Username:	Password:

Remember me

Forgot your username/password?

Print Email

Home>Articles

Andruw of Center Field

By Bill James

January 15, 2018

2018-3 1. Media

I will be all over the media this week. I taped an interview with Russell Roberts of EconTalk (Podcast) in late November or early December; that will be available to the public today. Dr. Roberts is very, very professional, meticulous in preparation for the interview; I hope it will be interesting.

Also, my yearly Top 10 contributions to the MLB panel are now running on MLB-TV, I think; I haven’t seen them but I am getting feedback about them, so I assume they are on. Also, I taped an interview with C-Span which should begin running this week or perhaps next week. So I’ll be around.

2. Andruw

After spending a couple of days on Twitter arguing about Andruw Jones’ qualifications for the Hall of Fame, I received this e-mail from my friend Yo Poz:

Bill, I have been sort of following your Andruw Jones thread on Twitter and what I have seen, I guess I would condense it as follows:

1. The only real argument for Andruw Jones as a Hall of Famer is if you are convinced he’s the greatest defensive centerfielder of all time.

2. The argument that he is the greatest defensive centerfielder of all time is specious and unconvincing.

3. There is no other real argument for Andruw Jones because he was not a notable or outstanding offensive player, base runner, leader, postseason performer, etc.

Do I have that generally right?

The three statements are accurate, but they dodge the real issue. The real issue is that people who see themselves as pro-analytical or post-analytical revolution, people who see themselves as sophisticated consumers of information, are in fact behaving in a manner which is identical to the pre-analytic arguments commonly used before 1975. They argue that Andruw Jones has 63 WAR or whatever it is and that other players who have 58 WAR are in the Hall of Fame, therefore Andruw should be in the Hall of Fame as well. This is no different than arguing that Herb Pennock won 240 games and he is in the Hall of Fame and Waite Hoyt won 237 games and he is in the Hall of Fame and Whitey Ford won 236 games and he is in the Hall of Fame, so David Wells, with 239 wins, obviously deserves to be in the Hall of Fame as well. It is precisely the same argument; it is just using a "new" statistical category, rather than an old one. Or, to apply it to a hitter, Yogi Berra drove in 1,430 runs, Charlie Gehringer drove in 1,427 runs, Joe Cronin drove in 1,424, Jim Bottomley drove in 1,422, Robin Yount drove in 1,406 and Ed Delahanty 1,400, and all of those guys are in the Hall of Fame, so how can you say that Joe Carter shouldn’t be in the Hall of Fame when he drove in 1,445 runs, you moron, you.

Andruw Jones is not going to be selected to the Hall of Fame this year one way or the other, so that is not a real issue, and it doesn’t actually make any difference whether he stays on the ballot or not; his chance of eventually being elected is the same either way, so that’s not a real issue to me, although I can understand that people don’t like the process because it is a flawed process. I agree that Jones was a superb defensive center fielder, so that isn’t the issue, either.

Most people are not "analysts"; they are consumers of information, and before they were using old information and now they are using newer, better information, so what’s the problem with that?

The problem is this. It ultimately damages the analytic community, and it will ultimately bring disrepute upon our community, if we allow people who pretend to be members of our community to make arguments which have obvious flaws, and which may be false because of those flaws. If we allow people who think they are "with us" to make arguments which may turn out in the long run to have been complete nonsense, without contesting those arguments, without pointing out the problems, then it ultimately comes back on us.

WAR looks to an un-skeptical reader as a "comprehensive" statistic—but what people don’t get is that pitcher’s wins, in their day, were understood to be a comprehensive statistic, as well. A pitcher's win total summarized EVERYTHING the pitcher did--until we realized that it doesn't. It was very, very common, 40 years ago, for people to say that RBI were the most important statistic for a hitter. Now it is WAR, but later it will be something else—or, at a minimum, some better, more carefully constructed versions of WAR.

Let’s say for the sake of argument that I am 100% comfortable with the runs saved estimates derived on behalf of Andruw Jones; in fact it might be 85% or 90%, but it doesn’t matter. Let’s say that we’re 100 confident in those numbers. The problem is that the Runs Saved estimates derived in this manner are much larger than the Runs Saved estimates derived by older methods, which are necessarily conservative because of the limitations of the data. They are not just larger for Andruw Jones; they are larger in general. They are larger for Andrelton Simmons than for Luis Aparicio. Andrelton Simmons’ dWAR as a first-year regular in 2013 was larger than Ozzie Smith’s career high.

Andruw Jones’ career dWar is shown as 24.1, in 17039 innings, whereas Willie Mays is shown at 18.1 in 24,427 innings. On a per-inning basis, Jones is being credited with saving twice as many runs as Willie Mays, compared to a replacement level center fielder.

Well, I believe that Andruw Jones was a fine defensive center fielder, but I don’t necessarily believe that he was twice as good a center fielder as Willie Mays. I’m a little skeptical. The problem is, though, that the people who are relying on Andruw Jones’ WAR as the basis of his Hall of Fame claim are, in fact, relying upon the conclusion that he was twice as good a defensive center fielder as Willie Mays. …or leave Willie Mays out of it; Garry Maddox. Maddox’ per-inning is about the same as Mays’, which is a little over one-half of Jones’. The people who are using Jones’ WAR to argue that he should be in the Hall of Fame are relying on that statement, but they either

a) do not understand that they are implicitly relying on this claim, or

b) will not admit that they are implicitly relying on this claim, or

c) don’t think that it’s a problem.

WAR seems to the less-educated public to be a fair and consistent method between generations, when in reality it is an apples-and-oranges comparison between generations. That’s not a criticism of WAR; it’s an inherent limitation of the facts available to us at this time.

If people do not understand that this is what they are doing. . . well, that’s fine; they’re not analysts, they don’t quite get it. 99.9% of people are not baseball analysts; it’s not a criticism of them that they’re not. It is our responsibility to help people understand this problem, not their responsibility to figure it out on their own.

But if they don’t understand this and they are pretending to be sophisticated data analysts, then that’s a problem. If you’re not an environmental scientist, that’s not a problem, most people aren’t. If you’re not an environmental scientist and you’re pretending to be an environmental scientist, then that’s a problem. Then it becomes a responsibility of the environmental scientists to point out that you are not an environmental scientist, and you don’t know what the hell you are talking about. The people who are insisting that Andruw Jones should be in the Hall of Fame because he was the greatest defensive center fielder of all time do not know what the hell they are talking about, and it is an absolute responsibility of those of us who are in this field to say so.

I wish I could stop there, because that’s a better closing sentence than anything else I have, but there were issues I didn’t get to. Some people will say "Maybe Andruw Jones really WAS a better defensive center fielder than Willie Mays. Maybe Andrelton Simmons really IS a better defensive shortstop than Ozzie Smith."

Well. . .yeah, maybe he was. Maybe he was, maybe he wasn’t. Let’s figure that out before we elect him to the Hall of Fame, rather than electing him first and then working on the gigantic hole in the argument for him later. That’s a way of saying that you don’t think it’s a problem, that we’re comparing apples to oranges. I think it’s a problem.

