UZR clarified

Dave · December 8, 2008 at 6:07 am · Filed Under Mariners 

Okay, so, Derek beat me to the announcement that we’ve started carrying UZR data over at FanGraphs. It’s pretty awesome, and hopefully, if you’re not already reading FanGraphs every day, you will start.

Leading up to the launch, I wrote a few articles about understanding defensive statistics and practically applying them. Not to be too self promoting, but I think you should read those. Defensive statistics cannot be viewed the same way as offensive statistics, and I try to make that clear in those posts.

For instance, in the comments of the last two sections, we’re seeing a lot of stuff that refers to Betancourt as “the worst defensive infielder” because his UZR/150 was -14.7 last year. However, you simply can’t make a claim like that. Here’s why:

1. UZR had Betancourt at -15 runs below an average shortstop in 2008. Shortstops are far and away the best defensive players in the game. His peer group is great defenders, which is just not true for 2B, 3B, or especially 1B. A -15 mark at SS is not the same thing as a -15 mark at another position.

2. Sample size – you cannot make concrete statements about a player’s true talent defensive level based on one season’s worth of data. Realistically, you need about 2,000 to 3,000 innings to eliminate enough of the variance to make a good judgment call. On it’s own, Betancourt’s -14.7 doesn’t mean an awful lot, and it certainly can’t be stated that he’s definitely the worst defensive SS in the game. He’s almost certainly below average, and he did have a poor defensive season last year, but don’t get carried away with valuing players based on a single season’s rating.

The fact that we have publicly available UZR data is a huge breakthrough, but please, use the data properly.


10 Responses to “UZR clarified”

  1. Mere Tantalisers on December 8th, 2008 6:23 am

    Perhaps this is more a question for someone over there, but what is the difference between BIS and STATS data sets? is BIS better or simply different?

  2. Dave on December 8th, 2008 6:56 am

    They’re both trying to do the same thing, but they use different staff. Different humans do different things, which is why the data is different.

  3. CMC_Stags on December 8th, 2008 7:35 am

    Here are some fun ones (I am only including years/positions where the player had at least 250 innings played). All data is UZR per 150 games for consistency.

    Yuni trend at SS (UZR/150):
    05: -0.4
    06: -0.5
    07: -4.3
    08: -14.7

    Jose Lopez at 2B:
    05: 3.3
    06: 3.5
    07: 1.0
    08: -3.3

    eat SS in 04 in 490 innings: 1.8!!! Eat that Yuni!! (actually, don’t eat that, lose 15 lbs instead)

    Jeremy Reed in CF:
    05: 14.9
    06: -9.5
    08: -3.5

    Ichiro! in RF:
    02: 8.4
    03: 10.8
    04: 8.5
    05: 2.3
    06: 12.0
    08: -0.2

    Ichiro! in CF:
    06: 0.6
    07: -1.8
    08: 3.3

    Beltre at 3B:
    02: 23.3
    03: 12.1
    04: 25.7
    05: 8.6
    06: 18.0
    07: -3.9
    08: 16.1

    LaHair at 1B:
    08: 13.3

    Russell Branyan career numbers (individual season samples are all very small):
    1B: 324 Innings, 4.4 UZR/150
    3B: 1831.1 Innings, -6.0
    LF: 720, 2.7
    RF: 357.1, -2.5

    I didn’t realize that he’d played so much LF when he came up with the Indians. He also played 328 innings in LF in 2000 and 2001 which are not included in the totals because there is no UZR information for those years. And his 1.6 UZR/150 in 364.1 innings at RF&LF for Tampa in ’06 is encouraging in that he could make an effective 5th/emergency OF for the team.

  4. CMC_Stags on December 8th, 2008 8:01 am

    Here is UZR/150 for some OFs that have been discussed as possible bargains/persons of interest for the M’s.

    Rocco Baldelli in CF:
    2003: -6.8 in 1322.2 innings
    2004: -8.2 in 1047
    2006: 10.0 in 749
    2007: 9.4 in 162 innings

    In 2008 he only played 36.2 innings in the corner OF positions so the data for that isn’t very relavent. The interesting thing for me is that the Rays didn’t trust his body to hold up in the field and used him almost exclusively as a DH. If the M’s get him, it seems like it should be as a DH / 4th OF. For reference, James has him at a wOBA of .345 in 09 and Marcel has him at .336.

