Sunday, June 14, 2009
Chien-Ming Wang’s release pointsPosted by Mike Fast
There has been some recent discussion around the web about Chien-Ming Wang's release point this year, particularly as shown in the PITCHf/x data. Matthew Sisson's original post at Baseball Daily Digest was picked up by Rob Neyer in his ESPN blog and subsequently challenged by Harry Pavlidis at Beyond the Box Score.
Matthew noticed that Wang's pitch release points, as measured by the PITCHf/x camera system, had shifted by a foot and a half in his June 10 start as compared to a start he made last year on April 11. Harry pointed out that you can't take those numbers at face value and assume they represent an accurate measurement of a shift in Wang's release point. So what do we make of those numbers?
|Chien-Ming Wang delivers a pitch during his first start of 2009 against the Orioles. (Icon/SMI)|
First, we should explain what "release point" means in reference to the PITCHf/x data. The PITCHf/x system describes the trajectory of every major league pitch by its initial and final positions, initial and final velocities, and a constant acceleration. The initial values are measured at a point 50 feet from the back of home plate, and the final values are measured at the front of home plate. So what is often termed the "release point" in discussing PITCHf/x data for a pitcher is actually a point of the trajectory of the pitch 50 feet from the plate, not the actual point where the pitcher releases the ball. Sportvision chose this distance so that its initial pitch velocities would match closely with typical radar gun readings for pitch speeds. However, Harry and I and other PITCHf/x analysts often find it more instructive to reconstruct the early part of the pitch trajectory from the given data in order to estimate the pitcher's release point at 55 feet from home plate. Again, we don't know if this exactly where the pitcher is releasing the ball in its trajectory, but it's much closer to that point than the 50-foot measurement reported in the PITCHf/x data.
Next, as Harry points out, there are shifts in the camera calibrations and other factors that cause shifts in the PITCHf/x data from one park to another park or even within one park over time. Generally speaking, the data is of very high quality, but in some cases there are problems. This was definitely true in Yankee Stadium during the first half of the 2008 season, so any comparison to PITCHf/x data from that time frame needs to consider that issue carefully. Let's get to the data.
I've plotted release points for each of the right-handed Yankee pitchers who've been on the staff through all of the PITCHf/x era, which for Yankee Stadium began on August 30, 2007. I show only the data from Yankee home games, not from road games, so as to keep the picture simple. As I mentioned before, I've adjusted the "release point" data to indicate the point on the pitch trajectory 55 feet from home plate, near but not exactly at the time the pitcher released the ball. First, let's look at the vertical dimension. Each point on the graph represents the average release point for each game pitched by the pitcher.
Chien-Ming Wang releases the ball at about 6.5 feet above the playing field. His data points are indicated with the black triangles and dashed lines. You can see that he missed the last half of 2008 for injury. In 2009, his release point does seem to be several inches higher than it was in 2007 or 2008. However, the vertical release points for other Yankee pitchers also seem to be shifted up by a couple inches in 2009, at least for Chamberlain, Rivera, and Hughes; so perhaps this is simply a small PITCHf/x system shift rather than anything unique to Wang. But what about the horizontal dimension that Matthew mentioned in his article?
Here we see something more dramatic. The data is all over the place. What is going on? For starters, let me explain what we're looking at. The x-axis of the graph is shown from the pitcher's perspective, so points to the left of the graph are closer to the middle of the pitching rubber. Second, there's a huge shift of about a foot to the right evident in the data from the first half of 2008. You can see it for each of the pitchers. Unless they moved the pitching mound in Yankee Stadium for that period of time, that's likely just a systematic error in the PITCHf/x camera system at Yankee Stadium. If you want to compare a pitcher's release point from a Yankee Stadium start during that time, you need adjust for this error. Matthew referenced the April 11, 2008 start by Wang, and it did occur during this period, so a large part of the difference Matthew noted should be attributed to measurement error and has nothing to do with any mechanical changes Wang may have made. However, if we correct for that error what do we see? I took the same graph from above but shifted the data from early 2008 back to the left so that we can come closer to comparing apples to apples.
Now most of the other Yankee pitchers line up nicely with consistent horizontal release points. You can see that Brian Bruney's release has drifted several inches to the left (i.e., back toward the middle of the diamond) and Joba Chamberlain's release has drifted slightly to the right; and Jose Veras, Edwar Ramirez, and Mariano Rivera have maintained pretty consistent release points in this dimension. Now we can see that Chien-Ming Wang's release point has in fact shifted around by as much as a foot between April 2008 and June 2009. However, it's hard to say what's normal or right for Wang since the data we have for 2007 shows a release point in between those two extremes. I also looked at data from Fenway Park and Jacobs/Progressive Field, and it showed the same trends as the corrected Yankee Stadium data.
I don't know what the shifts in Wang's horizontal release point tell us about his effectiveness or his mechanics. It's worth noting that pitchers can change their horizontal release points by shifting their starting point on the pitching rubber. I don't know if Wang has done this or not. I intended this post to dig deeper and set the record straight as best I could about what Wang's release point has actually been and how one would best go about determining that accurately. There's obviously more investigation to be done, including video analysis, if one wanted to determine what's "wrong" with Chien-Ming Wang this year. It does appear to be safe to say, though, that's he not releasing the ball from the same location in his last few starts as he did during 2008, for whatever that is worth.
Mike Fast is a Royals fan who enjoys investigating baseball questions using data of many sorts. He is a member of Complete Game Consulting. He welcomes comments via e-mail.