Another argument that people make is that if we can’t elect Andruw because of this issue, then we can’t elect anybody post-2000, because we’re always comparing apples to oranges. But of course we can elect other players post-2000. It’s a problem for Andruw because Andruw’s Advocates are making a remarkable claim for him, which rests entirely on data which is not fully understood at this time with regard to inter-generational comparisons. It’s not a problem for Miguel Cabrera, because Miguel’s Minions are not making any remarkable claim for him based on data that isn’t completely understood yet. It isn’t a problem for Miguel, or Albert Pujols, or David Ortiz, or Ichiro Suzuki, or Mike Mussina or Roger Clemens or Greg Maddux or Roy Halladay or Curt Schilling or Derek Jeter or Chipper Jones or Adrian Beltre or anybody else, because none of those men are making any remarkable claims based on data which is not fully vetted in this context. It’s a problem for Andruw because he is.

Thanks.

COMMENTS (64 Comments, most recent shown first)

MarisFan61
P.S. to Brock: I'm well familiar with Bill's "letter grades" on fielding in the Win Shares book. I find them extremely interesting (if sometimes shocking!) and I often cite them.
11:57 AM Jan 20th

MarisFan61
Gary: Is it still available? If so, can you give a link?
10:23 PM Jan 19th

garywmaloney
Win Shares Digital Update followed the publication of the main Win Shares book (which was published by the old STATS combine, at about the same time as Bill's main publisher released the New Historical Abstract.

It contained masses of numbers and tables that could not be fitted reasonably into the original Win Shares book (it would have doubled the size of the book.) It can be emailed to you -- not that large a file as i recall -- or put on Dropbox. However -- the Digital Update file did come with a PASSWORD -- I think I remember it.
9:47 PM Jan 19th

Brock Hanke
Checking MWeddell's data a bit further, I looked up Jones' defensive innings, 17078.1, and divided 85.5 by that and then multiplied by 1000. Jones, for his career, had 5.01 Defensive Win Shares per 1000 Defensive Innings. I had to include Jones' innings at positions other than the outfield, but there aren't very many of those. 5.01 is a high A+. The A line is 3.30. On the other hand, as of 2001, Jones had 6.47 DWS per 1000 IP, which is a gigantic number. (That number is found in a listing of 2001 Gold Glove winners in Win Shares.) So it's certainly true that Jones lost a lot of value in the last parts of his career. Also, 5.01, while way above the A line, isn't absurdly high. Torii Hunter, as of 2001, had 4.88. Jim Edmonds was above 4. Mike Cameron was about 4.5. None of those guys was done as a player. All those numbers are from the 2001 Gold Gloves. And none of those guys was Tris Speaker or Willie Mays or Curt Flood.
5:45 PM Jan 19th

MWeddell
Running through some quick and dirty math ...

The John Dewan article posted on 1/19 lists Jones as having 85.5 fielding win shares. Divide by 3 = 28.5 wins. Baseball-reference credits Jones with 24.1 dWAR, but that includes the positional adjustment so I believe he has 24.7 dWAR without the positional adjustment. [Baseball-refernence's presentation of WAR is confusing, with the positional adjustment included in both oWAR and dWAR but counted only once in total WAR.]

So, Jones' career WAR would be roughly 3.8 higher if one modified baseball-reference WAR to use Defensive Win Shares (but of course all the comparable players' totals would change too). Also, it appears that Bill's complaint really is that Mays, etc. have their defensive rating muted rather than that Jones' defensive value was pegged too high.
3:34 PM Jan 19th

Brock Hanke
Marisfan - No, the book Win Shares does not carry separate Fielding Win Shares anywhere. What it DOES have is a listing, by position, of most players who had a significant career, along with their "Grades" (A+ to F), based on their Fielding Win Shares per 1000 Defensive Innings. These listings are divide by position, and only use stats that the player compiled while playing that position. So there are many players who played more than one position, but only played enough time at one of those positions to show up on Bill's chart. You can figure out the grade for a more modern player by looking up the Fielding Win Shares on Baseball Gauge, looking up the Defensive Innings on BB-Ref, and dividing. The book Win Shares does list the breakpoints for the individual grades. This is VERY useful in getting a general feel for how good or bad a modern player is. When Win Shares was published, in 2001, Andrew Jones had the highest number in CF, by a large margin, much higher than any of the other A+ CFs. But, as Bill noted, he was only part way through his career.

I first worked this out when dealing with George Sisler, who was my father's favorite player. Sisler's defensive grade as a first baseman, for his career, is C-, which is completely inconsistent with his reputation. I speculated that his grade before 1923, when he missed the whole season with some sort of physical problem, might be much higher than his grade after 1923. This turns out to be true. Before 1923, Sisler grades out as an A. After 1923, he grades as an F. For the career, a C- makes sense. But what it really tells you is that Sisler, after 1923, was a completely different player, and much worse, than he had been before whatever happened to him. The George Sisler through 1922 is the one who got elected to the Hall of Fame in 1939, of course.
3:13 PM Jan 19th

MarisFan61
Gary: What and where is "Win Shares Digital Update"?
I never knew of it.
Seems like it would have been good to know -- and maybe still is!

Because, depending on who does the method and how it's done, it could be better than what we find on the site I mentioned.
Or vice versa.

Let me ask: What does anyone know about who and how, on either or both of those?
(It's important.)
1:16 PM Jan 19th

garywmaloney
MarisFan -- Win Shares Digital Update had many many tables (that was the reason it was digital only), the coolest ones were the leaderboards giving best fielders and hitters at each position for each league.

BaseballGauge supersedes all this.
12:32 PM Jan 19th

Guy123
Brock: yes, it's great that B-Ref has fielding innings. I was simply too lazy to track them down for all the players I was looking at. In broad strokes, PA is a reasonable proxy. But a serious study of these issues should definitely use innings.
12:01 PM Jan 19th

MarisFan61
Brock: The Win Shares book gives FIELDING Win Shares for all player for all those years, or even for many players for any of those years?
If it does, I'd love to know where.

(Remember, what we were wondering about here was availability of FIELDING Win Shares, especially for years prior to what's shown on this site, which is just since 2002.)
11:35 AM Jan 19th

Brock Hanke
Marisfan - Sorry about the understanding. I believe that Baseball Gauge carries Win Shares all the way back to 1876. I emphasized that it continues after 2001 because you can get WS for every player from 1876 on in the book Win Shares. Baseball Gauge lists those too, as well as WS from 2001 on.

Guy123 - You're right about Play Index not containing Defensive Innings at positions. But that's because it's in the player page WITHOUT needing play index. Just type in the player's name on the Home screen, and look through his page for the section titled "Standard Fielding." Not only are defensive innings there, but they are separated out by fielding position. An ENORMOUS contribution to research, at least in my experience.

Also, with Jones, you have to make some kind of discount for his not having a decline phase. People talk about that with regard to Curt Flood, but Flood played something like 16 years before he retired. Jones really DOES lack his decline phase, which makes any ranking look better than Jones likely was. I don't know the size of the discount - I've never seen anyone research it - but I do know it's there. In any case, what is becoming obvious is that the advanced metrics, if you are careful using them, do not show Jones as being an impossible defensive CF. That's reassuring, whether it impacts his HoF chances or not.
10:37 AM Jan 19th

MarisFan61
Gary: It's not just "post 2001" -- in fact, that's really the least of it since this site has it from 2002 and since.

(I wonder if maybe he meant to type 'post 1901'....)
7:21 PM Jan 18th

garywmaloney
Brock -- Thank you for unlocking the Win Shares for us, post-2001. I have written here on more than one occasion of how WONDERFUL the annual WS charts were, as provided in WIN SHARES DIGITAL UPDATE (2002 - own at all costs). And, belatedly, thank you for the Big Bad Baseball Annuals (I own every copy -- FFA '95!!)

