to show for it. Blanton and Kennedy are similar in ERA+, although Kennedy shows a much smaller area of release. Either way, each of these pitchers are showing they have control of their release points more than the other teams do.
So, in summary, it appears that the better the control of your release point, the better you will pitch. I’m sure this is just one factor, but it is probably one that is more coachable than some things that are out of their control. I surmise that if a pitcher gets in a good rhythm and all his mechanics are working well, the end result will be better, and if he is flailing about, not controlling his body, the pitch is going to go anywhere other than to a well-controlled point, and that will lead to better pitches to hit. A good pitching coach ought to be working on these things with the players, making them aware of how open they are getting when they release, and how variable they are. Age (and thus experience) do not appear to be factors, as none of the A’s are over 28, the Angels are split with a good and bad pitcher under 28 and a good and bad pitcher over 28 (and their best pitcher exactly 28), while the Rangers range from 23 to 32 with nothing resembling any kind of quality pattern.
Whether the Rangers have the personnel or ability to deal with this problem is unknown. Once more of this kind of data is available, we may be able to see patterns emerge showing that players can improve, or decline, in this ability. Eventually we may even be able to use this to evaluate pitching coaches. For now though, we can just look at what we have and see the beginnings of some interesting analysis to come.
[Update 6/14 at 7pm: I inadvertently switched the names of McCarthy and Loe when I pulled the dataset for this study. Although the results are still the same, I have corrected the Rangers chart and changed the names of Loe and McCarthy in the text where appropriate. I am going to take steps to correct this problem. My thanks to John Walsh of The Hardball Times for helping me discover this error through a related email I had sent him.]