from Amazon
Thursday, May 31, 2007
PZR
PZR is one side of the defense coin, with UZR on the flip side. You can find discussions on it in the archives:
The description of PZR is in post #1 here:
http://www.tangotiger.net/archives/stud0191.shtml
Extenive discussion on PZR:
http://www.tangotiger.net/archives/stud0014.shtml
In a nutshell, PZR doesn’t care about whether a ball was caught or not. It only cares about the distribution of balls allowed. How many were hit to point x,y, how hard was the ball hit there, what’s the tendency of the pitcher/batter to allow GB, what’s the park, and what’s the handedness of the batters/pitchers. Given all that, then we don’t need to know if Mike Cameron or Wily Mo Pena is playing CF. Or in the case of Roger Clemens last year and this: Everett or Jeter.
If 95% of balls allowed by RHP to angle 27 degrees by a RHH who has a tendency to hit GB on a grass surface are outs, then *all* RHP who gives up said ball gets 0.95 outs. If Clemens gave up 100 such BIP, and Everett gets 100 outs and Jeter gets 90 outs, we don’t care (for Clemens). Clemens gets credit for 95 outs. Everett gets +5 and Jeter gets -5.
Add up all the “virtual” outs, virtual singles, virtual doubles, virtual HR, and you get a park and fielder-independent stat line for a pitcher.
***
This really won’t work for hitters, since there’s a quality to hitting a ball that is simply not captured by the data recorders. While Wade Boggs has some control over whether a ball is hit at 22 degrees or 24 degrees, a pitcher doesn’t have anywhere near the same control. A hitter can hit based on the fielder positioning far more than a pitcher can make a hitter hit the ball based on fielder positioning. That’s why HZR won’t really work.