Those charts had a couple eye-popping numbers --

-- In one year (96, 97 or 98), Andruw Jones had the highest number of Fielding Win Shares for any outfielder, ever -- over 9. (Mays had years in the high 8s.)

-- Montreal shortstop Orlando Cabrera's 2001 Fielding Win Share, over 13, was (if memory serves) the highest ever, for any player.

Worth mentioning, I thought.

5:41 PM Jan 18th

MarisFan61
huh.....for some reason this page isn't now enabling italics, bold, or underline in the comments. (It did earlier.)
11:39 AM Jan 18th

MarisFan61
Brock: Thank you for coming back with the link for the Win Share data.

We ought to add this: Your post actually seems to [i]understate[/i] what data are shown there! (I assume you actually know that; not sure why you put it as you did.)
Your post looks to me like it's saying that the site only gives the Win Share data from 2001 on. Looks to me like it gives it for forever, including Fielding -- which was exactly what I had asked about. ([i]This[/i] site gives it from 2002 on[/i].)
11:37 AM Jan 18th

Guy123
Correction on previous post: The best (top 10% or top 20%) post-1988 OFs average 14% higher TZ ratings than earlier players (I mistakenly said the reverse).
11:10 AM Jan 18th

Guy123
It is useful to separate Bill’s two arguments here: 1) that we cannot and should not directly compare defensive WAR ratings based on play-by-play data (like DRS) with pre-2003 defensive ratings (Total Zone), and 2) there is no remotely serious HOF case for Andruw Jones once you recognize the validity of point #1. On point #1 I don’t think there is any real disagreement: it seems clear that the range of ratings is much larger using PBP data, resulting in much higher ratings for top fielders. Future research may allow us to some day make a reasonably accurate conversion of TZ to DRS/UZR, allowing for valid comparisons of fielders (and this would be a great project for young saberists to undertake). But unless and until that happens, we cannot validly compare pre-2003 and post-2003 fielders using defensive WAR.

However, this important insight tells us relatively little about Andruw Jones’ HOF case. That is because DRS and TZ provide very similar assessments of his fielding for 2003 to 2012. So if we simply ignore the problematic PBP data, and rely on TZ for Andruw’s full career (as we do for earlier players), his fielding runs decline by only 16 runs, from 236 to 220, lowering his defensive WAR by about 1.6 wins and reducing his total WAR from 62.8 to 61.2. In other words, he remains – at first blush - a borderline HOF candidate.

Now, Bill is also right that defensive ratings have larger error bars than offensive ratings, so any player like Andruw whose case rests largely on defense should receive additional scrutiny. Let’s start with the fact that there are actually two versions of TZ, TZ1 used 1953-2007 and TZ2 used 2008-date. As Charles helpfully notes below, there is somewhat more variation in TZ2. But it’s a difference of degree, not kind (unlike the transition to PBP metrics)—I think we can still make valid comparisons over the 1988 line. I compared the top outfielders in these two eras, and the best (top 10% or top 20%) pre-1988 OF average 14% higher TZ ratings than more recent players. While it could be true that the range of fielding talent is higher in more recent decades, let’s be conservative and attribute all of this to the TZ methodology. That means we should boost the TZ for great CF before 1988 (like Blair or Mays) by 14% if we want to compare them to Andruw.

What does this mean for Andruw? Actually, not much. The TZ2 metric, using more information, is presumably more accurate than TZ1, so what we are really saying is that earlier CF are a bit better than TZ1 says. How does Jones stack up if we make this adjustment to TZ prior to 2008? Here are the top 10 CF in defWAR per 600 PA thru the end of Jones’ career (yes, innings would be a superior denominator, but B-Ref Play Index doesn’t provide that):
Blair 1.91
Piersall 1.59
Pettis 1.57
Jones 1.56
White 1.20
Maddox 1.12
Johnson 1.04
Flood 1.03
Lofton 1.02
Mays .99

By this measure, Andruw is one of the top 4 CF in history, with Blair the clear leader in this group. Is this a plausible set of ratings? I think so. Still, it’s fair to say that we should carefully examine any HOF case individually that relies on extreme defensive value. If we just look at raw putouts, did Jones make outs at an elite level? Yes. Did Jones “steal” plays for teammates? As best I can tell, Atlanta 2B, SS, LF, and RF did not see any decline in plays when Andruw arrived (Brian Jordan put up incredible ratings in RF playing with Jones). As far as I know, no one (including Bill) has offered any evidence that would lead us to doubt that Andruw Jones was one of the 4-5 best CF in history, or that he created something like 22 wins with his glove. I think the burden now shifts to the skeptics to make their case, if they can.

A word on Willie Mays. Does TZ understate Mays’ defense, even after adjusting for the gap between old and new TZ methods? Maybe, I have no idea. But I don’t see why that should change our view of Jones’ HOF candidacy. Mays' presence in this discussion seems mainly designed to say “defWAR says Jones was twice as good as Willie Mays, so please ignore this metric” (along with Jones’ Hall candidacy). This is clever, as it appeals to the assumption most of us older fans have that Willie was the greatest, but if TZ is wrong about Mays that tells us little about Jones’ case. (Note that being “twice as good as Mays” on these metrics actually means making one more catch every 12 games, not as wildly implausible as it may first sound.) Of if Mays is really “only” about the 10th best CF (far more likely), that also seems irrelevant to Jones’ case. So for now, let’s set Willie aside (although a separate discussion on why TZ may get Mays wrong could be fascinating).

Finally, does this all mean Andruw Jones belongs in the HOF? That really depends on your HOF standards. Personally I’m not sure, but I’d lean toward no. But it’s not true that his case hinges on his being “the best CF of all time.” To me, 60 wins/11 playoff teams/top 5 CF is not a materially different case than 60 wins/11 playoff teams/#1 CF. So I think Bill is mistaken when he simply dismisses Jones’ case with comments like these:

“Center Fielders who have to get into Hall of Fame before Andruw Jones is considered: Jim Edmonds, Bernie Williams, Fred Lynn, Wally Berger, Kenny Lofton, Jimmy Wynn, Cesar Cedeno, Johnny Damon, George Gore, Mike Donlin, Carlos Beltran, Andrew McCutchen, Paul Hines, Brett Butler.”

“The argument for Andruw Jones as a Hall of Famer rests on the belief that he was a BETTER center fielder than anyone or almost anyone before him. I don't see ANY evidence for that proposition, thus don't see any case for Andruw.”

Bill is right about the big picture here: we need to be very careful in making defensive comparisons over time, especially with modern play-by-play metrics, and we should approach defensive statistical outliers from any era with appropriate skepticism. But Andruw Jones is a poor choice to be the poster child for these ideas. So, to steal a line from 1964, we should go part of the way with BJ.
10:39 AM Jan 18th

KaiserD2
Michael Humphreys's DRA (Defensive Regression Analysis) did find Andruh Jones to be the best center fielder of all time, although he rated Willie Mays quite a close second, based on total runs saved. I think DRE does a better job of comparing people across eras than anything else that we have.

