Interesting. Since I'm guessing changes in movement like that are pretty hard to discern in person, it would certainly seem to behoove teams to track this data via PitchF/X, at least until they get a clear idea of a pitcher's trend. I'd think teams willing to follow Gameday for immediate access to the data would be a little better off too, as clearly pitchers vary quite a bit from start to start.

First off, I love all this pitch f/x stuff. How precisely does it track pitches, though? I've always assumed those pitch graphs on gamecast are not reliable and use them only for entertainment purposes. Does anyone have a handle on the technology of this? Is the tech. uniform from stadium to stadium?

I know it's calibrated before each use, so I'd think so.

From what I can tell playing with the data a little bit:
- the numbers are surprisingly good most of the time, but occasionally the system simply blows a pitch
- comparability from stadium to stadium and even start to start is its worst feature

Still, normalizing everything to initial pitches would seem mostly to solve the comparability issue.

Coors field (thin air) is supposed to affect movement by up to 33% (less drop on curve balls). I think 1% in speed change would mean more to effectiveness than 1% in movement change.

I am glad that you guys enjoyed the article. Here are a few tidbits to clear up some confusion.

-PITCHf/x does vary from park to park a great deal. Pitchers can gain or loose 2-3 MPH on their fastball for instance. I currently have a pretty decent set of corrections that make things relatively uniform though that I use. Every article that you see on THT from me is using those corrections. This article however is much less depending on those corrections because I am fitting everything to the 10 pitch baseline. I'd imagine that following Gameday anyone could do a pretty accurate estimate since speed and total movement are listed. That should give a rough idea on how a pitcher's fastball has changed.

-Most of the difference from start to start in the same park you see are real differences. Sportvision does do calibrations but those alter the cameras very little and correcting for atmospheric difference is much more important. If a pitcher threw in 40 degrees one start then 80 degrees in the next you will see that in his movement.

-On that note Coors absolutely does effect the movement of the ball but more like 20% not 33%. You can also see the difference at Chase because of the high temperatures besides the altitude.

-On that note Coors absolutely does effect the movement of the ball but more like 20% not 33%. You can also see the difference at Chase because of the high temperatures besides the altitude.

Out of curiosity, what type of effect for Coors and Chase? Faster/slower or vertical/horizontal changes. Or I suppose both...

I know it's calibrated before each use, so I'd think so.

Maybe with respect to movement from the ball, but I have no faith or belief in a system that cannot refute an umpire's call. In the MLB gamecast, I have never observed a pitch within the confines of the "strike box" labelled as a ball because an ump missed the call. I've never seen a pitch outside the box labeled as a called strike. Also, I don't believe I've ever seen a pitch situated along the border. They may be tangentially on the border, but never cut across the ball, ala a secant.

MLB's gamecast appears to sort pitches as binary events and then place them in an appropriate location to validate the umpire's call. Until that's fixed, I place no faith in any contributory element.

Am I crazy, or was there an article in Baseball Prospectus today or yesterday about this exact same topic? Weird.

MLB's gamecast appears to sort pitches as binary events and then place them in an appropriate location to validate the umpire's call. Until that's fixed, I place no faith in any contributory element.

http://www.hardballtimes.com/main/article/the-eye-of-the-umpire/

Well, Messr. Harris, I just went through the Red Sox/Mariner's game on Sunday, and counted about a dozen pitches which looked like they were missed by the umpire, half and half roughly between balls which were strikes, and strikes which were balls, and a number of which were right on the edge of the corner. It is possible of course that the Pitch f/x system isn't properly locating a certain % of pitches, but it definitely isn't relocating them after the fact to match the call.

http://www.baseballprospectus.com/article.php?articleid=7641

Similar but different. Still, amazing that they were just one day apart. And at least I'm not crazy!

I have never observed a pitch within the confines of the "strike box" labelled as a ball because an ump missed the call. I've never seen a pitch outside the box labeled as a called strike.

FWIW, I have seen this a lot -- particularly the bolded part.

Back in my day, we didn't even bother looking at pitch f/x data until pitch #140. Damn pitch counts.

Out of curiosity, what type of effect for Coors and Chase? Faster/slower or vertical/horizontal changes. Or I suppose both...

Well less air to imped the ball so the ball gets to the batters faster but with less movement. This is true for all pitches. Chase is about 10% and Coors is about 20% depending on exactly how warm it is. Clearly the answer to all of Colorado's problems is a pressurized dome.

I really don't know what you guys are talking about. Go to mlb.com and get the Gameday for today's Sox/Orioles game, and check out Okajima's first AB to Bynum. Looked like 2 straight curveballs 3" outside which were rung up for strikes.