Thanks for the nice comments guys. I am glad you enjoyed the post. I found the info interesting and I figured you guys would too.
The data is from PITCHf/x and can be found at various places on the web. Fangraphs is the easiest to use and sort in convenient charts but their data is a little less reliable. There are some other sites that modify the data to make it more accurate. Brooks Baseball has a group that actually charts pitches to help insure the PITCHf/x data is correct. TexasLeaguers is another good site. Joe Lefkowitz has a site that gives you some sorting options that are not available anywhere else.
The PITCHf/x data is based purely on the movement of the pitches. If a guy throws a pitch that moves within a certain range it will be classified on that basis regardless of the grip or wrist action the pitcher uses. That can cause some pitches to be mis-categorized. A pitching coach would classify a pitch based on the hand grip and wrist motion (how the pitch is thrown), but PITCHf/x is only looking at what the pitch does after it leaves the pitcher's hand. To get around this you can go to Brooks Baseball for clarification. Change-ups in particular can be tricky because they have little movement and because there are so many different grips used by pitchers on change-ups.
Since RANDY IN INDY brought up the subject I looked it up, Homer Bailey does in fact throw a split-finger pitch that has very little movement and is slower than his fastball, so PITCHf/x labels it a change-up because it fits within their speed and movement criteria for the change-up class. However, Brooks Baseball classifies that pitch as a splitter because they have manually scouted some of his games and have double-checked his grips. Homer Bailey threw 164 splitters in 2011 and does not have a true change-up. I can't edit the original post in this thread anymore, but if I could I would move the change-up rate of 7.5% over to the splitter column.
Brooks Baseball has also reclassified some of Hector Santiago's pitches as screwballs. They show him as having thrown 40 screwballs in his career. I am guessing they don't break much because PITCHf/x is not picking up their reverse movement.
I am guessing there is some degree of blurriness to the data. For example, did Sean Marshall really throw .20% 2-seamers and .90% cutters? More likely those were just 4-Seamers with a little extra movement. How many of Marshall's sliders were actually curveballs with extra-sharp movement? I would imagine Marshall has several different grips and wrist actions for his excellent sliders/curves, each one slightly different from the others. Depending on the playing situation and conditions he may need to adjust his technique a little bit. Would the PITCHf/x system pick up the difference? No, I doubt it. But the tool is very useful and gives us a clearer picture of what makes each pitcher unique in baseball.
PITCHf/x is a pitch tracking system, created by Sportvision, and is installed in every MLB stadium since around 2006. This system tracks the velocity, movement, release point, spin, and pitch location for every pitch thrown in baseball, allowing pitches and pitchers to be analyzed and compared at a detailed level.