Using DRA for fielding and baseball-reference stats for offense (but without position adjustments), I show Andruh Jones with 4 seasons of 4 WAA or more. 4 such seasons is the most ambiguous figure, if you will, for the Hall of Fame. members of teh Hall with 4 such seasons includ Luke Appling, Robin Yount, Arky Vaughn, Tony Gwynn, Duke Snider, Frankie Frisch, Lou Boudreau, and Kiki Cuyler. They also include a lot of pitchers, led by Sandy Koufax. Players with 4 such seasons not in the Hall include Pete Rose, Minnie Minoso, Ken Boyer, Tony Oliva, Buddy Bell, Dave Parker, Willie Wilson, Bobby Grich, and Jesse Barfield. (This not an exhaustive list.) Also Don Mattingly, Manny ramirez, Rafael Palmeiro, Gary Sheffield, and Jason Giambi. Probably just about all of us could identify some people on that list whom we would like to see in the Hall, and quite a few whom we would not.

Now Chipper Jones, on the other hand, was a very good offensive player, playing in a very high-offense era, and a serious defensive liability. As a result, he had only 3 seasons of 4 WAA or more, one less than Andruh. I will regard it as a significant injustice if he is elected and Andruh is not.

This argument is not based on the idea that Andruh was by far the best center fielder ever, but on the overall value of his best seasons, hitting plus fielding. It does not suggest that he is an overwhelming choice for the Hall, but that he is, by historical standards, a reasonable one.

David K
8:38 AM Jan 18th

Brock Hanke
Riceman1974 - Thanks. That was useful to me. I didn't go through Win Shares or WAR just for results, because I know that the available statistics have changed over time. But, once you commented, I checked the book Win Shares for its formulas for the outfield. Bill uses the same formula from 1900 through to 2001, and the system, in Baseball Gauge, continues to the present. There are no stats used that aren't recorded in Speaker's time, much less Mays' and Flood's time. So there are no differences due to lack of data in Speaker's time. And sure enough, Jones does not come up as an outlier compared to Speaker. He comes up as comparable to Speaker in a shorter career.

And, since you mentioned Baseball Gauge and no one has given you a hard time for mentioning that on this site, I need to post this up for Marisfan61. I once said that there was a site that had Win Shares computed for every player for every year, all the way to the present. Marisfan asked where. I was worried about citing one site on another site's pages, so I didn't do that. But, now that it's out, here's how to get Win Shares for years after 2001 (it's not hard, just tricky and not obvious): Go to baseballgauge.com. On the home page, in the upper right hand corner, there is a little box labelled "Select Metric" or something like that, with the box filled in with "Wins Above Replacement." But there is a little slider bar at the end of the box, and if you click it, you get a selection of metrics, one of which is Win Shares. Select Win Shares. Now, Baseball Gauge will look the same in terms of design and presentation, but it will use Win Shares instead of WAR.

You can get a list of seasons easily, listing every player who played that season. The home screen makes it obvious how to do that. But getting an individual player is hard, unless you know the trick, which is similar to the slider bar trick to get Win Shares started. In the upper LEFT hand corner of the screen, you will see a little box labelled "Select Player" or something similar. There's no slider bar, but, if you just type in the name of the player you want, you get his whole career in Win Shares. If you just try the screen's obvious selection for players, you will get a list, starting with Babe Ruth, sorted by Win Shares. It will take you forever to find the player you're looking for. That little box in the upper left is the key to finding Win Shares for a player.

It also turns out that you can download this stuff, and it's not hard. Just run the cursor over the stuff you want, and use the Copy and Paste functions to put the data into a Word file or something like that. There may be a way to download the data while preserving columns, but I don't know it. I just cut and paste. In any case, the data presented for the player will include separate Batting and Fielding Win Shares, so you can get those defensive Win Shares, look up defensive innings on BB-Ref, and compute, using Bill's grading system in the book Win Shares if you want a grade.

Thanks, Riceman!
8:06 AM Jan 18th

sansho1
Regardless of where he ranks, this I know -- Andruw in his younger days was a joy to watch play the field. In person you could see his positioning, and so it wasn't a surprise to see him break left or right to snare low line drives to center. But on TV, where you don't see the entire field but you've watched enough to presume much of what you can't see, he retained the ability to shock with the angle he took as he arrived in the frame...not towards the infield, but sideways!
7:21 PM Jan 17th

BryanBM
Min 500 Games in CF through Age 30, how many innings it took the player to earn one fielding run according to baseball-reference

Juan Lagares 45.3
Andruw Jones 62.6
Jim Piersall 62.7
Devon White 67.4
Paul Blair 71.1
...
Willie Mays 1 Rfield per 116 innings

No idea on the accuracy of those numbers, but those are the numbers. Andruw is an outlier by a combination of empirically provable volume and subjective quality on par with past players even choosing an end-point that specifically favors Andruw.
6:09 PM Jan 17th

Riceman1974
Brock: Defensive Win Shares may provide such a method of comparison, as it essentially takes the same type of numbers and spits out a number at the end. According to The Baseball Gauge (which uses James's old Win Share methodology):

- Jones averaged 5.11 Defensive WS per/1,000 innings (87.1 DWS in 17038.2 inns).
- Speaker averaged 4.95 Defensive WS per/1,000 innings (117.3 DWS in 23698 inns).

But please note, Speaker was an excellent centerfielder through age 39, with 1-year of below average play at age 40, then he retired. Needless to say he was a tremendous hitter through age 39 as well.

Jones was a tremendous centerfielder through age 29, then fell off a cliff both defensively and offensively. He was out of the game by age 35. Through age 35, Speaker had 96.7 DWS in 18914 innings, or 5.11 WS per/1,000 innings, same as Jones.

I think it's safe to say Jones may have been among the 2 or 3 greatest outfielders the game has ever seen.

P.S: Mays averaged 4.29 DWS per/1,000 innings in his career.

3:56 PM Jan 17th

Brock Hanke
I'm afraid that this is a stupid question, but I've thought about it for a couple of days now, and it doesn't seem any stupider than when I first thought of it: If you're not happy with how far an outlier Andrew Jones' defense seems to be, and I am not happy about that, and you think that part of the problem is that we're using data in evaluating Jones that we just don't have for, say Tris Speaker (or Mays or Flood or DiMaggio) because it wasn't recorded at the time, then why not try to evaluate Jones by the method that you have to use to evaluate Speaker, INCLUDING that you do not use any data for Jones that is not available for Speaker, even if it is available for Jones? I don't have the database to try this, but someone here might. It's worth seeing what would happen, isn't it? There is a non-zero chance that it will clear up what is going on with Jones, which is what is needed, right?
1:58 PM Jan 17th

MarisFan61
(As I thought I emphasized loudly enough :-) .... I didn't mean "catcher ERA" per se, just something along such lines -- anything at all. What I'm criticizing is when a metric fails to include anything that address the effect on pitchers beyond those easily-measured things, which I don't think address much of the aspect.)
1:29 PM Jan 17th

CharlesSaeger
@MF61: Funny, I did actually mention that the Dodgers were better than the Yankees in walks and strikeouts every year Campanella was in the league:

"Campanella led in steals, caught stealing, strikeouts, walks, and non-strikeout outs; all the other stuff was small either way."

"Personally, I favor giving a bonus for walks and strikeouts rather than ERA (the Dodgers had more strikeouts and fewer unintentional walks than the Yankees every year Campanella was in the league)"

But blindly applying a crude tool like catcher ERA is making the matter worse; catcher ERA is so bad that 15 years ago nobody could figure out any predicative tendency out of it. It's part of why traditional fielding stats tend more towards zero: there's something of an inherent level of doubt, and saying that if I don't know, I'm not going to make a dramatic adjustment.
12:41 PM Jan 17th

those
I love this article. I'm biased because I've said a few times on here that people blindly treat WAR as proof that one player was better than another.

