Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Dialed In > Discussion
Dialed In
— 

Wednesday, November 02, 2005

D - ## !  D- ## !

I love the defense. As long as I have been stumbling around the internets, I have been arguing long and loud about how much defense was undervalued by Statheads. However, when I started I was arguing some things I would never say now, but I was at one point just as ignorant as Bill Plaschke.

Looking back, I was pretty amazed at the Re-Education of Chris Dial from November of 1997 to February of 1998. In June of 1998 , I devised some methodology to evaluate both sides of the ball using Zone Rating data, or about the same thing as SuperLwts/UZR.

I went through a few permutations of offense and defense, and began using Jim Furtado’s Extrapolated runs, and did a ton of work with Dale Stephenson (Google that name for the work of a top-notch sabermatrician) in determining runs for defense. Dan, we still need to get Dale to post his Peak Lists here.

Last year, I wrote a couple of “Who should be MVP?” articles. Right now the methodology link is dead, but it is pretty close to the above rsb post.

I wanted to improve what I had done. So I did some data-mining. I went back to the source – STATS - to better determine what ZR chances occurred at each position, and worked backwards from there to calculate defensive runs.

In doing so, I was able to come up with good averages for balls in play based on 3000-7000 defensive games (up to 60000 innings and 7000-22000 chances depending upon the position), and effectively draw a baseline of where a fielder’s production will lie based on his zone rating. Converting to runs is simple enough, once you figure out chances.

So I have done all this, worked on my calculations, and generated defensive runs. These runs saved (RS) are above average and specific to the individual’s playing time.

Yes, the first critique is: I have to do one of two things: normalize everyone to the same number of chances (the average) and indicate that the rate would result in so many more (or less) outs and runs. Or I use the average out conversion rate and subtract actual outs from average player outs. I’d certainly prefer to do it the second way, but I can’t. So I do it the first way.

I have some seasons where I can do it the second way. Working through the math, the difference between defensive plays at shortstop (the position with the most plays on average) converted to outs is plus or minus four plays. That’s three runs. So the first way is going to be within three runs of the second way from best to worst in 525 chances.

I took all the positions, made a nice spreadsheet, and have it ready to calculate runs saved above (or below) average.

The basic calculation is:
Player’s ZR times the average number of chances at the position over a full season times the average run value of balls hit to that position.

This yields a Player’s RS(cal), where (cal) is every inning of every game. We then subtract the league average RS(cal), and adjust those runs by the player’s actual playing time.

That yields Runs saved compared to what a player converting outs at X rate given an equal number of chances normalized to the playing time.

I know – how much does the normalization affect the data – as noted above, not very much.

With these formulas, you can generate good defensive value numbers on your own, anytime you need to. The results are going to be robust too.

What about UZR?

MGL has said that UZR *has* to be better than ZR because it is an extension of ZR. Well, it isn’t exactly. MGL converts all of the data from STATS zones to a different grid (Project Scoresheet) that is far less discriminating.

Envision this:
There are three zones (V, W, X) for a first baseman in ZR, covering about 24 feet. That is from the baseline to about 24-30 feet (depending on a fielder’s depth). These zones are assigned as “first base ones” because first basemen turn groundballs hit into those three zones into out more than half of the time. The next zone over, U, does not get half of the balls turned into outs, and thus it isn’t included as a first base zone. There are approximately 280 groundballs hit into those three zones. However, when you overlap these three zones with Project Scoresheet, only the first two zones, W, X, are the first baseman’s zone (Zone 3). The third zone, V, is combined with another two zones, T and U, creating a “34” zone and then UZR tries to average out the chances a first baseman should get from that. In UZR, unfielded groundballs in the T, U zones count the same against a 1B as those in the V zone (ZR zones T, U, V make up the Project Scoresheet (PS) 34 Zone) whereas they don’t count at all in ZR. That’s important. A ball hit 30 feet to the first baseman’s right is his responsibility in UZR. It is not in ZR.

STATS Zone Rating Grid

This doesn’t turn out too terrible because Zone T is the responsibility of second basemen in ZR. This is why UZR misses 2B the most (I think). Half of the groundballs hit in that portion of the PS 34 zone are in ZR by default at full credit, whereas in UZR they are not.

And so it goes around the diamond. In addition, the PS 4M zone (up the middle to the 2B side of second) is split in ZR. Half of it is the responsibility of the 2B, but the other half is not.

The 3B/SS side of the field sees the same problems. As I pointed out in Mike Emeigh’s great eight part series on Jeter’s defense, there was a zone assignment error from Project Scoresheet that rendered one season of Jeter’s data unusable. It’s a big issue.

Project Scoresheet Grid

In a nutshell, UZR is a very nice system. However, I believe the proprietary nature of the STATS’ raw data compels the data user to “tweak” it, and in my opinion the use of Project Scoresheet zones makes it less accurate than generating runs saved from ZR, as I have done.

Chris Dial Posted: November 02, 2005 at 04:45 AM | 111 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

Page 1 of 2 pages  1 2 > 
   1. Chris Dial Posted: November 02, 2005 at 09:16 PM (#1716484)
My data and Gold Glove ratings comparison is just about wrapped up.

Expect that in a day or so.
   2. TDF, situational idiot Posted: November 02, 2005 at 09:24 PM (#1716506)
I don't have one of those graduate-looking guys next to my name, so I might not be smart enough to comment. But...it struck me reading this that ZR may, in fact, do a very poor job of rating fielders' abilities.

Here's my thought. Let's say we have a 1B named "Al". Now Al plays for a team whose pitchers do a great job of jamming/pitching away from hitters. Such a good job, in fact, that the vast majority of his fielding chances are right at him (zones W & X, or zone 3 in the PS grid). In essence, every play is easy.

Now another team has pitchers that put 'em right down the middle. Let's call their 1B "Sean". Nothing is hit down the line; in fact, Sean could relocate in zone V (or 34), if he were fast enough to get to 1B to recieve a throw from other infielders. To make any plays, he has to make spectacular ones.

Sean may in fact be a better fielder than Al. He may be much better. But we'll never see that in ZR (or your improved ZR) because his chances are, on average, much tougher plays than Al's.

Or do we know (or assume) that fielding chances are evenly distributed among the zones?
   3. TDF, situational idiot Posted: November 02, 2005 at 09:26 PM (#1716514)
I missed a point:

Since Sean is trying to make tough plays all the time, he'll "miss" more balls in his zone, and actually look worse than Al in ZR and UZR (and your system).
   4. Chris Dial Posted: November 02, 2005 at 09:31 PM (#1716520)
We know that the zones of assignment are zones that in MLB, greater than 50% of balls hit to that zone are turned into outs.

We know that ball sit to Zone V are turned into outs by first basemen *most* of the time.

And as Mike Emeigh pointed out elsewhere, fielders will play where teh team believes the BIP distribution to be.

In adition, the 1B tends to stand in the middle of the zone. It's the same distance to the line as to the other side of Zone V (give or take a player's personal preference).

But yes, chances tend to even out over the course of a season.
   5. Chris Dial Posted: November 02, 2005 at 09:32 PM (#1716521)
Since Sean is trying to make tough plays all the time, he'll "miss" more balls in his zone, and actually look worse than Al in ZR and UZR (and your system).

That is a nice theoretical, but do you ever see a 1B positioned like that?

I mean *could he stand over beside the 2B* (if he were fast enough)? Sure, but no one does...
   6. AROM Posted: November 02, 2005 at 09:56 PM (#1716555)
Do you know what percentage of balls in play are counted in somebody's zone? What percent are not charged against any fielder?
   7. TDF, situational idiot Posted: November 02, 2005 at 10:01 PM (#1716574)
We know that the zones of assignment are zones that in MLB, greater than 50% of balls hit to that zone are turned into outs.

We know that ball sit to Zone V are turned into outs by first basemen *most* of the time.

And as Mike Emeigh pointed out elsewhere, fielders will play where teh team believes the BIP distribution to be.

In adition, the 1B tends to stand in the middle of the zone. It's the same distance to the line as to the other side of Zone V (give or take a player's personal preference)...

That is a nice theoretical, but do you ever see a 1B positioned like that?

I mean *could he stand over beside the 2B* (if he were fast enough)? Sure, but no one does...


The bottom line to what I'm saying is this: "Sean" needs to play further away from 1B to field the balls hit in his direction, because most are hit further from the base. He really can't, as you point out, because he could never field a throw from another infielder. So he has to be closer to base, which makes balls hit in his direction more difficult plays. Similar arguements could be made for all infielders, but not as much for outfielders.

But yes, chances tend to even out over the course of a season.

"tends to even out" shows me that there may be some truth to what I'm thinking.
   8. Chris Dial Posted: November 02, 2005 at 10:04 PM (#1716578)
"tends to even out" shows me that there may be some truth to what I'm thinking.

That there is "some truth" doesn't support "ZR may, in fact, do a very poor job of rating fielders' abilities."

No, it won't. Because it just doesn't happen that way.

So he has to be closer to base, which makes balls hit in his direction more difficult plays. Similar arguements could be made for all infielders, but not as much for outfielders.

But those balls won't count against his ZR, but they will count against his UZR.
   9. Chris Dial Posted: November 02, 2005 at 10:06 PM (#1716583)
Do you know what percentage of balls in play are counted in somebody's zone? What percent are not charged against any fielder?

I have some idea. I tried to do this a few years ago. I think teh answer is somehting like 400 BIP aren't in anyone's zone (or is a ball that a play isn't made on).

I'll have to find that spreadsheet.
   10. AROM Posted: November 02, 2005 at 10:14 PM (#1716600)
My wish list of stuff from zone rating:

Actual plays/opps for all players. Then I wouldn't need to estimate, and I'd know how many BIP aren't in a zone.

