Posnanski: Adam Jones the next defensive catfight

[DrungoHazewood]

Well, one of those metrics says he is a way better defensive CF on the road than at home. Is there some logical reason that would be true for any player? Obviously some parks are easier to hit or pitch in, but can a park be THAT much harder to field in?

(BTW, I agree with most of the comments and I "believe" the metrics, for the most part. Jones still plays a bit too shallow and doesn't always take great routes to the ball and occasionally gets turned around. I don't think he's an above average fielder. But when I see such a home/road differential in a stat it certainly gives me some pause as to the reliability of that stat, unless somehow the surface at Oriole Park is somehow harder to run on, like sand on the beach or something).

I agree that more work needs to be done to figure out the apparent home/road splits of some of the O's outfielders.

But why does this split not manifest itself with all Oriole outfielders? Corey Patterson had great UZR numbers in Baltimore. He had a year where he was +18 in center for the O's - that's great, that's a number that would lead the majors many years. Gary Matthews also had great UZR numbers for the O's, he didn't seem to be crushed by park effects. In fact, if you look at all regular outfielders for the O's for whom they have UZR data, the median is about 0.0. And that seems high to me - I'd assume that the average outfielder on a bad team with poor overall defensive effeciency numbers would be below zero, but it's actually average.

[DrungoHazewood]

Quick question I almost feel stupid for having to ask - I know Fangraphs has career UZR home/road splits. Where do you find year-to-year breakdowns?

[Frobby]

IMO Dewans +- is the best defensive metric system avaailble.

On what do you base your opinion that +/- is better than UZR, Rtot, etc.? Not debating you, I'd just like to know why you think so.

[crawjo]

Early in the season I felt like Jones's defense had improved this year, but the more I watch the more I'm becoming convinced that, overall, he's a pretty terrible defender. He does not get a good jump on the ball, especially on balls hit in front of him, and he takes bad routes to the ball as well. I wouldn't be surprised if he is the worst center fielder in baseball.

[CA-ORIOLE]

Quick question I almost feel stupid for having to ask - I know Fangraphs has career UZR home/road splits. Where do you find year-to-year breakdowns?

Never been able to find them. Total Zone (Rtot) has them. Perhaps they might useful enough to determine a split bias in cases with a lot of data. In Jone's case, they don't. They actually show him far more favorable at home than on the road.

[El Gordo]

Do your homework and answer the mail. You don't get to ignore my points and throw something else against the wall. It's Jeter versus Ripken at this point. It's a specific comparison utilizing the data that we have. Not to mention your premise above is flawed. Also, I did not say the stats were meaningless. So stop the strawmans.

Explain why Jeter's UZR away splits are so pronounced? Why is he a zero UZR on the road and minus 10 at home? Why is it unreasonable to assume he's closer to a zero UZR (based on away stats) instead of a minus 8 (rtot)?

Why should I not consider a superior metric like Jeters UZR (many in his latter years) in favor of his careeer rtot?

Why are Ripkens Rtot splits so pronounced in favor of his home park (more than 2 times in favor of his home park)? Based on that why can't I adjust his rtot from 10 to 7.3?

What reason should the park factors be ignored when Yankee Stadium was a notoriously fast infield and Memorial Stadium/OPACY are/were reputationally slower and the stats seam to indicate that?

Explain to me why my logic is flawed and why I can't say Jeter was potentially a minus 7 run differential between Ripken instead of an 18 run differential based on the above logic alone?

I'll give you a break on rationalizing all the flaws with rtot as compared to UZR.

After you answer the above, I'll touch on all the reasons why your analysis (including your latest) is flawed and answer your specific questions.

To begin with the basic point is that Cal was a much better defender than Jeter. Even accepting a 7 pont differential, that doesn't change the fact. Secondly you are using Jeter's UZR splits and comparing them to Cal's Rtot splits. It's a flawed argument, because the only way you can accurately compare them is Rtot because those are the only numbers available for Cal. There are general correlations between Rtot +- and UZR but there is not a 1 to 1 correspondence. To use Jeters URZ splits to reduce the difference in Rtot between them is specious. My general point is that when there is agreement between the three metrics, they indicate a clear trend,. In the case of AJ(to get back to the subject of this thread) the fact that they all show him to be a subpar CF convinces me that he is.

[El Gordo]

I like Dewan's because I have a better understanding of how they arrive at at their numbers and they give me more detailed information. In AJ's case I not only know his +- in terms of plays made above or below average, They give me info on his range, short, middle, and deep, and info on his arm in terms of kills and % of runners advancing. They also tell me the number of HR's he robs, and they translate his +- numbers into RS, for easier comparison to UZR and Rtot,

[CA-ORIOLE]

To begin with the basic point is that Cal was a much better defender than Jeter. Even accepting a 7 point differential, that doesn't change the fact.

I never stated Jeter was better than Cal. I stated Cal was probably better and implied not as significant as people think. Seven runs is a far cry from your 18 run differential isn't it? So assuming that Jeter is probably below average (range wise at least) that would make Cal only slightly above average wouldn't it?

I would also like to point out that potential 7 run differential was only a starting point. Realistically it's more than 7 based on those two assumptions but there was a lot more to consider. In the end I would not be surprised it to see a result in the 5-7 run range.

Secondly you are using Jeter's UZR splits and comparing them to Cal's Rtot splits. It's a flawed argument, because the only way you can accurately compare them is Rtot because those are the only numbers available for Cal. There are general correlations between Rtot +- and UZR but there is not a 1 to 1 correspondence. To use Jeter’s URZ splits to reduce the difference in Rtot between them is specious.

So let me get this straight. I'm comparing the same trait between player/sample A and player/sample B. I have one set of data that is the best we have and provides me very detailed information about sample A. In sample B I have a lot of data that is not nearly as accurate and is antiquated.