When Bill talked about Andruw on Twitter, Rob Neyer responded, "Huh. I feel like we spent years, six or seven or eight of them, believing (for various reasons) that he was maybe the best since Willie Mays. Were we all wrong? And how?"

Bill: If you thought he was the best since Willie Mays, the word "wrong" seems hardly adequate.

Rob: I'm saying that was, at the time, a consensus opinion. That he might well be the best since Mays, and that the numbers we had then said the same. If there's better information now, I want to see it.

Bill: The sabermetric community, to have impact on the baseball community, cannot treat its OWN opinions as consensus opinions. It is only the consensus of a minority. To impact the larger debate, we have to address the larger issues.

I think the last comment boils down what Bill is saying in this article -- that the advanced stats are one way of looking at things, and that we haven't reached an endpoint.

There's a guy on Twitter (I think his handle is Hot Stove Stats) who posts a lot of stuff like this. He posted something the other day listing Andrelton Simmons' defensive runs saved over his first six seasons, how it was twice as much as Ozzie Smith over his first six seasons.

The guy had no clue what he was talking about. Somebody questioned how the numbers were derived, and Stove Stats said baseball reference has found a way to compute them accurately across eras. In different posts, he used this data to show that "nobody has any idea just how good Simmons is."

There are numbers CONSUMERS, like Stove Stats and me, and there are legitimate people in the field.
11:09 AM Jan 17th

MarisFan61
Charles: I agree with this thing you said, and in fact I'm gonna use it against what you said: [smirk]

" if you’re not going to do a good job of it, you’re better off ignoring the matter."

I'm surprised that you (or anybody) thinks it's meaningful to come up with some defensive metric for catchers without trying to include something about the handling of the pitching staff (and I don't mean "pitch framing" which seems sometimes to be considered some big part of that).

The way this interchange started was that you expressed surprise about Berra coming out ahead of Campanella on some analysis of catching defense despite being behind him on all those things, none of which involve what I'm saying is key. IMO, what those factors add up to is 'not a real good job' -- do you really think they add up to a major portion of catching defense?? maybe you do; I don't. I also agree that the "WAR" system, as far as I can see, doesn't do a real good job of evaluating catching defense. I think Win Shares, broadly, does do quite a good job, although I can't say I understand exactly what's in there now. (I did with the original system.)
If a system ranks Berra as a better defender than Campanella, despite his being behind on those things you mentioned, we can only assume that other things were included in the analysis. Maybe they were good, maybe they weren't; but the basic thing was, I don't see why you'd think that those things that you mentioned should give some major idea of how catchers compare with each other or what a particular analytic system might show.

If you think there's nothing that can meaningfully take into account a catcher's influence on the pitchers -- anything about his catching defense besides those traditional easily-measurable things -- then, unless you don't want to resign yourself to "not doing a good job of it," I don't know why you would imagine any catching defense metric to be much good at all, or think you can gather much of anything from it in the first place.
11:08 AM Jan 17th

ksclacktc
“I dunno and I’m not gonna speculate” is often the best answer.

This may be the best, simplest post I've seen in a long time. Thank you Charles!
7:38 AM Jan 17th

CharlesSaeger
@MarisFan61: Oh, I didn’t misunderstand you. I just disagree. Personally, I favor giving a bonus for walks and strikeouts rather than ERA (the Dodgers had more strikeouts and fewer unintentional walks than the Yankees every year Campanella was in the league), but if you’re not going to do a good job of it, you’re better off ignoring the matter. Even taking one tenth of all runs saved by the pitching staff, which Pete Palmer does, can easily overwhelm the other, more reliable data. The best solution is likely a With or Without You number, weighted for recency, and regressed.

“I dunno and I’m not gonna speculate” is often the best answer.
7:35 AM Jan 17th

Riceman1974
Charles: Great work on the TZR data. And I agree Catchers ERA is overrated. I never knew Jones only played 400+ games after age 30, whereas Mays of course played a zillion or whatever. Of course Mays couldn't field like Mays in his late 30s and early 40s (although that he was still effective at center at that age is remarkable as you note), and this likely explains the Dwar difference between Jones-Mays.

It also shows us just how bad Jones became offensively after age 30, that even with his high-quality D in center he wasn't worth being in the lineup.
4:15 AM Jan 17th

MarisFan61
(sp -- dam, I wrote the wrong "affect" one of those times, I hate when that happens)
2:23 AM Jan 17th

MarisFan61
Charles: Now it's my turn to note that you mistook some of my evidently ambiguous wording! :-)

What I said:
Those catcher data from pre-1960 (as well as any others) don't include anything along the lines of "catcher E.R.A."?
If they don't, then they're even worse than I thought.
Not that they're great even with it, but....

What you answered:
I'm pretty sure they don't, mercifully. Catcher ERA is crap.

------------------

I agree! I mean not totally; I wouldn't say it's total crap, but yeah.

That's why I only said "anything along the lines of"!!

The stuff that you had mentioned, which Campanella would beat Yogi on, didn't include anything at all about the catcher's affect on the pitchers.
So, I was surprised at, well, a couple of things:

-- That you would be surprised that a catcher who's ahead of another catcher on all that other kind of stuff would fail to come out ahead on some 'large metric' -- because I always thought (and do think) that handling pitchers, and effect on the pitchers, is by far the #1 aspect of catching defense.
The things you mentioned, relatively speaking, are just 'noise.'

....in view of which.....

-- I'd be surprised if any 'large' defensive metric on catchers failed to include something on that.

If dWAR for years prior to 1960 (or ever) doesn't, then, as I said, it's even worse on catching defense than I ever imagined.

I don't disagree with you that "catcher E.R.A." is less than great.
But if there's nothing in the metric about the catcher's effect on pitchers, IMO the catching defense metric ISN'T EVEN WORTH USING, and it disqualifies the overall method (in this case, "WAR") from having any (ANY!!) meaningful assessment or rating or ranking of those catchers.
8:51 PM Jan 16th

CharlesSaeger
@Manushfan: One thing when thinking about Jones's fielding is that he doesn't really have much of a down phase. He only played 436 games after age 30 after playing 1,760 games through age 30. Fielding skill goes down fast in your thirties. Mays played until he was 42, which is more time for a down phase (though the fielding runs over on bb-ref.com don't show Mays dropping off afield until he was 36, when he dropped to being an average fielder for the last quarter of his career, which is actually a pretty remarkable record).
6:48 PM Jan 16th

CharlesSaeger
I figured the standard deviation of TZR for CF from 1974-1980, 1982-1983, 1986-1987 (I threw out the strike season, and I had ditched 1984-1985 for some data reasons that aren't all that important) and it's 9.91. For 1988-1992, it's 12.15. So it's higher, albeit not radically so.
6:41 PM Jan 16th

CharlesSaeger
I'm pretty sure they don't, mercifully. Catcher ERA is crap.
6:27 PM Jan 16th

Manushfan
So, for those of us in the back, then, the defensive part of the whole WAR analytic (izzat a word? no?) isn't as well developed or advanced as the hitting part? And the further back you go, the harder it gets to be as sure. That makes perfect sense to me.

