Friday, July 13, 2007

MLB First Half 2007 Defensive Ratings

One of the projects I've been working on this week has involved calculating plus/minus defensive stats for all ballplayers in the first half. I thought I'd share these data in case others are interested.


The stats are based on THT's ZR data. The conversion to a +- plays statistic is very straightforward--my methodology can be found here.

Here are the top three players at each position so far this season. Values are plays made above average.
1B
Albert Pujols +32.1
Casey Kotchman +12.0
Adrian Gonzalez +11.7

2B
Chase Utley +21.5
Mark Ellis +12.1
Brandon Phillips +11.4

3B
David Wright +32.1
Joe Crede +23.7
Pedro Feliz +19.6

SS
Troy Tulowitzki +24.3
Adam Everett +17.4
Jack Wilson +14.4

LF
Eric Byrnes +10.8
Norris Hopper +10.2
Matt Diaz +10.2

CF
Ichiro Suzuki +34.5
Andruw Jones +29.5
Curtis Granderson +18.7

RF
Shane Victorino +20.9
Austin Kearns +17.7
Andre Ethier +10.6
Nice to see a couple of Reds on that list. Hopper's been unreal in left field, though obviously he is best positioned in center field.

Also interesting to see Troy Tulowitzki overtaking Adam Everett at SS. Granted, Everett's hurt, but this Tulowitzski kid can pick it.

And Ichiro's transition to CF seems to have gone pretty well...

Update: There's been some good discussion of these data over at Baseball Think Factory (thanks to repoz for linking).

Update2: To my dismay, there were some problems with the spreadsheet for LF and RF. They have been corrected. The data I reported here was fine, but something got messed up when I uploaded the excel sheet to google (or something).

5 comments:

  1. Great Stuff. Thats really all I can say about it.

    ReplyDelete
  2. Very cool, J. And interesting to see what crummy fielders the starting LFs are (which makes sense). Any way to adjust the OOZ figures for the corner OFs to account for how "greedy" the CFs are?

    ReplyDelete
  3. Thanks guys!

    And interesting to see what crummy fielders the starting LFs are (which makes sense). Any way to adjust the OOZ figures for the corner OFs to account for how "greedy" the CFs are

    The low ratings for starting LF's came up in the BTF thread, and it's something I'll spend more time on in the coming days. I'm guessing that's real, but I'm not positive.

    As far as CF's...one thing that might be interesting would be to compare corner outfielders playing on teams with CF's with lots of putouts vs. corner outfielders playing on teams with CF's with fewer putouts. ... Or even better, compare within teams across years when CF's vary but corner outfielders differ. I'm guessing sample size would be an issue, but it'd be cool to investigate. :)
    -j

    ReplyDelete
  4. If you have faith in those fielding statistics then one could make the case that Norris Hopper is a more valuable player than Adam Dunn. To do this take the number of +plays NH would have made if he played as many inning as AD and then multiply it by the rate at which NH was making +plays. This would give you approximately 74 plus plays that Hopper would have made if he had played the same number of innings as AD. This seems too high so to err on the side of caution cut this back to +40. So Norris could be expected to make 50 more plays in left field if he played the same number of innings as AD. (+40 and -10). Is a hit taken away in the field as valuable as a hit garnered at-bat? IOW if a player saves a run in the field does that have the same net effect as if he created one while at-bat? I think you could make that case. So let's incorporate the plays in the field into a batter's offensive statistics. It is fair to assume that not all these plays would have been singles. Some were most likely balls in the gap that would result in extra bases. Let's make it a 70/30 ratio singles to doubles either saved or lost.
    Take Adam Dunn's offensive numbers and subtract 10 times on base and 13 total bases. Next normalize Norris Hoppers offensive numbers so that he has the same number of PAs as AD. To this add 40 times on base and 52 total bases. The result would be that Adam Dunn's OPS decreased from .903 to .841, while Norris Hopper's OPS increased from .635 to .874.
    Understandably there are some pretty huge assumptions made here. But I don't feel like they are way out in left field. =) Also, the number of +plays that would be made by NH were very conservatively estimated. Also, while a .900 OPS is in line with Adam's career average, I feel that if Norris was given the opportunity to play full-time he would OPS considerably more than .635.
    I am not completely sold on this line of reasoning. It is hard to accept the fact that Norris Hopper could be a more valuable player than Adam Dunn. But if these fielding metrics have much validity, a person might have to entertain that possibility.

    ReplyDelete
  5. texasdave,

    That's an interesting workup--thanks for posting it. :)

    I think Hopper's +40 plays ratin g might still be a bit liberal, but it's hard to know for sure, as we're basing it on a small sample size. I do think he's above average, and perhaps very above average, but even so...

    The other aspect of this is how much you weight offense and defense. I'm not sure it's always appropriate to simply sum their effect, as runs scored and runs allowed do not necessarily have a linear effect on wins (though admittedly I do this from time to time as well).

    Generally, my philosophy on building a team is to focus on defense up the middle, and let the corner spots be the primary source of offense (though obviously it's great if you get players who can do do both well).
    -j

    ReplyDelete