That way, if I see a team with a horrible DER but players with above average ZR, the first thing I'd want to know is if they have an above average # of balls not counted in a zone.

How many balls hit to each subset of the zone. That would answer the question of how well this evens out.

Stats did some of this on case by case basis when they published the Scoreboard. They did some great analysis on Griffey one year after he'd won a bunch of gold gloves despite subpar zone ratings.

I miss the Scoreboard.
   11. Chris Dial Posted: November 02, 2005 at 10:19 PM (#1716610)
Rally,
I know. Fo rme, I think I just need Plays made in zone. Plays made out of zone.

A zone-by-zone breakdown would be better, but I am not getting greedy. Even at that. I still think this approach is hte best one.

I have actual plays/opps for a bunch of players from 1998-2000 (sort of). It mixes out-of-zone plays with in-zone plays, but it's better than distorting the zones compeltely.
   12. Spivey Posted: November 02, 2005 at 10:35 PM (#1716642)
I wonder...if a team has a strong LHP staff, wouldn't that lead to more hard hit balls to the right side of the infield (not just more, but a higher percentage) on account of their being more pulled balls over there? Wouldn't it be the same with RHP staffs and the left side of the infield? Is this accounted for in ZR?

Note: I'm assuming pulled balls tend to be hit harder on average (particularly vs. opposite field grounders as opposed to in the air) than opposite field balls. This might not be correct.
   13. Chris Dial Posted: November 02, 2005 at 10:42 PM (#1716663)
I'm assuming pulled balls tend to be hit harder on average (particularly vs. opposite field grounders as opposed to in the air) than opposite field balls. This might not be correct.

There may be a small bias. MGL did some work that indicated that GBs to the right side from a LHB were converted at a lower rate. But he did that off PS Zones, so who knows.

I have in the past looked to see if there were any team linkages as what you are talking about, and when I post *everyone's* rankings, you can double check, but there aren't any IF biases that are readily detectable.

Mostly I investigated whether or not the Prime Braves got a proliferation of easy chances when Maddux and Glavine were at their peak, but the answer tended to be "No."

And we *know* Chipper gets fewer balls. We also know from Mike Emeigh's Jeter study that LHP staffs and GB staffs *don't* distribute them in any particular manner.

Do you know of a team with a strong LHP staff? Does their 3B have a poor ZR? Their SS? Pick an example, and we'll look, rather than hypotheticals.
   14. Spivey Posted: November 02, 2005 at 11:07 PM (#1716707)
What years do you have ZR data for? I have some free time and will look through the last 10 years or so, but I need to know if we can put it up against data.
   15. Chris Dial Posted: November 02, 2005 at 11:21 PM (#1716736)
Back to 1998.
   16. Spivey Posted: November 02, 2005 at 11:57 PM (#1716837)
Below I listed teams that skewed towards having a lot of their innings pitched by LHP. The percentage is the percentage of IP thrown by LHP.

2005- Pittsburgh (45.5%), Tampa (41.2%), Detroit (38.2%), Baltimore (38.1%),
2004- LAD (37%), NYM (36.7%), ARZ (38.8%), BAL (37.2%), DET (39.2%), KC (54.4%), SEA (38.8%)

Also, 2004 Boston is incredibly righthanded. Only 9.3% LHP. NYY in 2004 was only 8.8% LHP. I'll start looking at seasons prior to 2004 in a bit. I might have missed a team or two in here, but overall I think it looks pretty good.
   17. Chris Dial Posted: November 03, 2005 at 12:39 AM (#1716929)
What is the league averages in those years?
   18. Spivey Posted: November 03, 2005 at 12:57 AM (#1716956)
Highs:
2004- OAK (52.1%)
2003- FLA (38.5%), NYM (40.8%), KC (38.3%), OAK (53.2%), CLE (48.1%)
2002- ATL (38.2%), LAD (42.3%), NYY (38.4%), OAK (39.4%)
2001- COL (44.8%), TEX (40.6%), OAK (37.7%), NYY (38%)
2000- NYM (54.9%), ARZ (46.5%), TEX (41.9%), MIN (43.3%)

Lows:
2003- PIT (10.3%), MON (8.9%)
2002- CIN (8.2%), BAL (10.6%)
2001- SD (8.3%), LAD (5.3%), MIL (4.7%), BOS (7.7%), KC (7.6%)
2000- HOU (5%), KC (3.8%), TB (8%)



One thing to note is there were a ton of teams (I'd say at least a dozen, maybe more) who didn't have a LHSP in 2001. I'm not sure what to make of that. It shouldn't affect ZR but it could make some subconscious changes to the way people view plays if there is a bias with LHP/RHP. I also added Oakland for 2004. I forgot Redman was on that team. I'm going to do 1998 and 1999 later tonight, after I watch some TV.
   19. Spivey Posted: November 03, 2005 at 01:03 AM (#1716966)
I don't know. I'm just looking team by team to see if they're above or below what seems to be average. I'd say average is probably around 350-400 IP for LHP which would range from 24-28%.

The average appears to be going up, though. There seemed to be a lot more teams with no lefty starters in 2000 and 2001 than recently. In those 2 years, I'd wager it was probably more like 300-350 average (20-24%).
   20. Chris Dial Posted: November 03, 2005 at 01:43 AM (#1717011)
Okay, Spivey, so out of 400 chances at 3B, what do you suppose the distribution is from LHB/RHB? I'd suspect 90/10 or 85/15.

Since RHPs face 50% RHBs, I don't think LHPs generate these hard hit balls. In addition, are there that many platoons in MLB? Doesn't each 1B basically face the same set of hitters per league?
   21. Chris Dial Posted: November 03, 2005 at 01:44 AM (#1717015)
Don't get me wrong, I think you have a shot at something here, but I have never noticed a bias (or at least a significant one) for quality of pitching staff and the infield.
   22. Spivey Posted: November 03, 2005 at 02:06 AM (#1717040)
Okay, Spivey, so out of 400 chances at 3B, what do you suppose the distribution is from LHB/RHB? I'd suspect 90/10 or 85/15.

I presume you mean 10/90 or 15/85, right?

Since RHPs face 50% RHBs, I don't think LHPs generate these hard hit balls. In addition, are there that many platoons in MLB? Doesn't each 1B basically face the same set of hitters per league?

Don't get me wrong, I think you have a shot at something here, but I have never noticed a bias (or at least a significant one) for quality of pitching staff and the infield.

Since I agree it probably wouldn't be that significant, you'd need a lot of data. Since you only have 7 years (and I have 0 years data), that might not be enough. There were a couple cases it seemed where the same team went from very left-handed to very little lefthanded over a short amount of time. These would be the best to look at to see if there's any sharp change in the ZR of the fielders. Since the sample size is so small, it wouldn't prove anything but only show that there *might* be something worth investigating further.

Can I ask you for certain parts of the ZR data when I take a further look into this (later tonight, tomorrow maybe) if I think something could show up in the data?

I am assuming at this point that a RHB will tend to spray the ball more against a RHP than a LHP. If there are a fair number of platoons against the LHP, the left side of the infield might see not only harder hit balls but also *more balls* which I believe MWE said has some (slight) correlation to a higher ZR. These might cancel out making it hard to gather anything.
   23. Rickey! On a blog from 1998. With the candlestick. Posted: November 03, 2005 at 02:11 AM (#1717047)
I'm only posting because I get one of the Scholar icons.
   24. Chris Dial Posted: November 03, 2005 at 02:16 AM (#1717052)
MWE has said that, but I'm not sure I agree. More chances increases the smoothing of the data, but in general, the players with the most chances have lower ZRs than players with fewer chances.

That is: the players that get 900 innings average to have a lower ZR than the league's average at the position. Why? Because of defensive subs some, but those guys ZRs wouldn't go up with more chances - they'd go down (more likely) and approach the average.

I have chances - but not for tons of players.
   25. Chris Dial Posted: November 03, 2005 at 02:23 AM (#1717060)
I just checked that assertion:
Pos R
30.26
4-0.42
50.17
60.51
70.17
80.19
9-0.38


You can do the squaring on that. Now most of Mike's research, I think, is on the SS, where the largest correlation exists, but I would say that it's a definite *no* that more chances means higher ZR.
   26. chris p Posted: November 03, 2005 at 02:24 AM (#1717064)
one thing i think could add alot to the zone inormation is a measurement of how hard the ball is hit. this could probably done by timing how long hte ball takes to get back to the ground ( for fly balls) or how long it takes to get to the fielder ( for groudn balls) ... are there any systems that do anything like this? do you think this would add too much extra noise to the data?
   27. Chris Dial Posted: November 03, 2005 at 02:25 AM (#1717065)
WTF? I hate pre tags

3 : 0.26
4 :-0.42
5 : 0.17
6 : 0.51
7 : 0.17
8 : 0.19
9 :-0.38

Negative for second base and RF. How odd. This is off three year data. Maybe in single season data there is something to that. I'll have to go back and check.
   28. Chris Dial Posted: November 03, 2005 at 02:27 AM (#1717072)
are there any systems that do anything like this? do you think this would add too much extra noise to the data?

STATS tracks this, and I have scored games this way. I don't think it is a factor becaus eMOST balls are "medium". Think of how few balls are "smoked" that stay in the park, or are hit within a fielder's zone.

Not many.
   29. chris p Posted: November 03, 2005 at 02:38 AM (#1717089)
but isn't there a big difference between a lazy fly and a popup? or are there not enough popups to make this an issue?
   30. DSG Posted: November 03, 2005 at 02:59 AM (#1717116)
The correct split is actually 80/20 (or 20/80, whatever).