    Ryan Langerhans in LF:
    05: 26.4 in 379.1 innnings
    06: 23.5 in 706
    07: 18.1 in 233.1
    08: 10.1 in 141.2
    Career: 21.5 in 1465.1

    Langerhans in CF (limited samples)
    05: 23.7 in 77.1 innings
    06: 4.3 in 116
    07: -12.5 in 274.2
    Career: -3.7 in 482.1

    Langerhans in RF (limited samples)
    05: 28.4 in 356 innings
    Career: 23.5 in 488.1

    Great corner OF who looks like he’d be a league average or so center fielder. For 2009, James has him at wOBA of .343 and Marcel is .310. So you’re getting about a league average hitter who can play league average center or way above average LF or RF.

    I am still interested in Gabe Kaper, but that interest has been tempered a bit. His fielding totals are about 14 UZR/150 lower for all positions compared to Langerhans and his wOBA projections are .331 for James and .337 for Marcel. He’s also 4.5 years older than Langerhans and I’d rather take the younger player of the two all else being equal contract wise.

  5. Mere Tantalisers on December 8th, 2008 8:07 am

    Another quick question regarding methodology – is the average (ie +/- 0) player a set hypothetical or merely the mean of that season’s players?

    That is, are all players compared against an ‘objective’ standard or graded on a curve of sorts, like the one implicit in the +/- system? Both ways would have an advantage.

    For example if the league becomes suddenly flush with superb fielders at some one position it would make sense to recognize that and include it in their value. A run saved, after all, is a run saved.

    At the same time if the quality of the average fielder at that position improves, then the relative value of great defense diminishes because it is less rare.

    Now I’m just thinking out loud, and that’s never a good thing. All I want to know really is whether a 0 UZR is determined empirically each season.

  6. CMC_Stags on December 8th, 2008 8:07 am

    To put some mis-conceptions to rest (as much as you can with a single defensive stat):

    Griffey in CF:
    02: -9.8
    03: -18.9
    04: -42.5
    05: -22.6
    06: -33.8
    08: -14.4

    in RF:
    07: -22.0
    08: -24.0

    So, if the M’s get Griffey, he should be let no where near the outfield, ever. He can’t be a 4th OF. He shouldn’t be your 5th OF (which Branyan can cover better).

    For reference, Raul put up -5.3, -23.9, and -11.0 UZR/150 numbers the last 3 years. So according to UZR, Raul would better than Griffey in the field.

  7. DMZ on December 8th, 2008 8:47 am

    As the author of the previous post, I’d just like to say “my fault” — that wasn’t meant as serious “he’s the worst-fielding infielder” analysis, but “hey, if you sort on x, Betancourt’s terrible!”

  8. Edgar For Pres on December 8th, 2008 10:29 am

    If Yuni is really -15 runs at SS then does he have any business playing defense anywhere? SS are good defenders but 2B aren’t slouches either. The positional adjustment for SS vs. 2B is less than 15 runs. Is our adjustment wrong or is Yuni just a horrible IF?

    (I understand switching positions isn’t as trivial as just assigning a run value to figure out the difficulty of a position)

  9. Broadcast James on December 8th, 2008 11:39 am

    I would assume that the more difficult and significant positions would return results much higher on the scale. It’s going to be pretty easy to get down below -10 at SS, while a 1B may need work really hard to have such a significant impact on runs allowed.

    One thing you might have noticed is Betancourt getting an even lower number than Ibanez, I think it’s pretty easy to imagine that a below average SS could cost you more runs than a really bad corner outfielder. This’d apply to a 2B/SS comparison on a smaller scale…

    Somebody stop me if I’m wrong. 🙂

  10. 3cardmonty on December 9th, 2008 1:19 am

    Orlando Hudson at -4.9 UZR/150 over the past 3 years?! Does that mean he’s no longer a USSM-approved acquisition?

Leave a Reply

You must be logged in to post a comment.