April 22, 2008
I posted an analysis of Jose Valverde’s early-season troubles on THT Live yesterday.
Also, Dan Brooks (Jnai) has been doing some very good work with PITCHf/x over at the Sons of Sam Horn discussion board. His PITCHf/x wiki is well worth checking out.
As part of the discussion of Dan’s work at the Book blog, I made a chart of pitch speed vs. spin deflection angle for a typical right-handed pitcher.
April 23, 2008 at 4:45 pm
Mike,
Just a curious question…I don’t know if you have done this or not. In one of his analyses, Alan Nathan posits that it would be “more correct” to subtract off an estimated drag from ax and az, and re-calculate pfx_x and pfx_z (and spin rate and axis) from that rather than the current model. It’s my belief that doing so will fix a few things that bug me a bit…like some cutters that break inside a small amount. I think it will also give tighter clusters, but I don’t know for sure yet.
Anyway, you might want to give it a try. The reason I say this is because the current pfx_x,z values are somewhat dependent on where the ball is thrown…take the example just the x component of a cutter thrown from a righty in on a lefty. It has a magnus force pointing somewhat in to the lefty, and a drag force away from the batter, and the two nearly cancel. But if the same pitch is thrown away from a lefty, the x component of the drag is lessened, and we would infer from just the pfx values that the pitch broke more…which did not actually happen.
April 23, 2008 at 5:06 pm
Ike,
I am using formulas based on Alan Nathan’s new paper that take into account drag in x and z. I haven’t seen any dramatic improvements in clustering, however.
I don’t see why Sportvision/MLBAM don’t use those adjustments to pfx_x and pfx_z. They are not very hard to make.
I am very curious to see what you’re finding about correcting the data for camera distortion or whatever the problem may be. There are some weird things going on this season in the data from Cincinnati and Detroit in particular. Do I have to wait until May 10 to hear what you’ve been learning?
April 23, 2008 at 7:49 pm
Sadly, what I’ve been learning lately is not a lot. I have a number of directions I want to try to go with this, and not a whole lot of time to dedicate to it. It looks like I’m going to have some amount of time in May before the summit to work on it, so, unfortunately, yeah, it may not be before May 10. If I find something sooner though, I’ll let you know.