My logic to use sample A data and utilize/reconstruct the sample B data as best I can is specious and flawed? In your considered opinion, the accurate data must be ignored in favor of the inaccurate data. From this approach we can gain more accuracy? I know I'm benighted and specious, so correct me if interpreting you correctly.

As far as Cal’s rtot/rtz numbers go, I took them at face value (pretty generous assumption) and simply adjusted for the home/away splits.

My general point is that when there is agreement between the three metrics, they indicate a clear trend,. In the case of AJ(to get back to the subject of this thread) the fact that they all show him to be a subpar CF convinces me that he is

Rtot/Rtz agreeing with the other two could easily lead you down the wrong path and it has imo. You can’t seam to put aside the generalizations and get into the specific details and extenuating circumstances of this particular situation. There are multiple factors to look at. First of all, Rtot/Total Zone often lacks correlation with UZR / +/- and other range stats such as RF/RF9, particularly when looking at the range element/components. Total Zone is an expanded Zone Rating. Unlike RZR, it’s not even that good, because in historical cases it’s estimating how many hits might have been hit into a zone. Unlike RZR it does not assign a percentage, it rewards players runs for total number plays made in the zone above average over a season. In Ripken’s case you have a guy that played every inning of every game and he’s going to be unfairly rewarded for making a lot of plays simply because he had the most opportunities. Play complexity, easier field, a favorable pitching staff and Ripken’s lack of range are going to be mitigated in Ripken’ favor by Total Zone. If you want to get a better feel for Ripken’s range look at the RF9 (not RF) stats and compare it to league RF9 you’ll get a better opinion of Ripken’s range skills. The RF9 will calibrate range to 9 inning increments and offset Ripken’s biased high game/inning counts. Ripken had a very near to average RF9. From there you can look at field adjustments and pitching staff adjustments and figure that Ripken was realistically a below average ss (rangewise). Obviously his low error rate is going to help. Now take a look at Ozzie Smith’s RF9 differentials and note he was significantly above league RF9 (especially in his prime years). No, Mr Positioning was not in Ozzie Smith’s League.

With Jeter/Cal you’re probably looking at a .40 advantage for Cal on the RF9 differential as compared to league average up to age 35. Pretty significant but when you get into field adjustments, pitching staffs etc. the gap is probably narrowed quite a bit.

My general point is that when there is agreement between the three metrics, they indicate a clear trend,. In the case of AJ(to get back to the subject of this thread) the fact that they all show him to be a subpar CF convinces me that he is.

In Jones case it’s pretty confusing when you look at the details.

Lets isolate the range rates:

2008 UZR +1.9 , 2008 Rtz +10

2009 UZR -13.5, 2009 Rtz -3

2010 UZR -6.7, 2010 Rtz +2

We have an approximate 9 run differential on the range rates. I guess you could say there is some sort of correlation there, even if it’s not very good. But how do you explain that UZR has him as a -15.7 at home and -2.9 away while Rtz has him as a +5 Home and a -17 road. Whatever correlation is there seams to be coincidental.

Feel free to have the last word on this.

[DrungoHazewood]

Ill keep this simple. I dont look at UZR or WAR or even Rtot. I just know great when I see it. Paul Blair was great. Adam Jones is NO Paul Blair.

Roy, that works great when you're comparing Adam Jones to Paul Blair. Or maybe Mark Belanger to Freddy Bynum. But if you're comparing, say, Brian Roberts to Dustin Pedroia... well, good luck with that if you have no reliable numbers.

[El Gordo]

Roy, that works great when you're comparing Adam Jones to Paul Blair. Or maybe Mark Belanger to Freddy Bynum. But if you're comparing, say, Brian Roberts to Dustin Pedroia... well, good luck with that if you have no reliable numbers.

BRob and Ratso is easy. Ratso is on the field and BRob is on the DL. Now Cano and Ratso...

[Phantom]

So many misconceptions about UZR (not just here, everywhere). It's crazy.

As crazy as a post that makes this claim without giving any examples?

[Hallas]

Jones has gone from being ~-19 in UZR/150 to being -14.8. His overall UZR was worse than -10 before, now it sits at -8.7.

Based on what I know about fielding, it seems that fielding is hard to pin down because most MLB players are pretty close, and because a small number of missed plays leads to a large difference in runs saved/given away. This makes intuitive sense; if you remember Jones' inside-the-park HR, it was considered to be a catchable ball. A catch versus a HR is a swing if 1.6 runs, which is a huge number given that it's 1 play out of hundreds he'll make over the course of a season.

[CA-ORIOLE]

Jones has gone from being ~-19 in UZR/150 to being -14.8. His overall UZR was worse than -10 before, now it sits at -8.7.

Based on what I know about fielding, it seems that fielding is hard to pin down because most MLB players are pretty close, and because a small number of missed plays leads to a large difference in runs saved/given away. This makes intuitive sense; if you remember Jones' inside-the-park HR, it was considered to be a catchable ball. A catch versus a HR is a swing if 1.6 runs, which is a huge number given that it's 1 play out of hundreds he'll make over the course of a season.

Yep, 1 year UZR is approx 3 years of offensive stats...at least. Not worth getting too wrapped up about at this point. UZR won't have 1.6 run swings on a singular play. I believe the range is about .35 to .57 or so (.45 average). Wagging here from memory. Seams logical that outfielders will have more variation with heavier weighting than infielders. Fielding Bible/Plus-Minus has a homerun saved category so that could be a large swing, but I don't know how FB scores it (or if it's even scored).

[SrMeowMeow]

As crazy as a post that makes this claim without giving any examples?

Crazier!!