Strasburg makes his first ML start tonight; live PitchFX data will be
here.

http://captnsblog.wordpress.com/2010...ed-performance

Over at the Hardball Times, Harry Pavlidis, who has been charting every pitch thrown in 2010, has provided an early benchmark for pitch types. Pavlidis also breaks out his data into a variety of different outcome-based and performance-based measurements. For example, according to his data, more batters swing and miss at the splitter than any other pitch, while the curve is the most likely common pitch to be taken for a called strike.

Because pitch selection is often predicated by the count, the performance-based metrics are not necessarily indicative of the relative effectiveness of each pitch, but they still offer an interesting look into how much damage is being done on each type. Pavlidis’ preliminary data indicates that the most damage (as measured by SLG% on balls in play) is done on the four seam fastball, while the least amount of pain is caused by the splitter.

There are several other interesting early inferences that can be drawn from the data, but the point of this post is to examine what pitches are currently being thrown to Yankee batters, and how each performs relative to the diet they are being fed. Again, without a breakdown based on counts, the value of pitched-based performance is mitigated somewhat. Still, the general picture provides some insight into what’s going on each time a Yankee batter steps into the box.

It's usually not that difficult to identify good pitches -- by location (often at the corner of the strike zone), by result (swinging strike being the most sought-after), and sometimes by velocity and movement. Identifying bad pitches by result, location and (lack of) movement isn't that hard either. I started a SoSH thread on Papelbon's awful slider to Wigginton on that topic after the Red Sox were swept by the Orioles.

The problem with looking at one pitch is that all pitches are thrown in context. A good changeup works because of its contrast with past and anticipated future fastballs -- changeup movement is important, but deception in the context of an expected fastball is probably more important. How can one quantify context using pitchfx? Dan used pitch leverage as one variable, and pitchers do show definite tendencies to use different pitches in different counts and situations. Are there other ways to compute context (ie, pitch count, men on base, runs leading or trailing, inning and number of outs, and sequence of pitches within an at-bat, and from one at-bat to the next) that would give a fuller representation of the effectiveness of a given pitch and pitch type?

If you have recently written (or have seen) an article using PitchFX or other Pitching Data on any other website, please write (or paste) a short blurb about the article (an "abstract", we'll call it) in our "Across the Web" forum and provide a link for us.

We'll promote salient articles to the main page (i.e., the page you get when you click "Home"), which will hopefully increase viewership and get some discussion of those articles going in the broader community.

The more information you provide in your first post about the article, the greater the likelihood of the article being promoted to the front page (because then it won't be just like promoting empty text).

Also, in case you hadn't noticed, this forum provides a way for individual members to create and maintain their own blogs. So, if you haven't had an outlet for your work before, you do now! Just click "Blogs" and "Create a New Post". If you write something you want to share with the rest of us, don't hesitate to also provide us with a link here (and an "abstract" as well).