April 15, 2008
More Data, More Research
Mike Fast at The Hardball Times looks at how new tracking technologies might change the game of baseball. I must admit I haven't been keeping up with PITCHf/x as much as I should. My first thought is a probabilistic model of the strike zone. Given handedness of the batter and pitcher, the count, velocity, release point and coordinates of the pitch when crossing the plate boundary, (a line extending out from the front of the plate), what is the probability of:
- A strike call
- Contact on a swing
- Fair in play
- Type of ball in play
- A base hit
There are a lot more pitches than balls in play, so one should be able to develop a better model than PMR with the data for a season. Then you can see which pitchers and hitters perform above or below average on these types of pitches. This will keep us busy for years to come.
Posted by David Pinto at
10:33 AM
|
Statistics
|
TrackBack (0)
As a Computer Scientist, PF/X data is a gold mine. I have an SQL database holding the 2007 data. There were 190266 at bats, 139381 contact events, 163770 runner events, and 716994 pitches (323526 of these pitches have PF/X data). There's enough data that I can plot the path of each pitch in 3D or guess the pitch type.
The only pieces of information they are missing that I believe would be helpful would be spin rate and direction of the baseball.