But Chris, I think you're trumping ZR a bit too much. There are 2 basic problems with ZR:

1. It lumps together all balls in play, not making an allowance for how hard they were to turn into an out. So a hard hit GB into zone V counts the same a soft grounder to zone X. UZR adjusts for this.

2. It doesn't count balls of zone, which means that BIP distribution plays an extremely important role in ZR. If for whatever reason, a first baseman plays closer to zone U and makes a bunch of plays there but not as many in zone X (as average), it counts against him even if he converted the same number of GB into outs. JC Bradbury had a good post on this on Sabernomics a while back. Unlike #1, this is a problem unique to ZR.

I'm not sure how much the PS zones impact UZR. If I remember correctly, MGL was going to change his program to accomodate Stats zones. Maybe he already did.

Don't get me wrong--I like ZR--but IMO, UZR is clearly superior, and Range is a very good supplement (and possibly even better).
   31. Chris Dial Posted: November 03, 2005 at 03:08 AM (#1717126)
The correct split is actually 80/20

How do you come by that?

It lumps together all balls in play, not making an allowance for how hard they were to turn into an out. So a hard hit GB into zone V counts the same a soft grounder to zone X. UZR adjusts for this.

UZR doesn't adjust for that. For starters, UZR doesn't differentiate between balls in zone V versus balls in zone U. So a hard hit ball in zone U counts against the 1B as does an easy hit ball. Not so in ZR. That's a much larger problem, IMO.

a first baseman plays closer to zone U and makes a bunch of plays there but not as many in zone X (as average), it counts against him even if he converted the same number of GB into outs.

But this is the same effect on UZR? This is a problem. I personally believe it is alleviated by the proper assignment of zones. Yes, I think ZR could be better, but the destruction of the BIP distribution, yielding the Swisher rating is atrocious.

How can you even hint that Range is complimentary when you miss a player by *54* runs? That's not a "small" difference.

And besides, what Chone did (and what I do) correlate with UZR better than your methodology. I don't know why you'd think Range was more complimentary.

Are you looking at my description of what UZR does? Are you looking at the grid differences?

Yes, ZR could be better, but what UZR does doesn't make it better.
   32. Chris Dial Posted: November 03, 2005 at 03:09 AM (#1717127)
I like ZR--but IMO, UZR is clearly superior

I don't mind that. I disagree, but that's mostly a difference in our experiences.
   33. Spivey Posted: November 03, 2005 at 03:40 AM (#1717146)
Dial, can you post Randa's ZR for 1999, 2000, 2001, 2002, 2003, and 2004? Can you post Sweeney's for the same years, including 2005?

I'd also like to see Chavez's ZR from 1999-2005, if you have it.

I'd also be interested in Tino Martinez, 1998-2001.

Jack Wilson, 2002-2005.

Karros, 1998-2002.

Beltre, 1998-2004.

If you could post those or email them to me (my BTF email is valid), I'd be thankful.
   34. AROM Posted: November 03, 2005 at 04:22 AM (#1717171)
Spivey, you can get zone rating from 2001 on from the Lahman encyclopedia.
   35. Chris Dial Posted: November 03, 2005 at 04:23 AM (#1717172)
Aren't those at ESPN, spivey?
   36. Spivey Posted: November 03, 2005 at 04:25 AM (#1717174)
Does ESPN have ZR's dating that far back?
   37. Chris Dial Posted: November 03, 2005 at 04:26 AM (#1717176)
yes
   38. Spivey Posted: November 03, 2005 at 04:26 AM (#1717177)
ESPN seems to only have ZR since 2001, and I can't even get any of the pages from 2001-2004 to work.
   39. mgl Posted: November 03, 2005 at 05:22 AM (#1717201)
I'll chime in. Chris' criticism of UZR is that the zones I use are not all that great. I can't see how that is a big deal. As long as the zones are not ridiculous, it shouldn't really matter, and the PS zones are not ridiculous. The fact that UZR uses "sub-zones" (as compared to ZR) and all zones (not just ones in which a fielder fields more than half the balls) plus the fact that it adjusts for handedness of the batter, speed of the batted ball, baserunner configuration, outs, defensive park factors, and the exact run value of a ball hit into each zone, and ZR uses none of those things, IMO, renders it much better than ZR, even if the zones that it uses (again, the old Project Scoresheet/Retrosheet zones) are not to Chris' liking. In fact, I don't see how it can be anything but MUCH better than ZR, again, unless it uses ridiculous zones, and I don't see how ZR can add virtually anything to UZR. If someone else other than Chris (no offense to him; it is just that we have already heard his) has an opinion on this, I would love to hear it.

If it will make Chris or anyone else happy, I can easily change the UZR methodology to use the STATS zones rather than the PS zones. That would take me about a half an hour or less. In that case, would there be any doubt that UZR would be a zillion times better than ZR, and that ZR would add nothing to UZR? There are two reasons why I use PS zones (actually STATS zones converted into PS zones). One, I originally did not have any STATS data. Two, I wanted to consolidate zones anyway to reduce sample error. In fact, given that I am not using PS zones anymore (Chris: read this), I am really just consolidating the STATS zones (to make them larger), and there should be nothing wrong with that whatsoever. I would venture that if I change the UZR methodology to "crunsh" each of the STATS zones separately (IOW, not combine or consolidate any of them), the results would be almost the same as the results I get now. And I am not sure that the new ones would be any better, as again, if I do not combine those small STATS zones, I am risking adding a whole lot of noise to the baseline data (the league averages). Since STATS uses 22 fair slices and 4 foule slices and then adds in the number of feet from home plate (in 5 feet increments I think - it used to be 10 I also think), I would probably consolidate the distance grids anyway - I don't think I would want to differentiate between a 250 ft fly ball and a 255 ft fly ball. IOW, since I am going to consolidate STATS sectors anyway, the way I do it now ("converting" them to the PS data) is probably fine.

Chris, if you were under the assumptiuon that I was actually using PS data (as far as I know, they [Palmer/Gillette] no longer have any hit location data anyway), then perhaps you need to rethink your criticism. If you know that I use the STATS data and simply "transcribe" and "consolidate," I still think you need to re-think your criticism as I don't see the merit in your argument, especially as it relates to comparing UZR to ZR, where the latter is so much "coarser" it is not even funny...
   40. DSG Posted: November 03, 2005 at 05:35 AM (#1717210)
How can you even hint that Range is complimentary when you miss a player by *54* runs? That's not a "small" difference.

Who are you talking about?

And besides, what Chone did (and what I do) correlate with UZR better than your methodology. I don't know why you'd think Range was more complimentary.

That's not true. ZR correlates with UZR at about .65 (using a two year sample, based on MAH's work). Range should correlate at about .8 with a two year sample, based on the work that I have done.

More so, Range compliments ZR because ZR's biggest problem--that it does not look at balls out of zone--is covered by Range. That adds something to ZR. As does DRA, or DFT or w/e. Though Range, IMO, is the clearly superior fielding system out there.

<i>Are you looking at my description of what UZR does? Are you looking at the grid differences? </i>

Yes, and I understand that. The difference is that I tend to think of it as a lesser problem than you do. There's room for disagreement here.
   41. Chris Dial Posted: November 03, 2005 at 05:58 AM (#1717239)
There are two reasons why I use PS zones (actually STATS zones converted into PS zones). One, I originally did not have any STATS data. Two, I wanted to consolidate zones anyway to reduce sample error. In fact, given that I am not using PS zones anymore (Chris: read this), I am really just consolidating the STATS zones (to make them larger),

Uh, how are you consolidating them? Do you not see why that doesn't make sense? Do you consolidate teh STATS zones in teh manner that they "match up with" PS zones?

the way I do it now ("converting" them to the PS data)

Well, which is it?

if you were under the assumptiuon that I was actually using PS data

I never was.

Did you read my example in the article? Is that description true or not? Do you consolidate Zone V (in STATS-speak) with Zone T, U? Do you do UZR based on the zones of assignment (see my other articles on ZR)?

If it is only going to take 30 minutes, crunch it that way...
   42. Los Angeles Waterloo of Black Hawk Posted: November 03, 2005 at 06:00 AM (#1717244)
I just now realized that the number signs in the article title represent a fence.

That's about all I have to contribute to this discussion.
   43. Chris Dial Posted: November 03, 2005 at 06:01 AM (#1717246)
Who are you talking about?

Swisher. MGL has him at *+37* You have him at *-23*. That's not possible.

That's not true. ZR correlates with UZR at about .65 (using a two year sample, based on MAH's work). Range should correlate at about .8 with a two year sample, based on the work that I have done.

Of course what I said is true. I'll demonstrate it soon enough. You'll pardon me if I don't accept MAH's work and do it myself.

More so, Range compliments ZR because ZR's biggest problem--that it does not look at balls out of zone--is covered by Range.

You can't look at balls out of the zone because you don't know which balls are out of the zone.
   44. Chris Dial Posted: November 03, 2005 at 06:06 AM (#1717250)
I said this in the article:
MGL converts all of the data from STATS zones to a different grid (Project Scoresheet) that is far less discriminating.

MGL said this in Post 39:
I use PS zones (actually STATS zones converted into PS zones).

How is what I said not true?

Oh, I skipped this part:
the exact run value of a ball hit into each zone,

If you mean to say you use the exact run value of each particular play after it has happened (a liner gets by the RF for a 2B, you count it as a 2B rather than the league avg for a ball hit to that location), then you are doing it wrong.

If you use the actual run value of the play you introduce a significant portion of the adjacent fielder that backs up the play.

