Joseph Pawlikowski at River Avenue Blues this morning has up the kind of post of which I heartily approve–the kind that says a highly abstracted stat doesn’t always tell the whole story, a refreshing admission from a Sabrematrician.
Pawlikowski’s post comes in response to the storythat Mark Teixeira’s UZR (that’s “ultimate zone rating,” a defensive stat for those who don’t know) was lower than average last year. As Joseph notes however, the stat doesn’t mean Teixeira’s a worse than average first baseman.
I agree with Joseph, Tex is an exceptional first baseman with quick reflexes, good food work, great hands, super judgement, and, most useful of all, a fantastic scoop that turns potential throwing errors into outs with robotic regularity.
So then what to make of his worse than average range UZR in two of the last three years? (I’m not going to take the time to explain UZR in full–it’s a complex abstraction that involves assigning various values to various offensive events, splitting the field into 70 something “zones” any number of which a given position player might be responsible for patrolling, then charting what balls hit into which zones a player was responsible for against how many run equivalents those hits produced.)
I’m on record as deploring highly abstracted stats. Ok, maybe not deploring them, but being skeptical of the over reliance on them–in this case the stat’s designer has created a method of assigning a value to a single, or a homer, or a stolen base and a method of assigning field zones to a given player. But these values and zones don’t necessarily reflect the full complexity of baseball situations (for example a single in the bottom of the 9th with a runner on third in a tie game has a higher value and would be something a team would seek to prevent more than a solo HR in a blow out; player defensive assignments are shared, etc). Measurements of any sort are useful if they really quantify what we can plainly observe, mystifying (and probably flat out wrong) when they contradict the observable. These stats also try to index one players performance against a league average instead of against a constant–introducing a still further degree of discretion in trying to quantify league average–a level of discretion, and a multple variable, which I think tends to make these stats less useful not more so.
But, if UZR is intended to measure range, it seems to be more accurate than less in this case. As great a first baseman as Tex is, the flaw in his defensive game is range–he’s not the most mobile of first baseman. He certainly doesn’t get around into foul territory, RF, the second base hole they way the likes of Keith Hernandez and Don Mattingly did.
But so what? Who wouldn’t take Tex at first? Perhaps its true that the fielding stats we use to measure infielders are less useful for first baseman–whose hands are more valuable than their range. I don’t know. What I do know is that ever more granular statistic abstractions at their best don’t do much more than tell us what we already know–Tex is one heck of a first baseman who doesn’t have the greatest range.
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_b.png?x-id=f1a6ceee-e6b0-4327-be35-31c1221407e8)
