UZRs for 2007
DMZ · June 1, 2007 at 2:35 pm · Filed Under General baseball
By way of MGL (direct link to the spreadsheet). There’s some… we’ll call it counter-intuitive results in there.
By way of MGL (direct link to the spreadsheet). There’s some… we’ll call it counter-intuitive results in there.
Well, Beltre has committed a lot of errors this season, no? The rest seems to fall into line, although I had no idea Lopez was that good.
Oh, and there is a .zip link to a spreadsheet of all the results from 03-07.
http://www.tangotiger.net/mgl/UZR0307.zip
Beltre has saved the 8th most runs over the last 4 years.
Edit to say seventh. Gahhhh i r tarded
Lopez probably isn’t that good. UZR needs heavy regression to be a useful estimator of true talent, and usually takes about 3 years of data before there’s any kind of confidence, so a 3rd of a season probably isn’t indicative of true talent.
Well if you look at the 03-07 number it has him for 232 games averaged out to 8 runs/150.
At shortstop, it does have him at -13 runs per 150, but only in 43 games.
Alfonso Soriano??
I love that Prince Fielder is a lousy.
Does Adam Everett finally win a gold glove this year?
I would definitely focus on the 2003-07 data.
Questionable results:
Betancourt’s range factor of +1
Ichiro’s +5 range, but Sizemore is a +12? Really?
Tango, a huge “THANK YOU” for posting these. With all of the fishy fielding stats out there, it’s nice to see the most thought out one out there.
Interesting to see Utley up there, normally people rag on his fielding.
Robinson Cano as +2. Really?
These numbers aren’t necessarily telling of how good the player actually IS, but rather how good he’s been during the time frame.
How does a player changing positions within that time frame effect their range rating (i.e. Ichiro moving from right to center)?
There is a reason we cannot rely on defensive metrics as a reliable indicator of defensive value. UZR is no exception.
#13, yup, and not to mention the ammount of regression required to trust UZR or any ZR stat is huge. I agree w/ Tangotiger #9, use the whole data set not 07.
#14, there is usually some loss from moving to a new position, so either Ichiro is a natural or he’s losing some points from the move. I suspect it’s the former not the latter w/ Ichiro! He was meant for center.
Would the best way to evaluate defense be look at all the numbers and watch some games and put it all together? It seems like every defense ranking has some kind of defect.
Matt Holiday at +7? He looks slow at times out there and I have seen him take some bad routes and get bad jumps at times.
Ichiro’s played plenty of center, after all, just not in the U.S.
#17, the best would be to track the players and ball at all times, to get the speed, trajectory and angle of the ball, as well as positioning of the fielder.
19- That would be the best, can’t imagine how much work would go into that though.
Alfonso Soriano??
Robinson Cano as +2. Really?
Ichiro’s +5 range, but Sizemore is a +12? Really?
Reggie Sanders is hitting .367 so far this season. Kelly Shoppach is hitting .373. Paul Konerko is hitting .229.
This doesn’t make batting average a useless stat. It just means it’s early in the season.
I’m shocked Manny Ramirez is second from last in the list.
This may be way out there, but has anyone ever thought about using WPA for a defensive rating?
#23, you’d need to figure out how to properly assign fielding to WPA (vs dinging the pitcher). Also if you gave them a WPA increment for plays made, you’d still need to know which plays they didn’t make and assign those with negative WPA values. If you had that data, you could construct a neutral defensive stat w/o using WPA.
23. How would such a thing work? WPA is based on game situations and related to leverage index, neither of which have much to do with defensive ability.
#25, it could work if you could reliably divide responsibility for each play to the fielder and the pitcher. The problem is, on a play by play basis that’d be very hard to do accurately.
I realize that you would have to assign a certain amount of credit for each change in WPA to each defender involved in the play (and maybe even a coach!) – and figuring out how to assign that credit is kind of the Holy Grail anyway.
I was just thinking that when a fielder makes a great OOZ out, all the WPA still goes to the pitcher. It’d be cool to say “that catch was worth x wins.”
Splitting WPA between pitchers and fielders would be highly subjective, like scorring errors. What you probably could do is apply league averages on balls in play and split between pitchers and the fielders making the play on that basis. Or apply run values of batted ball types if you want to refine the split. Whether that’s all that worthwhile is a different question – if RBIs is not a good evaluative tool for context dependency, this applies just as much to WPA.
Wow, even having moved to RF, Griffey still sucks. Guess his legs are gone. Cincinnati must be purgatory for pitchers. It’s a hitters park and you have an OF of Dunn, Freel, Griffey, and Hamilton, each of whom is -6 so far.
Hmm, Freel and Hamilton are the two worst CF in baseball and Griffey has been the worst the past few years; is it possible there’s a ballpark or pitching staff effect that isn’t being accounted for?