Andruw twice the CF as Mays? Really that's a fail from the start. Just...no. Great D sure but..is he better than say Fred Lynn or Edmonds or whatever? you tell me.

And I'll go back to the tired Pie Traynor thing--Jaffy's book has him dinged as a minus whatever for his career defense, and looking at any survey of commentary from his peers etc-that's nuts. If you don't have real proof of that, then you might want to just leave that alone until better evidence comes along. Just saying.
5:44 PM Jan 16th

MarisFan61
Charles: Those catcher data from pre-1960 (as well as any others) don't include anything along the lines of "catcher E.R.A."?

If they don't, then they're even worse than I thought.
Not that they're great even with it, but....
5:23 PM Jan 16th

sansho1
Mine is not a question of whether anyone played shallow before -- I'm familiar with Speaker. It's 1) whether shallower positioning is in fact better, and 2) whether Andruw employed superior positioning to Mays, either directly or in comparison to their respective peers.
5:08 PM Jan 16th

CharlesSaeger
I should note that even the catchers, who you would think are cut and dried, are messed up before 1960 or so. On a Facebook group, someone noted that Berra ranked as a better fielder than Campanella in spite of Campanella's stellar OCS rate, and after a look at the detailed stats, I concluded that there was no way in hell that this ranking made any sense, unless you placed an undue weight on wild pitches, which was the only big gap in favor of Berra (Campanella led in steals, caught stealing, strikeouts, walks, and non-strikeout outs; all the other stuff was small either way.)
4:33 PM Jan 16th

bhalbleib
Sansho1, the problem with that is that playing shallow was in no way a new phenomenon, in fact here is the quote from the HOF on Tris Speaker's defense:

Speaker was known for playing a shallow center field, which helped him lead AL outfielders in assists three times, while his ability to cover ground on balls hit over his head helped him lead the league in putouts seven times.

“I still see more games lost by singles that drop just over the infield than a triple over the outfielder's head,” Speaker said. “I learned early that I could save more games by cutting off some of those singles than I would lose by having an occasional extra-base hit go over my head."
4:18 PM Jan 16th

MarisFan61
P.S. Charles - I just noticed that in an earlier post, you noted that there's also an additional 'break' around 1950 in how dWAR is done -- and that it's not clear exactly what's done for the pre-1950 game.

Hardly anybody who puts forth historic dWAR data ever makes sure to indicate any of this, and as I said, I have doubts that anywhere over a tiny fraction of them realize it.
The stuff is cited as evidence for all kinds of things as though (1) it's reliable, and (2) it's well comparable for different eras.
3:34 PM Jan 16th

MarisFan61
Charles: Thanks for mentioning that.
No, I didn't know that "Total Zone" had a different formula for pre-1988 than post, and I doubt that most others did.

So, yes indeed -- For at least some dWAR formulations, I guess, there's an additional discontinuity in the meaning of dWAR besides the 2003 watershed.
3:22 PM Jan 16th

sansho1
Mays was a bit before my time, but I watched 100+ games a year of Andruw's Braves career. I wonder if those who are dubious that Andruw could have been the more valuable defensive CF take into account his famously shallow defensive positioning. He may not have been faster, or got better jumps, or went back on deep balls more gracefully than Mays....but it is possible he employed better initial tactics, and put himself in position to make the most of his range. Maybe all those singles he cut off were of greater value than the doubles that got over or past him. The deep CFs in Mays' time would have prevented him from setting up as shallowly, but I see no reason to think someone else couldn't come up with a different and better fielding strategy that would redound to "value".
2:48 PM Jan 16th

CharlesSaeger
I'll have to check to see if DRS has a wider range than TZ-hit location when I get home. (I work from a spreadsheet for 2004-2009 to see how I can reverse-engineer modern ratings from traditional data.) There wasn't a big difference between them IIRC.

Y'all are aware that Total Zone has two different formulas, one 1988-present and another before 1988, right? Pre-1988, it didn't use any hit location data, and instead apportioned the hits based on the hitter's out distribution. Or, if 15% of Jim Rice's outs in play (much better recorded in PBP than hits) went to shortstop, Total Zone would apportion 15% of all Rice's hits to shortstop. This kind of works for infielders, not so much for outfielders.
2:46 PM Jan 16th

MWeddell
It's the comparison between Jones and all of the pre-2003 center fielders to which Bill objects. It might be that DRS is a reasonable estimate of player's defensive performance (at least for outfielders) but that it's a mismatch to compare dWAR using DRS to dWAR using TZR.
1:27 PM Jan 16th

MarisFan61
MWedd: I agree with what you're saying and I imagine just about everybody will.
But, re what you're saying at the end there, your "either/or" doesn't cover it.
(Pardon if I'm making more of it than you meant -- maybe you didn't mean it as covering it.)

What it says:
"Is it merely that DRS yields a wider range of defensive rating that TZR does? Or that DRS is wrong (as Bill previously used to argue about DRS’s evaluation of the effectiveness of shifts)?"

First of all, both of these are true.
And also, it's not just DRS that's 'wrong'; TZR also is, and probably more so.
12:23 PM Jan 16th

MWeddell
Copied and pasted from a Reader Post (since comments weren't turned on initially for this article). Sorry it covers some of the same material previously mentioned.

* * * * *

Some of us may read Bill’s article as being against WAR, but I don’t see it that way.

Bill also says that he doesn’t know how players rank until he ranks them. He consistently is against a one-off, ad hoc approach to player evaluation. If someone wants to argue that Roy Halladay’s pitching peak was as good as Sandy Koufax’s peak, Bill’s approach would be let’s collect a lot of data, decide how to best measure “peak,” look at our method’s results to see if they are reasonable and tweaking them perhaps, and then as the last step look to see what the results tells us.

WAR is a systematic way to evaluate players. It has the virtue of not being designed today merely to advance today’s desired conclusion.

I think the problems are
(1) Bill would say that when it comes to WAR’s CF defense evaluation, someone should have stopped to say “Hmmm, Andruw Jones is evaluated to be twice as good per inning as anyone else in history – is that reasonable?” If not, how do we fix the problem? If it’s a data limitation that can’t be resolved, how do we better communicate that one can’t compare defensive WAR of recent players to defensive WAR of older players?”
(2) While it may be expecting too much of most users, let’s not invest too much into any particular yardstick. How does Jones’ evaluation compare not just in bb-refWAR but also in fWAR and WARP and Win Shares?

* * * * *

This is why many of us are more skeptical of defensive valuations. Bill criticized one or more of us a few weeks ago on this very issue. When we compare Andruw Jones’ career offensive value across the multiple WAR and WAR-like metrics, we’re likely to find greater consensus than when we compare Jones’ defensive value. That’s why we often have a tendency to want to regress defensive value toward the mean on an ad hoc basis.

* * * * *

It’d be nice to know what baseball-reference.com’s WAR is doing wrong (assuming it is) regarding Jones’ defense. The most obvious problem is that from 2003 – now, bb-refWAR uses Baseball Info Solutions’ Defensive Runs Saved and before that (when DRS isn’t available) it uses Sean Smith’s Total Zone Rating.

Is it merely that DRS yields a wider range of defensive rating that TZR does? Or that DRS is wrong (as Bill previously used to argue about DRS’s evaluation of the effectiveness of shifts)? Today’s article doesn’t say. I’d guess that Bill isn’t particularly interested in picking apart someone else’s evaluation system. He’s more interested on developing his own systems, his own metrics, answering his own research questions.
12:11 PM Jan 16th

MarisFan61
I'm loving what I'm seeing in the discussion here. I don't think I've seen such broad and loud resonance here before on the issues of fielding WAR, and certainly nothing like it about the differences in what it is for the different eras.

