THE BOOK--Playing The Percentages In Baseball

This may seem like a silly question and I have searched for answers before but never found one.

Does UZR or any other fielding metric account for reach?  Can it be assumed that if a first baseman has a longer reach he can not only field more bad throws, but also shorten the distance of a throw by a few inches with his reach?

UZR doesn’t “account” for anything specific like that.  It counts the number of balls that he FIELDS and compares it to the number of balls an average 1B would have fielded, given the same parameters as the batted balls.

IIRC, UZR doesn’t count infield line drives at all. Is that accurate?

If so, is it possible that Teixeira is above average at snaring liners, thereby impressing fans but evading quantification on those plays? I doubt that would account for the whole gap between the common perception of Teix as a solid defender and UZR’s lukewarm rating, but it might help reconcile some of the difference.

The current incarnation of UZR tracks LD for 2B and SS only, as when I correlated “line drive catches per line drive in the area” for each position (1B, 2B, SS, and 3B) only 2B and SS had a decent year-to-year “r”.

Here is one of the “problems” with any metric you are designing that includes lots of things you are tracking:

Since each of those things has their own signal to noise ratio, you are faced with a decision as to whether you want to include certain things have a low ratio.  That is the essence of FIP or DIPS.  Those metrics “decide” not to include a pitcher’s BABIP because it pollutes everything else due to the fact the signal to noise ratio is so low (and because it involves things largely out of the pitcher’s control, like defense).

So too with UZR. I don’t include infield pop-ups because it would just dilute the final numbers.  Anyway, that is why I don’t include line drives for 1B and 3B, but I do include them for SS and 2B.  I could have just as easily left them out for everyone.  It does not mean that there is NO skill in catching line drives for 1B and 3B.  It DOES mean, however, that the difference in skill among all 1B and all 3B is likely small.

I like Tango’s characterization that there are no “flaws” in UZR.  There are some things that could be improved, some of which I already know and some of which I don’t.  It is likely that any improvements would be marginal at this point.

I am adding a few of them for this season, and I will write another, comprehensive primer on Fangraphs detailing the additions as well as the original methodologies.

The discussion is good, BTW, between Mike and Tango.  Again, Mike is providing useful information for him and for his readers, through Tango, and in a back-handed way (rather than asking, “What about such-and-such?” he says something like, “Surely such-and-such is this (it sucks), isn’t it?” I suppose it doesn’t really make any difference as long as he engages someone like Tango, and as long as the end result is accurate information and education.

I have to give Mike some credit for allowing this, and for actually soliciting and “listening” to the answers. I have been pretty harsh on him. In some sense, he is like Murray Chass, and in another sense, he is willing to engage, listen and learn.