If you don't mean that, then nevermind.
   45. Chris Dial Posted: November 03, 2005 at 06:09 AM (#1717254)
In that case, would there be any doubt that UZR would be a zillion times better than ZR,

Well, of course. If I can't check your work, I can't be completely assured you are doing it correctly. You expect *to be able to* check everyone else's work, no?
   46. VoiceOfUnreason Posted: November 03, 2005 at 06:55 AM (#1717281)
If it will make Chris or anyone else happy, I can easily change the UZR methodology to use the STATS zones rather than the PS zones. That would take me about a half an hour or less.

Big Wave!

At least to get some sense for how much it depends on zone selection. Put it back afterwards if the current scheme proves superior, but I'd love to read a report on the deltas (assuming that it isn't much work).
   47. DSG Posted: November 03, 2005 at 07:15 AM (#1717291)
Swisher. MGL has him at *+37* You have him at *-23*. That's not possible.

His ZR is just about average (I have him at -1 according to ZR). So what does that mean? That's almost 40 runs away from what UZR is saying and more than 20 from what I'm saying. That's just not possible.

Look, Range has some problems w/ right field, as do all non-PBP systems. I don't know why this is, but perhaps someone will care to enlighten me (I wish MAH were here: he knows the answers to many of these questions). In fact, here's what MAH wrote in his excellent DRA series on THT:

"there are still lingering problems in right field (though the same is true for ZR and DFT, which seem to do even worse there)"

I bolded the important parts. Clearly, UZR ratings differ sharply from all other systems in RF. ZR correlated at .35 w/ UZR in the corner outfield spots in MAH's study.

Chris, it becomes a little annoying, to be honest, when you refuse to acknowledge the flaws in ZR. It's not a perfect system, and it is not as good as you make it out to be. ZR has a multitude of problems, and in terms of defensive metrics, I don't think it's any better than DRA, Range, or maybe even DFT. ZR has its advantages (that it's based on PBP data), but it has serious drawbacks as well. The problems should not be played down or ignored. It's irresponsible to suggest that ZR is so much better than any other system that no other metric can even compliment it, let alone replace it.
   48. mgl Posted: November 03, 2005 at 08:48 AM (#1717335)
ZR is what it is. It is a little better than range factor in that it takes into consideration a fielder's opportunities, insofsar as the number of balls hit within a certain distance of a typical positioning for each defensive position. That is not a small thing of course. It also distinguishes between putouts and assists by infielders, which RF does not do - another important thing of course. What ZR does not do, which are large negatives, is distinguish among all balls hit in a rather large zone (a fielder's area of responsibility that is part of his ZR may be balls hit right at him or may be balls hit way to his right or left), account for balls hit outside a fielder's "zone," and account for the speed of balls hit, the handedness of the batter (which is somewhat a proxy for positioning and somewhat a determinate of speed of the batted ball), the baserunner and out state, park factors, and maybe a few other things as well. So yes, there are a multitide of flaws in ZR (maybe flaws is a bad term, as it does not purport to account for these things - it is what is advertised). UZR or other similar "true" PBP metrics do not have these flaws. That is a huge difference, especially with regard to four things - one, park factors, two, balls hit outside zones, three, balls hit in different locations within zones, and four, the speed of the batted balls.

The non-PBP, but sophisticated, metrics, like DSG's, which attempt to estimate each fielder's opportunities, are a poor man's ZR. I don't think they add anything to zone rating, but I think they will yield results very close to that of ZR. I consider them a notch below ZR. I think they are great, BTW, and are critical in coming up with decent defensive evaluations for fielders where no detailed PBP data are available, but things like a pitching staffs GB/FB ratio are. They are far, far better than range factor, which is practically worthless without at the very least having a rough estimate of opportunities.

As far as what "zones" I use for UZR over the last 4 years or so, and I apologize if I never made that clear, here is what I do:

I use the STATS hit location data only, which divides the field into 26 equal "slices" (A-Z), 4 of which are in foul territory. A and B, and Y and Z are in foul ground. They then use distance in 5 foot increments from the back of home plate. What I do before the data is crunched is "convert" the STATA hit location data into Retrosheet format, only because my original programs were writtent to accomodate that format. I could have easily re-written the programs to accomodate the STATS locations, but I didn't. I probably would have consolidated the STATS "sectors" anyway, as there would be like 1800 fair sectors (22 slices and around 75 "distance increments").

Anyway, what I do is the following (these are some examples of the "conversions" I do):

STATS slice "C" between 90 and 110 feet "becomes" Retrosheet (RS) sector "5L"
STATS "D" less than 20 feet is RS "2L"
STATS "D" less than 50 feet is RS "25"
STATS "D" less than 90 feet is RS "5S"
STATS "D" less than 150 feet is RS "5"
STATS "E", "F", or "G" less than 20 feet is "2L"
STATS "E", "F", or "G" less than 50 feet is "25"
STATS "E", "F", or "G" less than 100 feet is "56S"
STATS "E", "F", or "G" less than 160 feet is "56"
STATS "T", "U", or "V" less than 20 feet is RS "2R"
STATS "T", "U", or "V" less than 50 feet is RS "23"
STATS "T", "U", or "V" less than 110 feet is RS "34S"
STATS "T", "U", or "V" less than 155 feet is RS "34"
STATS "X" and between 90 and 110 feet is RS "3L"
STATS "W" or "X" and less than 20 feet is RS "2R"
STATS "W" or "X" and less than 50 feet is RS "23"
STATS "W" or "X" and less than 90 feet is RS "3S"
STATS "W" or "X" and less than 145 feet is RS "3"

Again, these are just some examples of how I consolidate in the IF. Basically I combine 2 or 3 STATS slices into one and anywhere from around 20 to 50 feet in distance into one.

In the OF, I combine 2 slices into one and combine about every 50 feet in distance into one.

Although this "scheme" is not perfect, I don't think it is all that bad, and I don't think it will differ all that much from if I were to use all 26 slices separately and maybe use distance sectors in 10 or 20 foot increments.

If I get the chance I might try using much smaller sectors and see what happens. One of the things that will happen is that I will get better results for smaller samples. For larger samples, the results from each methodology will converge, as all of the things that smaller sectors will pick up on will "even out" in the long run. Same thing is true for ZR. In the long run, zone rating will be quite good and will be about the same as a true PBP metric like UZR...
   49. dcsmyth1 Posted: November 03, 2005 at 01:31 PM (#1717374)
---"In the long run, zone rating will be quite good and will be about the same as a true PBP metric like UZR... "

Do you have an estimate of how many seasons or BIP you think it would take for this to be true?
   50. Chris Dial Posted: November 03, 2005 at 02:06 PM (#1717381)
mgl,
thanks for the clarification.

That is essentially what I thought you were doing. Where my concern comes from is the BIP in zone V.

I kind of shrugged off TDF in post 2, but I spent a good deal of last night thinking about how Swisher got his rating, and the old STATS UZR ratings. I doubt this is an event in the IF, but the OF has a bigger chance.

I think what you see with Swisher *is* what TDF was suggesting.

What I believe happens is that the BIP distribution for the OAK RF is skewed enough (or that Swisher simply positions differently enough) that a large portion of his plays are in an "odd" zone, that RFs *that don't play behind teh OAK pitching staff* never see.

That is Swisher BIP distribution *is* different enough so the difference between UZR and ZR are big.

Now obviously I an't look at the actual zone data, but I would think any RF playing for OAK may perform similarly. Maybe not. Who played RF last year? Did OAK's pitching staff change radically in the off-season (yes), but did their profile?

Mike indicated that OAK allowed some 200 fewer singles (was that GBs only?). Perhaps Swisher playing shallow and cutting those off had that effect.

I also like teh "Zone" concept. As it provides a baseline, irrespective of positioning because it is based on "typical" positioning.

I think in Swisher's case, if ZR went:
Swisher
BIZ = 200
OIZ = 150
OOZ = 50

ZR = 1.000

it would do a better job of describing defense. In that case, Swisher would end up with a "Dial rating" of approximately +37. (I re-figured that math - it has to be some other distribution than that exactly)

Does that mean UZR is dead-on, and ZR is way off? It's possible. This extreme (and these data) were used in 1999 in alt.sports.baseball-atlantabraves when arguing over Andruw Jones and whether or not *Andruw primarily caught a different subset of balls in play*.

Now, it takes an extreme - but I don't know if it is an extremely good fielder.

Also, if you want to see exactly what MGL is talking about, print out the two grids in my article and overlay them (hold them up to the light) and you get a very good idea of what MGL is describing.
   51. Chris Dial Posted: November 03, 2005 at 02:27 PM (#1717390)
His ZR is just about average (I have him at -1 according to ZR). So what does that mean? That's almost 40 runs away from what UZR is saying and more than 20 from what I'm saying. That's just not possible.

Sort of. One of us is probably right. But I am not claiming some kind of "complimentary" effect here.

ZR correlated at .35 w/ UZR in the corner outfield spots in MAH's study.

Shrug. That assumes UZR is the *correct* one.

Chris, it becomes a little annoying, to be honest, when you refuse to acknowledge the flaws in ZR.

Good thing I'm not doing that then.

It's not a perfect system, and it is not as good as you make it out to be.

You are entitled to that opinion.

ZR has a multitude of problems, and in terms of defensive metrics, I don't think it's any better than DRA, Range, or maybe even DFT.

I would attribute that to your inexperience or ignorance. Possibly to believing more in what you have read than you should have.

ZR has its advantages (that it's based on PBP data), but it has serious drawbacks as well.

I think it has some drawbacks, yes. I think they can be easily rectified with the data MGL has. I just don't happen to think STATS (or MGL) treat the data quite right, and me given the same data, would treat it differently, but as well or better.

The problems should not be played down or ignored.

Sheesh. You (and all non-pbp methods) downplay the serious drawbacks of your methodology like crazy - you *have* to, because you cannot identify anything from BIP distribution with basic stats, and all the gymnastics in the world can't change that. That's a HUGE, and IMO, drawback. You cannot overcome it with present technology. Swisher *really* points that up. Sure, ZR could be wrong and UZR be right. However, ZR could be right and UZR could be wrong. It is probably somewhere in-between.

*THERE IS NO CHANCE RANGE IS RIGHT*. That's right - 0%. You completely missed the mark on him. Not even close. So, now let's consider that effect on any rankings you do where there is no UZR or ZR for "feel" purposes. Just ponder that for a moment. How do you know when your rating is just right, or if it is off by 60 runs? How do you know?

It's irresponsible to suggest that ZR is so much better than any other system that no other metric can even compliment it, let alone replace it.

Good thing I'm not doing that then.
   52. Chris Dial Posted: November 03, 2005 at 02:45 PM (#1717399)
Chris, it becomes a little annoying, to be honest, when you refuse to acknowledge the flaws in ZR.

I just want to come back to this.

I have been analyzing and discussing the value and quality of ZR, and other systems (DFT, DA, UZR) for nearly a decade.

If you think I haven't discussed every flaw about ZR ad nauseum, and gone through about every theoretical, and never acknowledged its flaws, well, you know squat about my posting history wrt ZR.

Whether or not *you* think so, David, I have discussed these flaws in depth all over the internet. You can look it up. That you think I downplay it is fine with me. I don't think it is "downplaying" - I view the issues as "less significant than others".

There was a world of baseball research on this topic before Baseball Think Factory, before Baseball Prospectus, in just as much depth, and in general, with a much more focused collection of critical minds.

Why? Because the noise and the arrogance of upcoming statheads that think they have re-invented the frickin' wheel - when it is a wheel we've had for centuries. Okay, maybe not that, but it doesn't help.

I have demonstrated *repeatedly* that "new" ideas presented here were discussed in USENET a decade ago.

Me? I'm just a moron compared to the paragons of rsb - names like David Grabiner, Clay Davenport, Dale Stephenson, Ron Johnson, Harold Brooks, Roger Moore and so forth (bah, there are too many to list - Szym can jump in here). Heck, MGL was there - where do you think he got UZR from?

Don't read that to mean there is nothing new to discover and that rsb (and assorted asb groups) knew everything under the sun, but before you thump your chest so hard about how great your work is, or spout off about what someone has or has not acknowledged, do a better lit search.

I'm sorry if that seems harsh, but watch your accusations.
   53. AROM Posted: November 03, 2005 at 03:47 PM (#1717458)
This extreme example, along the lines of TDF's hypothetical in #2, could explain Swisher.

The zones closest to where the RF normally plays, balls are caught 80% of the time, so any ball that is not caught there results in a -0.20 for the player. For some reason, the A's have very few plays hit into these zones, so Swisher doesn't position himself there.

Instead, Swisher puts himself in areas where leaguewide, few balls are hit, and when they are, few rightfielders catch them (20%). When Swisher catches these balls, he gets a +.80 to his UZR.

An extreme example, but could a strange ball distribution pattern combined with good positioning explain the wild differences between ZR and UZR for this player?
   54. Chris Dial Posted: November 03, 2005 at 03:50 PM (#1717459)
Anaheim,
isn't that what I said here:
"I think what you see with Swisher *is* what TDF was suggesting.

What I believe happens is that the BIP distribution for the OAK RF is skewed enough (or that Swisher simply positions differently enough) that a large portion of his plays are in an "odd" zone, that RFs *that don't play behind teh OAK pitching staff* never see.

That is Swisher BIP distribution *is* different enough so the difference between UZR and ZR are big."

8-D

Yes, it may be possible.

I think zone based systems are better for issues like this - because, I suspect *any other RF* in OAK would position himself there.
   55. TDF, situational idiot Posted: November 03, 2005 at 03:56 PM (#1717465)
Chris:

Wow, I guess I wasn't just talking out of my butt (although, in fairness, I said I thought this would be more of a problem with infielders, while your example points to an outfielder).

And I don't want to say that "just" ZR has the problem; it seems to me that <u>any</u> non-PBP system would run into a problem like this.

Now that I've added something of actual value to a discussion around here, can you send me that Strato info?
   56. Mike Emeigh Posted: November 03, 2005 at 04:06 PM (#1717481)
Mike indicated that OAK allowed some 200 fewer singles (was that GBs only?).

It was all singles, not just GBs.

Two things:

1. The comparison I was doing was OAK to TEX - best to worst. OAK didn't allow 200 fewer singles than average.

2. I should have made clear that the numbers for singles are estimated from other data. I have team TB allowed, H allowed, and HR allowed, but not yet the exact breakdown of non-HR hits. My estimates are usually within 5-10 of the totals, but they aren't exact. I generally don't have my PBP database updated until sometime after the first of the year, but thanks to a guy who posts under the Yahoo ID Seaver78 on Ray Kerby's old A.S.S. support group, I have a complete set of basic PBP files which only need to be updated with the Retrosheet info fields and cross-checked against the MLB.tv archives to get BIP types for errors and a couple of other plays. I hope to complete that before the end of the year. I might actually have time to put some PS zone data into them, too, using the MLB.tv archives.

-- MWE
   57. AROM Posted: November 03, 2005 at 04:23 PM (#1717504)
Anaheim,
isn't that what I said here:


Yeah. I was just trying to grok it.
   58. AROM Posted: November 03, 2005 at 04:32 PM (#1717517)
I might actually have time to put some PS zone data into them, too, using the MLB.tv archives.

Are you talking about watching the games and doing the scoring yourself? Or does MLB have some sort of database available?
   59. Chris Dial Posted: November 03, 2005 at 04:41 PM (#1717532)
Now that I've added something of actual value to a discussion around here, can you send me that Strato info?

I have to locate it. I can dig out my notebooks.
   60. Spivey Posted: November 03, 2005 at 05:03 PM (#1717568)
Dial: Could you send me the ZR from post 33? ESPN seems to have taken down their ZR pages before 2005 and Sean Smith's suggestion only dates back to 2001.
   61. Chris Dial Posted: November 03, 2005 at 05:16 PM (#1717588)
Spivey, no, they haven't.

[url="http://sports.espn.go.com/mlb/players/stats?playerId=3138&c>Randa</a>
<a href="http://sports.espn.go.com/mlb/players/stats?playerId=3360&c]Sweeney[/url]
[url="http://sports.espn.go.com/mlb/players/stats?playerId=3953&c>Chavez</a>
<a href="http://sports.espn.go.com/mlb/players/stats?playerId=2388&c]Tino[/url]
[url="http://sports.espn.go.com/mlb/players/stats?playerId=4627&c>Wilson</a>
<a href="http://sports.espn.go.com/mlb/players/stats?playerId=3878&c]Beltre[/url]
<a href="http://sports.espn.go.com/mlb/players/stats?playerId=2590&c>Karros</a>

you goofy Texan...
   62. Spivey Posted: November 03, 2005 at 05:22 PM (#1717594)
Ah. When I tried to access them season by season (listing everyone) the page wasn't found.
   63. Mike Emeigh Posted: November 03, 2005 at 05:41 PM (#1717619)
I think zone based systems are better for issues like this - because, I suspect *any other RF* in OAK would position himself there.

Very likely true - and I think it raises the basic question of how we can best evaluate fielding skill without first determining what the optimal position is for a fielder given the BIP distribution against his team. If the BIP distribution is significantly different from the norm, the optimal positions for the fielders - and the zones that they should be expected to cover - will also be different.

Simple examples:

We know there are hitters like Barry Bonds, Carlos Delgado, Jim Thome, etc. against whom teams position three infielders on the right side of the diamond. We also know that, against the average hitter, that positioning isn't close to being optimal. But because those guys have a BIP distribution that is skewed rightward, both from an actual standpoint and from a cost-to-the-team standpoint, teams adopt these (typically) non-optimal defensive alignments.

We also know that during the early innings of a game, teams will position their 3Bs off the lines. But in the late innings, if that game is close, many teams (not as many as used to) will push their 3Bs into "guarding the lines". Does the actual BIP distribution change in the late innings? If it does, I can't see it, and the fact that some teams are backing away from "guarding the lines" suggests that it doesn't. What changes is the cost-to-the-team aspect; teams choose to deploy their defenses in such a way as to minimize the cost to the team when a BIP is not converted into an out.

Fielder positioning has one purpose, to minimize the cost of a BIP to the team. It's not just about maximizing the probability that a BIP will become an out, but also minimizing the cost of a BIP that does not become an out. The cost function is variable, and situation-dependent; you play the outfielders in when the winning run is on 3B with less than two outs in the bottom of the ninth because the cost of a long fly ball to the OF is the same whether it is caught or not, while the cost of a shallow fly ball to the OF is much larger than normal if it drops.

The reality of measuring fielder performance is that, until we have the data to derive the cost function for a specific team, we're relying on methods that use a static cost function, and that will misrepresent some fielders.

-- MWE
   64. Mike Emeigh Posted: November 03, 2005 at 05:42 PM (#1717621)
Are you talking about watching the games and doing the scoring yourself?

Yes. MLB.tv has "condensed games" which cover about 90% of the plays, and which run about 8-10 minutes each.

-- MWE
   65. Chris Dial Posted: November 03, 2005 at 05:43 PM (#1717622)
I knew what happened, Spivey. That's why I repeated they were there before doing your research for you.
   66. Spivey Posted: November 03, 2005 at 06:00 PM (#1717648)
Yes. MLB.tv has "condensed games" which cover about 90% of the plays, and which run about 8-10 minutes each.

How many games do you plan on watching? Do you have any concern about being unable to tell where some zone boundaries are when you watch the games on mlb.tv?
   67. Mike Emeigh Posted: November 03, 2005 at 06:17 PM (#1717674)
How many games do you plan on watching? Do you have any concern about being unable to tell where some zone boundaries are when you watch the games on mlb.tv?

1. Very likely, I'll focus on teams for which I have a particular interest - right now that's Oakland, Pittsburgh, and the Yankees. It's about 15-20 minutes of effort per game.
2. Using the PS zones, no (I've done it before for a handful of games). On questionable plays, I can go full screen on my monitor. The STATS zones would be a lot harder to map.

-- MWE
   68. AROM Posted: November 03, 2005 at 06:46 PM (#1717736)
Mike, I have no doubt in your abilities to score the zones correctly.

I do have doubt in you being able to score more than a small subset of games for those teams, unless its your full time job.

There just aren't enough days in the offseason to do 162.
   69. Spivey Posted: November 03, 2005 at 06:48 PM (#1717742)
If the games only take 15 minutes a piece, there's enough time. It will take about 120 hours to do those 3 teams, though. It will be a significant investment of time.
   70. Los Angeles Waterloo of Black Hawk Posted: November 03, 2005 at 07:20 PM (#1717807)
A question:

Have any trackers of zone data ever endeavored to record what zone the player stands in?

Looking at the Project Scoresheet grid, it appears that your 2B will normally start off at position 4. However, from time to time, the 2B will be in a shift position, and actually be lined up in 34D. When Robbie Alomar first came up, he always used to line up at what looks like 4D on the grid.

Now, it strikes me that, for any batted ball, we might have some idea of what the average left/right/forward/back range will be for a position. Perhaps, on a medium groundball, when a 2B lines up at 4, your average 2B will make an out on 95% of the balls hit to 4, 60% of the balls hit to 4M, 75% of the balls hit to 34, and 35% of the balls hit to 6M; I don't know.

But when the 2B lines up elsewhere, like 34D, those percentages would be radically different in terms of zones, but we might still have an idea of what his range should be, in terms of how many zones he should cover in each direction.

Does UZR account for this? My interpretation of the above conversation is that it does not.

In the case of a shift for a power-hitting LHB, that's obviously a pretty rare case -- I doubt your average infielder will have more than 30 or 35 BIP per year where he lines up in an extreme shift. But it could have a large impact in the outfield, where outfielders make pretty stark changes in positioning for every batter, and also play deep in later innings (1B and 3B are also prone to guard the line more in innings 7-9, except for Mike Schmidt, who guarded the line 1-3).

I take it that this is the sort of thing that "evens out", theoretically, after two or three years of UZR data.
   71. caprules Posted: November 03, 2005 at 07:49 PM (#1717865)
From a complete amateur observation, one significant difference between how I understand ZR and UZR is that ZR attempts to show what a player did in a given season compared to the average, while UZR is really designed as an evaluation for projection purposes.

As I understand it, UZR ignores BIP that would be regressed to a value of near zero(IIRC that includes catcher and pitcher fielding, infield line drives and outfield fouls). I don't know what percent of BIP that involves, but it seems to me that ZR is better at capturing the performance of a given player in a given season, while UZR could be better over time. Or to put it a different way, if I wanted to use a rating to determine the GG winner for a season, I would use ZR, but if I wanted to look at multiple seasons to see which player I would sign to a contract, UZR would be preferred.
   72. Chris Dial Posted: November 03, 2005 at 08:18 PM (#1717916)
Mike, I have no doubt in your abilities to score the zones correctly.

Clearly you've never met Mike.
   73. Chris Dial Posted: November 03, 2005 at 08:31 PM (#1717939)
Have any trackers of zone data ever endeavored to record what zone the player stands in?

When Mike and I attend ballgames, or get together for meetups, this is what we talk about a lot. How much positioning changes from player to player.

I tend to think it isn't much (and that it doesn't matter with static zones -if the resulting data is treated correctly). Mike (who will correct me if I misstate) believes there is enough positioning changes to affect BIP results (from a ZR view).

I take it that this is the sort of thing that "evens out", theoretically, after two or three years of UZR data.

I thinkthat sort of things evens out anyway. Every team will employ a similar shift against said hitter - and that shift is also dependent upon the game state.

Every batter doesn't get to see the same mix of fastballs/curveballs, nor have to face the toughest pitchers (say, they duck Randy Johnson) - every fielder isn't going to see the same mix of BIP - but over 500 BIP (per position), these things will even out.
   74. Mister High Standards Posted: November 03, 2005 at 08:48 PM (#1717971)
If anyone wants an easier to read print out of the PS zones try HERE
   75. Chris Dial Posted: November 03, 2005 at 08:54 PM (#1717982)
Thanks, Mr. H!

Mine is about 12 years old.
   76. Los Angeles Waterloo of Black Hawk Posted: November 03, 2005 at 09:03 PM (#1717994)
Chris, what now strikes me about your argument is that you claim, in reference to Swisher:

What I believe happens is that the BIP distribution for the OAK RF is skewed enough (or that Swisher simply positions differently enough) that a large portion of his plays are in an "odd" zone, that RFs *that don't play behind teh OAK pitching staff* never see. [emphasis added]

But also say:

I tend to think [how much positioning changes from player to player] isn't much (and that it doesn't matter with static zones -if the resulting data is treated correctly).

We have more than a few examples of how positioning does affect zone-derived ratings; Derek Jeter and Andruw Jones are the most notable, but it now appears possible that Nick Swisher (and possibly other Oakland RF) is another example.

So, I guess my question is, what could MGL do with the Swisher data to correct what you hypothesize to be the problem? Simply analyze the BIP distribution so that we can see if he had a large number of balls hit to unusual zones?

MGL is of course under no obligation to do so (and in fact it may violate the proprietary nature of his data), but a "walk-through" of Swisher's UZR this season would be pretty instructive to the rest of us.

Every batter doesn't get to see the same mix of fastballs/curveballs, nor have to face the toughest pitchers (say, they duck Randy Johnson) - every fielder isn't going to see the same mix of BIP - but over 500 BIP (per position), these things will even out.

And here we have the difficulty. An opponent can "target" a batter by having him face tougher pitchers, and they can tailor their pitch selection and location to his weaknesses. No batter goes out there with the intent of hitting a ball to the 6M zone to exploit Jeter playing SS, unless that batter is Rod Carew.
   77. DSG Posted: November 03, 2005 at 09:36 PM (#1718034)
Chris,

I don't want to derail what is a very interesting series of posts, but your reply led me to believe that you thought I was being arrogant or talking down. I wasn't. I simply interpreted your opinion as downplaying the flaws in ZR rather than not seeing them as big problems. I like ZR, but I think the basic problem is that we disagree on how problematic its flaws are. I tend to believe that the answer is very.

Let's say that the average ZR at a position is .800, and the average player at that position sees 3.5 BIZ/game. So, on average, he will get maybe 2.7 plays in zone/game, and another .5 plays out of zone/game. Player 1 gets to 3.0 BIZ/game and .1 out of zone. Thus, his ZR equals .861, and he'll be +26 runs/150 games. Another player gets to 2.5 BIZ/game, and .6 out of zone. His ZR will equal .756, -18 runs/150 games. Both players get to 3.1 balls/game and see the same number of chances (we're assuming the second part). Both are slightly below average. Yet one is going to be the best at his position, the other the worst. Yes, I'm maginifying the problem, but not by that much, and my point is that ZR has a lot of problems like this.

That's what bothers me most about ZR--beyond that it's a great system. But my point is that this kind of problem does not exist in Range and if if you combine the two systems, you'll be smoothing out most of the anomalous ratings in both. All I'm saying.
   78. Chris Dial Posted: November 03, 2005 at 10:18 PM (#1718108)
So, I guess my question is, what could MGL do with the Swisher data to correct what you hypothesize to be the problem? Simply analyze the BIP distribution so that we can see if he had a large number of balls hit to unusual zones?

Yes. I claim they aren't unusual overall, and I suspect MGL re-aligning the zones to STATS will see a big chunk of that +37 evaporate. If Swisher catches balls at the edge of his zone more often, MGL will have them valued too high.
   79. Chris Dial Posted: November 03, 2005 at 10:30 PM (#1718126)
and .6 out of zone.

This is the problem - that is too many. I have serious doubts that a player gets 25% of his balls OOZ. The reason is that the zones are defined *specifically* to balls that players do get to.

They aren't andomly drawn by what someone thinks a SS should get to - they are determined by years of data indicating that *this* is the set of zones that a player turns into outs at least 50% of the time.

Many people focus on this flaw without recognizing there is no evidence (or very little) that that occurs. In terms of analyzing the data, your numbers are also WAY too high.

BIZ = 3.25 at SS, easily the most (on average).

Yes, I'm maginifying the problem, but not by that much, and my point is that ZR has a lot of problems like this.

I don't agree. You list this as a hypothetical. Can you demonstrate that it occurs (I know the answer is 'no')? You are listing a *potential* flaw, not a flaw. That something *can* be wrong with a system doesn't mean there *is* something wrong with the system.

And it also depends on how much this is an issue - WHICH IS AVAILABLE in the raw data.

It's the present presentation of the data that has the potential flaw, not the data itself.

And you say "a lot of problems like this" - that's one potential way it can be messed up. What are the others?
   80. DSG Posted: November 03, 2005 at 10:36 PM (#1718138)
my point is that ZR has a lot of problems like this

I meant that this problem applies to many players. I would suggest that any time UZR agrees with another well-constructed metric that looks at all BIP (this could be Range, PMR, DFT, DRA, CAD) and disagrees with ZR, the reason is probably the BIZ/BOZ problem. This flaw is unique to ZR in that all other defensive systems look at all BIP.
   81. mgl Posted: November 03, 2005 at 11:05 PM (#1718204)
I must say that this is an interesting and informative discussion and I hope it continues for at least a few more days (maybe Dan can bump it up to the top). I have not kept up on all the defensive work and discussion for quite some time, for various reasons, and I realize that UZR is not nearly as good as it can be (there has to be an "ultimate" methodology given a certain set of data).

I am somewhat committed to updating (improving) the UZR methodology. I have some ideas which I'll throw out there in the next few days, and hopefully I'll get some workable suggestions as well...
   82. AROM Posted: November 03, 2005 at 11:46 PM (#1718266)
My understanding is that if a player makes a play on a ball out of his zone, it counts towards his plays and opps, otherwise the play is ignored.

ZR would be much, much better if STATS gave us 4 columns:

Opps - Plays made- ZR- Plays out of zone

Plays out of zone would not be counted in ZR, but reported as a complementary stat.

The data is there, I just wish it were reported that way.
   83. Chris Dial Posted: November 04, 2005 at 12:03 AM (#1718303)
mgl,
there's no doubt that you are quite instrumental to the discussion, as you have the data.

I agree with Rally:
I like
BIZ, OIZ, OOZ, ZR. I can subtract things myself - I just need to get a rough idea of how many OOZ plays there are.

DSG's math (#77) is the correct critique. The question that stands to be resolved is: how much of a flaw is it? If MGL's data can say "less than 10%" or "greater than half the players", then we can really identify whether or not ZR is a quality measure.
   84. Spivey Posted: November 04, 2005 at 12:12 AM (#1718320)
Well, I looked at the 7 players I asked for, and there was no trend at all. Obviously only 7 players will make it very difficult to see anything but noise anyways, but I wanted to give it a look. I might end up looking at more 3B and 1B in the future (and possibly RF and LF).
   85. DSG Posted: November 04, 2005 at 12:14 AM (#1718328)
Yes! I hadn't thought about this before, Chris. MGL, could you tell us what % of outs fall outside of zones where more than 50% of balls are caught for each position?
   86. Chris Dial Posted: November 04, 2005 at 06:56 AM (#1718582)
In looking at Swisher's mark, and writing my Fielding Awards piece, I googled for MGL's methodlogy. I can't find it anywhere (fix that, Jim/Tango).

But I did find a link to last year's UZR data.

Hey! Look who led teh AL in RF - the Oakland A's Jermaine Dye. Dye is a decent fielder, so it is possible.

Of course, that matched my ratings last season (Dye at +12/150 DG).
   87. dcsmyth1 Posted: November 04, 2005 at 01:40 PM (#1718664)
Thanks, Chris. I missed that interesting discussion.
   88. Ron Johnson Posted: November 08, 2005 at 08:59 PM (#1723964)
Chris,

Much of this discussion reminds me of a few DA (and MGL's work reminds me a lot of DA -- particularly Dale Stephenson's work converting DA to runs) vs ZR discussions on RSB and RSBA.

I recall a great back and forth between Dale Stephenson and Keith Woolner that pretty much covered the whole nine yards.

Bottom line is that where there's a big difference in evaluation it's likely to be because ZR in effect places greater weight on converting the routine (or semi-routine). Charlie Saeger (who came up with a pretty decent fielding system himself) said that ZR weights hands over range while DA places the greatest emphasis on range. (And for *most* players it comes out in the wash)
   89. AROM Posted: November 08, 2005 at 09:10 PM (#1723997)
ZR weights hands over range

I've heard this before and assumed it was true. Now I'm not so sure. Thanks to Chris's articles I see that ZR covers almost all of the field. Everything in the infield except straight up the middle and the 2 holes, including even down the lines, and for flyballs, everything in the outfield.
   90. Chris Dial Posted: November 08, 2005 at 09:20 PM (#1724025)
Rally, to be fair to Ron, Dale, Keith and Charlie, the original zones for ZR were slightly smaller than what they have been changed to - and this was a discussion from the early 90s, rahter than aroudn today's ZR.
   91. Mike Emeigh Posted: November 08, 2005 at 09:21 PM (#1724028)
Bottom line is that where there's a big difference in evaluation it's likely to be because ZR in effect places greater weight on converting the routine (or semi-routine). Charlie Saeger (who came up with a pretty decent fielding system himself) said that ZR weights hands over range while DA places the greatest emphasis on range. (And for *most* players it comes out in the wash)

I think this summarizes the issue quite well.

If a player is trading in-zone plays for OOZ plays, it's likely that the tradeoff will *hurt* his ZR, even if he winds up making more plays as a result. Suppose league-average ZR is .750. Player A sees 100 balls in-zone, and converts 75 of them. Player B sees the same 100 balls in zone, and converts 70 of them - but he gets to 5 out-of-zone plays. Even if player B made 5 OOZ plays in exchange for just 2 in-zone plays - making 73 plays in zone and 5 OOZ, vs. the 75/0 that player A makes - player A still has the ZR advantage.

I don't know how often this happens (my sense is that, like Ron suggests, it does all come out in the wash for most players, so it doesn't really happen very often), but I think that when it does happen we do need to take a closer look at what's going on. In general, I think that's true for any analytical method; it's easy to dismiss an anomalous result as "just" an outlier, but we ought to be looking at all of the anomalous results just to make sure there isn't something we missed.

-- MWE
   92. Chris Dial Posted: November 08, 2005 at 09:41 PM (#1724077)
I don't know how often this happens (my sense is that, like Ron suggests, it does all come out in the wash for most players, so it doesn't really happen very often), but I think that when it does happen we do need to take a closer look at what's going on

As is noted, MGL has the actual data, so this question is in hand - we just have to read it.

I don't know how much of a problem it is, but when you make hypotheticals you need to approach realism more - total chances and 5 converted - otherwise, while hte math is easier, it's exxaggerated for effect.
   93. DSG Posted: November 09, 2005 at 10:54 AM (#1724756)
There's one other difference in the way UZR rewards range and ZR does not, which is that each BIP in UZR is given a seperate run value while ZR aggregates all plays together. This isn't really a big deal, but a guy that gets to a lot of would-be doubles and triples in exchange for allowing more singles will be rewarded in UZR, and get no credit for that in ZR.
   94. Chris Dial Posted: November 09, 2005 at 01:16 PM (#1724775)
UZR is given a seperate run value while ZR aggregates all plays together. This isn't really a big deal, but a guy that gets to a lot of would-be doubles and triples in exchange for allowing more singles will be rewarded in UZR, and get no credit for that in ZR.

I know. We have discussed this a good deal in rsb with Dale and the gang.

It's a reasonable position to claims that catching balls in zones that have a higher run value (on average, obviously). Defensive Runs back in the day used to rate "extra base hits prevented" based on the same zones that MGL is using. Unfortunately those zones are too large to differentiate what you are claiming.

However, it is not a good idea to use the actual run value of balls not caught. Using a league average is a better idea because when you use actual value, you introduce the skill of the other outfielders. If Austin Kearns miss a ball in the gap. he has to wait on Ken Griffey to start running over there (triple), while Jacque Jones already has Torii Hunter fielding it on a bounce (single), neither of which had anything to do with the fielder being rated, for a significant change in run value.

I would also suggest that UZR and traditional methods do *the exact opposite*. They rate cutting off short loopers about as highly as they rate deep hit balls due to the "percentage" of plays made there.

My "average run value", IMO, and the sample size covers what you are suggesting.
   95. DSG Posted: November 09, 2005 at 01:26 PM (#1724778)
Chris,

Probably but not definitely. Obviously, as sample size increases, the differences are smoothed and eventually dissapear. But that's probably true of all systems. Over small samples, there could be large differences between ZR and UZR for the reason I gave. Your criticisms, of course, have merit, and suggest that UZR may not be doing a better job than ZR over those small samples (though I personally do think it does).
   96. Chris Dial Posted: November 09, 2005 at 01:37 PM (#1724782)
DSG,
I think UZR will do a better job over some of hte small sample sizes and worse over others -because in some of the UZR plays, MGL is counting BIZ as BOZ (effectively - by adding a In-zone Zone to a Out of Zone zone).

That changes the value of "percentage of balls caught by this position in these two zones" from a 20%(OOZ)/60%(BIZ) to merging both to a 40% value. Which is wrong to do.

It overvalues the BIZ plays by 20% and undervalues the OOZ plays by 20%. Depending on teh BIP distribution, this can have a dramatic effect on runs saved.
   97. Los Angeles Waterloo of Black Hawk Posted: November 09, 2005 at 08:12 PM (#1725292)
This isn't really a big deal, but a guy that gets to a lot of would-be doubles and triples in exchange for allowing more singles will be rewarded in UZR, and get no credit for that in ZR.

I have often wondered about this, as ZR is basically OBP Allowed, and we have no measure for SLG Allowed. For middle infielders, of course, that will be close enough to equal for government work, but is that true for outfielders?

Last offseason, I looked at some old Defensive Average/Fielding Run data from the rec.sport.baseball heyday.

This was data that went from 1988-1992. Using <a href="http://groups.google.com/group/rec.sport.baseball.analysis/browse_thread/thread/6662e89f72d587a5/908d1d9b6b6674c4?q="chris+dial"+dpi&_d>the Dial method</a> ... well, let's back up a second. I'll come back to that.

If you search Google groups somewhere, you'll get Defensive Average data like the following, featuring Brett Butler. DA is Defensive Average, NHS is Net Hits Saved, NEB is Net Extra Bases Saved, and FR is the Fielding Runs (IIRC, this is Dale Stephenson's work).
Year   DA   NHS   NEB   FR
1988  .594   12    17  13.69
1989  .592   -3    13   2.33
1990  .595    0    11   3.68
1991  .591    7    14   9.24
1992  .556  -29     5 -19.40
TOT   .586  -13    60   9.54
Okay, now we can go back up to Dial's method. The purpose of it, as far as I can tell (and not to speak for Dial), is to estimate how many runs a defender saved in the absence of knowing how many extra bases he allowed. Basically, just going off the hits saved only -- because the NHS figure is the only one we can guesstimate using ZR.

So, let's add too more columns: EFR (Estimated Fielding Runs [using NHS only, and the weights devised by Dial]) and Diff (EFR - FR):
Year   DA   NHS   NEB     FR    EFR   Diff
1988  .594   12    17  13.69   9.74  -3.95
1989  .592   -3    13   2.33  -2.43  -4.77
1990  .595    0    11   3.68   0.00  -3.68
1991  .591    7    14   9.24   5.68  -3.56
1992  .556  -29     5 -19.40 -23.55  -4.15
TOT   .586  -13    60   9.54 -10.56 -20.10
So, just going off of this one guy, you can see that using just NHS will get you withing +/-5 runs for a season.

But here's something else: amongst center fielders from this data set, this is by far the most dramatic case of NHS Alone missing a player's value. Well, Roberto Kelly's kind of close -- he had -32.78 FR and -18.68 EFR in that period, which is basically thrown off by one season. Griffrey was off by about -13.50 runs over the course of those seasons.

John Shelby was off, too, actually, by 17.76, but he wasn't a regular for one of the years included (for him, it's only 1988 and 1989). But everyone else is pretty close.

Even if you just have zone rating data, you will almost always be within 5-10 runs, even in the most extreme cases. And that's for a CF, where you might expect that ZR doesn't accurately measure extra-base prevention vis-a-vis hit prevention.

The only other position I managed to work through was 1B; if you just have ZR, you're going to be within a run or two, at worst, for a season. For five seasons, you might end up being off by five or six runs.

Maybe I'll get to LF and RF one day, but I'm fairly comfortable with guestimating runs off of a ZR or a DA.
   98. Los Angeles Waterloo of Black Hawk Posted: November 09, 2005 at 08:18 PM (#1725302)
Actually, there's one CF I left out of that: Devon White. He had 138.15 Fielding Runs over that period, but "only" 110.44 EFR; he's actually a bigger miss than Butler. Still, the worst you get for any single season is 1990, where you're off by 8.53 runs. And this is a pretty extreme player ... and you're still within 5-10 runs for a single season, which isn't terrible.
   99. Los Angeles Waterloo of Black Hawk Posted: November 09, 2005 at 08:42 PM (#1725350)
To further illustrate the above, there are 126 CF seasons in the data set I looked at. Here they are in descending order of the aboslute value of the difference between Fielding Runs Estimated by NHS and actual Fielding Runs:
Yr      Player                    FR      EFR  ABS(Diff)
88Shelby,John          10.21-0.8111.02
92Felix,Junior        10.50.0010.50
90McGee,Willie        -11.48-21.119.63
90Kelly,Roberto        -3.136.509.63
90Dykstra,Lenny        10.7819.498.71
90White,Devon          17.468.938.53
89Gallagher,Dave      -11.73-3.258.48
89Samuel,Juan          -4.833.258.08
89White,Devon          41.0733.297.78
88McGee,Willie        -16.86-9.747.12
89Griffey,Ken_Jr      -16.04-8.937.11
91Lankford,Ray        -12.71-5.687.03
89Shelby,John          3.49-3.256.74
88Wilson,Willie        -14.64-8.126.52
88Young,Gerald        -11.21-4.876.34
92White,Devon          26.5320.306.23
88Cotto,Henry          -13.44-7.316.13
91McRae,Brian          -3.282.445.72
89Dykstra,Lenny        0.876.505.63
91Devereaux,Mike      -1.91-7.315.40
89Young,Gerald        -9.41-4.065.35
92Jackson,Darrin      15.910.565.34
92Martinez,Dave        -1.284.065.34
88Yount,Robin          23.1717.875.30
92Johnson,Howard      -14.94-9.745.20
90Henderson,Dave      26.1721.115.06
88Moseby,Lloyd        -5.52-10.565.04
89Carter,Joe          -16.22-11.374.85
92VanSlyke,Andy        -8.06-3.254.81
89Butler,Brett        2.33-2.444.77
89Kelly,Roberto        -24.86-20.304.56
89Yount,Robin          -6.25-10.564.31
89Moseby,Lloyd        -18.5-22.744.24
88Wynne,Marvell        -1.762.444.20
92Butler,Brett        -19.4-23.554.15
92Dascenzo,Doug        -0.04-4.064.02
92Puckett,Kirby        -14.56-10.564.00
88Butler,Brett        13.699.743.95
89Thompson,Milt        -3.070.813.88
92Johnson,Lance        -7.94-4.063.88
91Johnson,Lance        39.5735.733.84
89VanSlyke,Andy        4.348.123.78
88VanSlyke,Andy        10.8914.623.73
92Nixon,Otis          13.459.743.71
90Butler,Brett        3.680.003.68
91Burks,Ellis          -2.850.813.66
92Lankford,Ray        -0.383.253.63
91Gonzalez,Juan        -7.66-4.063.60
91Butler,Brett        9.245.683.56
91White,Devon          35.2131.673.54
90Moseby,Lloyd        3.036.503.47
89Wilson,Willie        -0.23.253.45
89Espy,Cecil          10.657.313.34
91Walton,Jerome        4.120.813.31
88Wilson,Mookie        -9.77-12.993.22
88Thompson,Milt        -0.05-3.253.20
90Griffey,Ken_Jr      -25.06-21.933.13
90Johnson,Lance        7.4610.563.10
90Boston,Daryl        -8.62-5.682.94
91Cole,Alex            9.396.502.89
92Gonzalez,Juan        -19.82-17.052.77
92Griffey,Ken_Jr      -9.97-7.312.66
88Burks,Ellis          -3.47-0.812.66
88Carter,Joe          -18.85-16.242.61
89Henderson,Dave      16.118.682.58
92Lewis,Darren        3.946.502.56
90Puckett,Kirby        -23.53-21.112.42
91Puckett,Kirby        -34.07-31.672.40
89Pettis,Gary          -3.97-1.622.35
90Yount,Robin          -3.35-5.682.33
91Grissom,Marquis      6.344.062.28
88Puckett,Kirby        -5.94-8.122.18
92Lofton,Kenny        28.1525.992.16
92Wilson,Willie        19.0221.112.09
91Henderson,Dave      5.257.312.06
89Murphy,Dale          -2.87-4.872.00
92Grissom,Marquis      16.5714.621.95
88Gallagher,Dave      7.615.681.93
88Davis,Eric          -3.83-5.681.85
89Martinez,Dave        -2.62-0.811.81
92Yount,Robin          4.716.501.79
90Yelding,Eric        1.483.251.77
92Devereaux,Mike      11.2412.991.75
92Dykstra,Lenny        2.344.061.72
90Burks,Ellis          -11.3-12.991.69
88White,Devon          17.8816.241.64
88Henderson,Dave      24.3622.741.62
88Webster,Mitch        -0.660.811.47
89Davis,Eric          -11.53-12.991.46
92Finley,Steve        -0.18-1.621.44
88McDowell,Oddibe      -1.85-3.251.40
90Pettis,Gary          -2.75-4.061.31
90Wilson,Mookie        -0.470.811.28
92McRae,Brian          7.78.931.23
92Cuyler,Milt          -9.3-8.121.18
90Martinez,Dave        21.420.301.10
89Nixon,Otis          13.5514.621.07
89Walton,Jerome        -2.2-3.251.05
88Dykstra,Lenny        27.7826.800.98
91VanSlyke,Andy        -17.7-18.680.98
88Washington,Claudell  -7.47-6.500.97
90Devereaux,Mike      12.3211.370.95
91Gant,Ron            -8.17-7.310.86
91Griffey,Ken_Jr      -14.45-13.810.64
92Sanders,Reggie      -8.72-8.120.60
88Pettis,Gary          16.4717.050.58
90Walton,Jerome        -9.46-8.930.53
91Williams,Bernie      -14.3-13.810.49
91Finley,Steve        3.574.060.49
90Gant,Ron            -10.08-10.560.48
88Martinez,Dave        -9.4-8.930.47
89Devereaux,Mike      15.0714.620.45
89Puckett,Kirby        -17.45-17.050.40
91Yount,Robin          -3.61-3.250.36
90VanSlyke,Andy        -7.77-8.120.35
89Gwynn,Tony          -2.93-3.250.32
91Davis,Eric          -5.98-5.680.30
91Pettis,Gary          8.678.930.26
91McGee,Willie        -7.51-7.310.20
91Young,Gerald        3.443.250.19
90Webster,Mitch        26.9926.800.19
91Jackson,Darrin      9.078.930.14
91Cuyler,Milt          12.0912.180.09
92Kelly,Roberto        -4.79-4.870.08
90Carter,Joe          -1.66-1.620.04
89Burks,Ellis          -1.62-1.620.00
Of the 126 seasons:

126 (100%) are within 15 runs;
124 (98.4%) are within 10 runs;
107 (84.9%) are within 5.5 runs;
99 (78.6%) are within 5 runs;
50 (39.7%) are within 2 runs;
39 (31.0%) are within 1.5 runs; and
28 (22.2%) are within 1 run.
   100. Los Angeles Waterloo of Black Hawk Posted: November 09, 2005 at 08:48 PM (#1725362)
Damn you, Live Preview, and your lies ... it would take me forever to re-format that chart. Just believe me.
Page 1 of 2 pages  1 2 > 

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
robinred
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 0.7567 seconds
70 querie(s) executed