Charles: Thanks for clarifying what you meant by those lower and upper limits on the players' "fielding runs."
It's funny ([b[very[/b] funny) -- that word, "might," which you intended to indicate just that you were tossing out some casual numbers, could also mean -- and I took it rather to mean -- that that's the supposed EXACT 'range,' especially since you didn't give round numbers.
i.e. "might," meaning that the actual number could be anywhere in that very specific range.

=======================

A little off the subject, but:

I've wondered if a big part of WAR's "problem" and a big part of the problem for the makers of the metric (all of us should have such problems) is how widely accepted it is. I know (or assume) that they do keep trying to improve it, and I hope they approach it with a fully open mind, as opposed to the defensiveness that we sometimes see. But I imagine they're somewhat hamstrung by the wide acceptance. Maybe it gives them a certain conservatism and extra protectiveness about it; and about this other thing that I sometimes criticize them for, I'd guess that it must make them extra reluctant to trumpet forth too loudly about how a lot of the public so misunderstands it and makes too much of it. I mean, if I had a product that millions of people were using and talking about and writing about and were assuming to be uber great, I wouldn't be too eager to try to make them realize that it wasn't all that, and that they shouldn't keep using it the way they do.

11:43 AM Jan 16th

raincheck
Something seems off here. Willie played in an era with more outs on balls in play. So even if they are exactly equal as fielders it seems unlikely that Andruw could have had MORE value in the field than Willie. Willie also played for a long ass time and made a lot more plays (he had over 2,000 more chances than Andruw).

For comparing accross eras, we would need to use the methodology pre and post 2003. That seems like an obvious starting point.
11:26 AM Jan 16th

smbakeresq
This is a very good piece IMO. I understand the defensive metrics however I believe in many cases the value of a players defense is overstated compared to observation and opinion of him. Your observation of him is always colored by your experiences of course, people who saw Willie Mays apply a "Willie Mays" standard for example.

I grew up watching Paul Blair run down balls in Baltimore. So for me base upon my experiences I have a Paul Blair standard, and Jones might or might be as good a defensive player as Blair was. Certainly I don't think Jones was clearly better than Blair, I don't really think can prove one way or the other. When Gary Maddox came along I heard the hype as it was and thought he is about the same as Paul Blair. Watching them both I couldn't say which one was better.

I don't think WAR tells us either.
11:15 AM Jan 16th

CharlesSaeger
ksclacktc: You're absolutely right. You have no idea how many times I look at the output from Statcast bandied about and ask myself, "What is this crap? What does this mean? Is a 104.1 m.p.h. exit speed good or bad?"
10:18 AM Jan 16th

CharlesSaeger
Wizardry is admirable for its perspiration, but not for its inspiration. His refusal to evaluate individual errors throws the infielder ratings into serious question. However, I prefer his outfielder ratings to TotalZone pre-hit location, which assigns hits based on the out distribution of the batter who hit them, which causes huge issues with corner outfielders since outs do not track hits in the air as outs tend not to be hard-hit balls. (It isn't a problem for infielders, however; outs on the ground track hits.)

Regardless, evaluating pre-PBP fielding stats is going to look like a lot of work with loads of sub-formulas. Because it is.
10:16 AM Jan 16th

ksclacktc
I don't expect to be heard or received by some of you. I would just ask that you consider the following:

EXPERTISE BREEDS OVERCONFIDENCE

The pendulum has swung, at one time we SABR people were the ones on the outside and not the establishment. We were the ones on outside throwing the shit inside. NOW, we are the establishment and throwing shit outside.

The level of overconfidence I see is downright scary.
10:04 AM Jan 16th

steve161
One analyst who explicitly uses era-specific methodologies to evaluate defense is Michael Humphreys in Wizardry. He details them in a fairly dense introductory chapter, which gives the reader the opportunity to decide what he thinks. Especially for earlier times, his approach involves a number of steps, each of which introduces a margin of error (which he does not attempt to calculate). Even if each step is 90% accurate, which is pretty optimistic, the cumulative error potential invites skepticism as to the results.

Before Statcast, the only way to judge how an outfielder reacts to a fly ball off the bat was to watch him live in the ballpark; television showed it only rarely. I'm old enough to have seen Willie Mays in his prime and for the longest time, I didn't think I'd ever see a centerfielder get as good a jump as he did. Then I saw Andruw Jones a few times--in Atlanta, in the conditions he was most familiar with--and he seemed comparable to Willie. I had a similar impression of Jackie Bradley Jr in a couple of games in St Louis last year.

But those are just hints. There's a whole bunch of CFs I've never seen live, given I attend about 10 games a year on average, so the evidence of my eyes is of severely limited value. Statcast will be a wonderful thing, but until someone invents a time machine, it won't tell me much about how close Byron Buxton comes to Willie Mays.
9:28 AM Jan 16th

CharlesSaeger
dWAR since 2003 onward is based on DRS. Honestly, I'm not sure that this is a good choice; when I compared it with UZR and TZ-hitlocation, it sometimes was the odd man out, especially for corner outfielders. But it goes something like this:

2003-present: DRS
1988-2002: TZ-Hit Location
1951?-1987: Total Zone (pre-Hit Location)
1871-1950?: ¯\_(?)_/¯ It's some range factor-based something that Sean never disclosed that I've seen.
9:02 AM Jan 16th

3for3
Copied from my post in the Reader Posts section:

I find it interesting that Bill is assuming all of the Andruw lovers are blinded by his WAR.

My own view is that his war (low 60s) is one that automatically gets him in the discussion. Meaning, you have to at least think about him.

The 2 main questions beyond that are:

1. Is his WAR an accurate depiction of his skills? For Bill, the answer seems to be no. For me, I can see his point about the magnitude of Jones' greatness in the field.

2. Did he have impact beyond his WAR? Here is where Bill is doing exactly what he is accusing the WAR people of. I think he had substantial bordering on enormous impact. He was a key player on an historically great team. His team averaged close to 100 wins in his first 8 years. We all know the story of how poorly the Braves did in the 3 tiered playoff system. Put them in the 80s, or better yet the 50s, and they'd be in the World Series most seasons. Yes, they had other great players on those teams. But he was great then too.

As a side note, I also like to give a boost (from WAR) for players with one memorable skill. Sure, Dwight Evans was a great player because he did everything well. But, someone who is historically great at a central skill, like Jones, Ozzie, Brock, etc, DO get a boost, at least from me.
8:38 AM Jan 16th

Mike137
MarisFan wrote: "Defensive WAR" is a completely different thing for very recent times than for the past.

That is a very important point. If Method A is applicable for all data and Method B is only applicable for a certain range of data, one should not switch between the two methods. The results from Method A should be given for all data and those from Method B should only be given for the data to which it is applicable.

BBref seems to do that for some of fielding stats. So Rtot (total zone fielding runs above average) is given for all years and Rdrs (defensive runs saved above average) is given only for 2003 on. But then they mash it all together in dWAR. Just another example of "we don't know how to do this right, so we will just do it the way it is convenient".
8:00 AM Jan 16th

CharlesSaeger
@MarisFan61: You're putting too much stock in those ranges. They were pulled out of my ass for sake of example. Hence the word "might" in them.

But yeah, they are pulled more towards the center, as we're hedging our bets. We don't know whether Tris Speaker or Heinie Wagner did more to save hits for the Red Sox, so we assign them pretty much the same. We can do things like look at doubles and triples allowed to help split between the infield and the outfield, and whether Speaker or Wagner was the one who got to more outs, but we're hindered.
6:38 AM Jan 16th

Riceman1974
I think pre-2003 Bball Ref Dwar uses Sean Smith's "JAARF" (Just Another Adjusted Range Factor) merhod in calculating defensive runs saved. FanGraphs uses "UZR" (Ultimate Zone Rating) for their pre-2003 defensive runs saved. Both sites use Baseball Info Solutions's (BIS) DRS method post-2003. JAARF and UZR are generally more conservative than BIS's DRS, in some cases strikingly so as Bill mentioned above. I am not qualified to take apart each method, but it seems to me that most of our problems with WAR in general originates with BIS's DRS run values. These values significantly increase the Dwar of post-2003 players, leading to conclusions as Bill mentions, that Jones was twice the centerfielder as Mays, which seems ludicrous.
6:37 AM Jan 16th

wovenstrap
A trap that I find myself falling into, speaking as a 40-something man who's been reading Bill's work since 1982, is that there's a difficult tendency to "see" the stats and not the game itself. That is, when you are watching someone at the plate, you don't see what is in front of you and "see" instead your prior understanding of him as a free swinger, pull hitter, whatever. As Bill points out endlessly, the truth is always more complicated than our mental constructs for it. I think that many of the Andruw boosters are doing something like this, they're not watching with fresh eyes, they're substituting the statistic or the widespread understanding of Andruw as an unbeatable fielder for their own perceptions.

In 1946, before he was famous, George Orwell wrote an essay with the title "In Front of Your Nose." Here was the key section of the concluding paragraph: "To see what is in front of one's nose needs a constant struggle. One thing that helps toward it is to keep a diary, or, at any rate, to keep some kind of record of one's opinions about important events. Otherwise, when some particularly absurd belief is exploded by events, one may simply forget that one ever held it." It's hard to see what is right in front of you.

One of the things that amazed me so much about the Ballantine run of Abstracts was that Bill would re-formulate his player rankings system every year. To abandon a system that seemed to work, just to test your own premises. This was very inspiring. The WAR genuflectors need to attack this issue more in that spirit.
5:27 AM Jan 16th

MarisFan61
Charles: That's good and interesting information, and I think that by only a slight extension we can roughly infer this from it: Fielders from longer ago tend to be calculated by "fielding runs" as being more toward the center than they really were.
i.e. Good fielders probably usually were better than shown, and poor fielders probably usually worse than shown.
("Roughly" meaning not just only approximately but also not uniformly.)
I think that's probably implied by what you're saying, and I think it makes sense.

BUT, I think you take it too far in stating any kinds of specific figures for the possible upper and lower limits for any long-ago player.
(BTW I'm purposely avoiding the term "range" that you used -- like Mays's likely "range" of fielding runs -- because of the word's meaning with regard to fielding. It's the right word for what we're talking about, but it could lead to misunderstandings.)

I don't think it's possible to give anything like such precise estimates of the possible upper and lower limits for that -- especially upper.

I also think you're giving way too much credit to the system with regard to recent accuracy, and that you're mistaken in thinking that the figures for Andruw (or any current or recent player) are so reliable.
You say:
"Jones played in the 21st century and got 220 fielding runs.....we're pretty sure of that. His range might be 210-230 runs for his career."

We're NOT pretty sure of that, and IMO the possible limits are much more widely spread.

I say all of the above on a theoretical basis.
I would add this extra thing on a seat-of-the-pants basis:

There's a very loud thing in your figures that seems to beg for criticism of how you're seeing the issue. You're saying that the extreme upper possible limit of Willie Mays's "fielding runs" is barely more than the amount that Andruw Jones "pretty surely" had.

That doesn't meet the sniff test.

......even more so, when we consider that Mays had a much longer career and therefore had greater 'mass.' He had over 40% more innings in the outfield. So, what your reasoning is saying is that not only was Andruw a better fielder than Willie; he was so much better of a fielder that Willie, even in >40% more innings, didn't accumulate as many "fielding runs." I would say confidently that this fails on a simple Emperor's Clothes basis.

You put forth a very good basic framework, but you extend it into specifics that just don't work. I suspect that the basis of the error is very simple: You are way overestimating the accuracy and reliability of the system. You are way underestimating the degree to which old-time players' actual performance can differ from the calculation, and, although I concede that the method is far more accurate for more-recent times, you're way overestimating the accuracy and reliability of what the method shows even for our times.
12:01 AM Jan 16th

CharlesSaeger
The standard deviation of fielding runs is bigger now since we have better measures of fielding since about 1988 or so. As we go back farther, we're forced to rely more on the things we can measure better: team DER, fielding percentage, outfield assists, range factor. Team DER to reflect individual range means that players cluster around the team rate. Think of, say, Tris Speaker's fielding runs as being the midpoint of a range. BB-ref.com has his career for all positions (which is about 99.5% in CF) as 92 runs better than average. The range on that might really be 22 to 172.

Willie Mays had 176 fielding runs. After WW2, we have more data, though not perfect; the range might be 116-236 runs.

Now, Jones played in the 21st century and got 220 fielding runs. By contrast, we're pretty sure of that. His range might be 210-230 runs for his career.
10:15 PM Jan 15th

sayhey
A more basic argument against Jones--which you may or may not agree with--is that HOF players should continue to build their résumés through their 30s. They have to fill out the scenery, as Bill has put it. They don't have to win MVPs or Cy Youngs, but, at a minimum, they have to continue producing at a level that makes for a reasonable addendum to their peak years. Even guys like Vlad and Puckett, who both had short careers, they were productive players for the first half of their 30s. Koufax gets a pass.

Andruw Jones was essentially finished at 30. If he'd been been Mike Trout in his 20s, I might give him a pass too. But he wasn't.
8:48 PM Jan 15th

MarisFan61
This is the first fleshing-out that I've seen of a thing that I've noted countless times on Reader Posts, which has received virtually no resonance:

"Defensive WAR" is a completely different thing for very recent times than for the past.
......which has a couple of different important implications, both of which seem usually to be ignored when fielding-WAR figures are cited, and which I've suspected most people who cite them aren't even aware of:

-- The stuff is probably much less accurate and reliable for anything prior to about 2005 (give or take) than to the years since then -- so, to whatever extent those data may be quite good for recent times, don't assume anything about earlier times.

and.....

-- Beware of relying on such data for comparisons of players from different eras, especially comparing recent players to long-ago players.

I think the chronological dividing line for when the more-detailed fielding information starts being applied is somewhere in the first few years of this century.
I've never been able to get a clear straight answer on whether there are also multiple other prior chronological dividing lines on how the data are determined, or whether the method is pretty uniform for everything prior to the last decade-and-a-half or so. I suspect it's the former, and that it's quite helter-skelter -- which would mean that hardly anyone who uses the data really knows what has gone into them for any period or year and how much reliance should be placed on them.
8:24 PM Jan 15th

Andruw of Center Field

COMMENTS (64 Comments, most recent shown first)

Leave a comment

Report inappropriate comment


Type of Abuse:
Comments: