Who was best at playing the field?

After the passing of Orioles’ legendary manager Earl Weaver last month, I found myself spending a good amount of time scrolling through the rosters of those dynasty clubs from Baltimore in the late ’60s and early ’70s. It shouldn’t surprise you that the name of Mark Belanger kept showing up, considering he spent 16 years with the O’s while playing some of the most defensively magnificent shortstop that baseball has ever seen.

Belanger has always fascinated me, though I never got the chance to actually see him play—unfortunately, as his career was winding down in the early 1980s, my obsession with the game of baseball was only just beginning. As a result, my understanding of Belanger’s contribution to the game is limited to the more sophisticated modern estimates of defensive value.

The most widely respected of these retrospective defensive metrics, Baseball Reference’s “Rfield” measurement (which uses the very popular Total Zone to measure range) credits Belanger with the best defensive season for a shortstop in the history of the game:

Greatest defensive seasons since 1920

Name Year Pos. Defensive runs
Gary Carter 1983 C 26.9
Albert Pujols 2007 1B 31.0
Frankie Frisch 1927 2B 37.0
Mark Belanger 1975 SS 35.1
Brooks Robinson 1968 3B 32.6
Darin Erstad 2002 CF 38.7
Barry Bonds 1989 LF 36.9
Ichiro Suzuki 2004 RF 30.0

Yet I doubt that Belanger’s name recognition reflects that amazing feat. When standing side by side with the other positions’ single-season leaders, his stature seems almost dwarf-like in comparison. In fact, aside from Belanger, it would seem that Darin Erstad is the only other player on this list who is neither currently in the Hall of Fame nor otherwise bound for it. But certainly Erstad’s defensive reputation was similarly untouchable.

So, as a tribute to all the potentially forgotten Erstads and Belangers in the darker corners of baseball history, I wanted to dedicate this morning to shedding light upon more of these elite but unsung defensive superstars.

Of course, I want to do this without championing the ever-lurking “fluke” defensive season. Even the best of our modern defensive metrics can be occasionally seduced into granting undue credit or blame to a fielder in smaller samples. One solution to this problem might be to instead celebrate the greatest defensive peaks in history, rather than individual seasons.

Since three seasons of defensive statistics are generally considered as reliable as a full season of offense, I compiled a list of the best consecutive three year defensive peaks as determined by Baseball Reference, using the following criteria:

{exp:list_maker}Each three-year peak required at least 80 percent of the player’s team’s games played at the position in question.
For catchers I made the requirement just 75 percent of their team’s games.
Players were ranked by their defensive runs per 150 games from their three-year period to account for the differences in the length of season from the different eras.
I also limited this search to the live ball era. {/exp:list_maker}

Greatest three year defensive peaks since 1920

Name Years Pos. 3 year Defensive runs/150 3 Year Defensive runs
Ivan Rodriguez 1996-1998 C 21.8 65.0
Albert Pujols 2006-2008 1B 22.0 66.0
Orlando Hudson 2003-2005 2B 23.1 62.0
Mark Belanger 1975-1977 SS 27.6 82.6
Brooks Robinson 1967-1969 3B 27.8 88.3
Barry Bonds 1989-1991 LF 26.9 83.0
Andruw Jones 1998-2000 CF 29.9 96.0
Jesse Barfield 1985-1987 RF 21.8 68.7

Belanger hangs onto the top spot for shortstops with his remarkable stretch of defensive excellence from 1975-1977. Albert Pujols, Brooks Robinson, and Barry Bonds also maintain defensive supremacy at their respective positions, but Gary Carter, Frankie Frisch, Darin Erstad, and Ichiro all prove they were not up to the task of sustaining their defensive wizardry over the course of multiple seasons.

That Andruw Jones replaces Erstad at center field should surprise no one, especially not those who witnessed Andruw’s playing days during the turn of the century. Jones’ elite range and strong arm during those years in Atlanta places him on another plane of existence far above other center fielders in history. Paul Blair’s second place showing is almost 30 runs behind Jones. Devon White, Jim Piersall, and even Garry Maddox also make the top 10, but Jones stands out as a man amongst boys:

Greatest defensive peaks, center field

# Name Years Pos. 3 year defensive runs/150 3 year defensive runs
1 Andruw Jones 1998-2000 CF 29.88 96.0
2 Andruw Jones 1999-2001 CF 27.06 87.3
3 Paul Blair 1968-1970 CF 24.73 67.6
4 Paul Blair 1967-1969 CF 22.50 64.2
5 Devon White 1991-1993 CF 22.48 67.9
6 Andruw Jones 2000-2002 CF 22.31 70.8
7 Devon White 1992-1994 CF 21.99 57.9
8 Jim Piersall 1955-1957 CF 21.56 65.1
9 Paul Blair 1969-1971 CF 21.40 59.5
10 Garry Maddox 1977-1979 CF 19.93 57.4

(Erstad, for what it’s worth, did not meet the criteria to qualify here. From 1999-2001 Erstad amassed 81 defensive runs, but played a majority of his games at a different position in each season. When running the same query for “outfielder” excellence, rather than specifically LF/CF/RF, Erstad’s 28.8 runs/150 ranks him just below Jones and Barry Bonds. Willie Wilson scores a few points by that method as well.)

At catcher, Ivan Rodriguez dethroning Gary Carter on the merits of his fantastic run from ’96 to ’98 isn’t exactly a shocking upset either. But to Carter’s credit, it should be noted that the two together can lay claim to eight of the top 10 defensive peaks for catchers. Only Charles Johnson (16.0) and Jim Sundberg (15.5) came even remotely close to breaking up this monopoly:

Greatest defensive peaks, catcher

# Name Years Pos. 3 year defensive runs/150 3 year defensive runs
1 Ivan Rodriguez 1996-1998 C 22.78 65.0
2 Ivan Rodriguez 1997-1999 C 21.28 60.0
3 Gary Carter 1981-1983 C 20.21 53.5
4 Ivan Rodriguez 1995-1997 C 20.19 56.0
5 Gary Carter 1982-1984 C 16.50 48.4
6 Ivan Rodriguez 1994-1996 C 16.13 40.0
7 Charles Johnson 1996-1998 C 16.04 40.0
8 Gary Carter 1979-1981 C 15.74 40.6
9 Jim Sundberg 1976-1978 C 15.45 45.0
10 Gary Carter 1983-1985 C 14.90 42.7

At second base, I will admit the appearance of Orlando Hudson just about knocked me out of my seat. I would never have guessed that Hudson was actually on that sort of level at the keystone. I did find it interesting that the perpetually underrated Chase Utley narrowly loses out to O-Dog with his terrific run from 2006-2008, despite accumulating a superior 67 defensive runs overall during that span. Frankie Frisch, incidentally, wasn’t too far behind. Among these three, the battle for greatest defensive peak at second base was extremely close:

Greatest defensive peaks, second base

# Name Years Pos. 3 year defensive runs/150 3 year defensive runs
1 Orlando Hudson 2003-2005 2B 23.13 62.0
2 Chase Utley 2006-2008 2B 22.48 67.0
3 Frankie Frisch 1926-1928 2B 21.48 60.0
4 Chase Utley 2005-2007 2B 20.92 59.0
5 Chase Utley 2007-2009 2B 20.52 61.0
6 Orlando Hudson 2004-2006 2B 20.36 57.0
7 Lonny Frey 1938-1940 2B 18.61 49.0
8 Hughie Critz 1932-1934 2B 18.17 51.0
9 Lonny Frey 1939-1941 2B 17.90 50.0
10 Bobby Grich 1973-1975 2B 17.45 54.9

(Nap Lajoie, interestingly, would have beaten them all with his startlingly awesome run from 1906-1908 had we included the dead ball era in this competition.)

At first base however, no one is even close to challenging Albert Pujols. In Pujols’ best three-year span he was worth a total of 66 defensive runs above average. His closest challengers were George Scott and Keith Hernandez, both of whom totaled under 40 defensive runs in their three-year period. John Olerud ekes out an appearance at the very bottom of the list:

Greatest defensive peaks, first base

# Name Years Pos. 3 year defensive runs/150 3 year defensive runs
1 Albert Pujols 2006-2008 1B 22.45 66.0
2 Albert Pujols 2007-2009 1B 21.66 66.0
3 Albert Pujols 2005-2007 1B 20.11 61.0
4 Albert Pujols 2008-2010 1B 14.02 43.0
5 George Scott 1972-1974 1B 13.48 39.9
6 George Scott 1971-1973 1B 13.22 38.7
7 Keith Hernandez 1981-1983 1B 12.98 34.6
8 Keith Hernandez 1983-1985 1B 12.92 39.1
9 Albert Pujols 2004-2006 1B 12.31 37.0
10 John Olerud 1998-2000 1B 12.09 38.3

In right field, Jesse Barfield is a name I have not thought of in a long time. I remember Barfield as a respected, above-average outfielder with a strong arm that made up for a merely adequate bat, but I would not have expected him to have ranked as the greatest defensive right fielder of all time. His peak really ran five seasons, from 1985-1989, and it establishes him as one of the best outfielders to play the game rather conclusively. Al Kaline, Tony Armas and Brian Jordan all make an appearance in the top 10, but none ever put Barfield in any danger. Ichiro, incidentally, fell to eighth best, despite his record-breaking season in 2004:

Greatest defensive peaks, right field

# Name Years Pos. 3 year defensive runs/150 3 year defensive runs
1 Jesse Barfield 1985-1987 RF 23.05 68.7
2 Jesse Barfield 1986-1988 RF 22.85 65.8
3 Jesse Barfield 1987-1989 RF 21.28 60.3
4 Al Kaline 1956-1958 RF 20.07 56.6
5 Tony Armas 1980-1982 RF 20.03 52.6
6 Brian Jordan 1999-2001 RF 18.86 53.3
7 Jesse Barfield 1988-1990 RF 17.98 50.7
8 Ichiro Suzuki 2003-2005 RF 16.42 52.0
9 Bobby Abreu 1998-2000 RF 15.98 47.3
10 Roberto Clemente 1966-1968 RF 15.91 45.5

In left, Barry Bonds is not a name we often associate with elite defensive ability. In a way it’s a shame—his otherworldly achievements at the plate (along with a multiverse of controversy) have overshadowed the fact that Bonds was also possibly the best defensive left fielder of all time. He’s followed by some other big names in Carl Yastrzemski, Rickey Henderson and Pete Rose, but we are also blessed with a pair of lesser known defensive superstars in Rocky Colavito and Warren Cromartie:

Greatest defensive peaks, left field

# Name Years Pos. 3 year defensive runs/150 3 year defensive runs
1 Barry Bonds 1989-1991 LF 27.36 83.0
2 Barry Bonds 1988-1990 LF 24.78 72.7
3 Carl Yastrzemski 1966-1968 LF 22.34 70.3
4 Carl Yastrzemski 1967-1969 LF 20.01 60.7
5 Rickey Henderson 1980-1982 LF 18.68 50.3
6 Barry Bonds 1990-1992 LF 17.67 51.6
7 Pete Rose 1972-1974 LF 16.61 52.7
8 Rickey Henderson 1981-1983 LF 16.36 42.0
9 Rocky Colavito 1961-1963 LF 15.00 45.2
10 Warren Cromartie 1977-1979 LF 14.74 46.1

One of the more remarkable things I noticed in making these lists is that at one point in time the Baltimore Orioles had the greatest shortstop playing beside the greatest third baseman of all time, though their peaks didn’t necessarily overlap. (Paul Blair, owner of the second best center field peak, was also in Baltimore during this same period.)

Prior to running these queries, I fully expected Brooks Robinson to take the prize at third base without much opposition. That Clete Boyer and Buddy Bell show up with the second and third best third base peaks was a bit unnerving. Neither was close to usurping Brooks’ amazing run from 1967 to 1969, but I do admire their gall. Robin Ventura and Graig Nettles are predictable showings, but it’s also nice to see Evan Longoria make an appearance here at No. 5 overall. He is certainly the youngest active player to appear in a top 10 for any of the positions here today:

Greatest defensive peaks, third base

# Name Years Pos. 3 year defensive runs/150 3 year defensive runs
1 Brooks Robinson 1967-1969 3B 27.83 88.3
2 Clete Boyer 1961-1963 3B 25.32 74.1
3 Buddy Bell 1981-1983 3B 22.03 58.0
4 Brooks Robinson 1966-1968 3B 21.54 68.5
5 Evan Longoria 2009-2011 3B 21.18 61.0
6 Robin Ventura 1998-2000 3B 21.12 64.5
7 Graig Nettles 1970-1972 3B 19.74 60.8
8 Brooks Robinson 1968-1970 3B 19.34 61.1
9 Graig Nettles 1971-1973 3B 18.94 58.7
10 Brooks Robinson 1971-1973 3B 18.41 56.7

At shortstop, Belanger is similarly far above the competition, as you might have expected. One combination or another of Belanger’s stretch of seasons from 1973 to 1978 accounts for the top four defensive peaks for shortstops in the live ball era. Only Ozzie Guillen’s three-year span from 1986 to 1988 puts an end to Belanger’s dominance. And just in case you were wondering how the Wizard fared, Ozzie Smith’s best three-year span rates only as 10th best per Baseball Reference:

Greatest defensive peaks, shortstop

# Name Years Pos. 3 year defensive runs/150 3 year defensive runs
1 Mark Belanger 1975-1977 SS 27.72 82.6
2 Mark Belanger 1973-1975 SS 26.58 81.7
3 Mark Belanger 1974-1976 SS 25.89 79.4
4 Mark Belanger 1976-1978 SS 25.84 73.9
5 Rey Sanchez 1999-2001 SS 24.64 69.8
6 Dick Bartell 1936-1938 SS 23.31 62.0
7 Ozzie Guillen 1986-1988 SS 22.82 70.3
8 Marty Marion 1942-1944 SS 21.84 61.0
9 Travis Jackson 1927-1929 SS 21.68 61.0
10 Ozzie Smith 1988-1990 SS 21.50 63.5

The entire list of three-year peaks, including the top 50 at each position, can be viewed as a google doc here.

Note about the differences in eras

There is something we ought to consider when using Total Zone.

Baseball Reference essentially uses four separate methods of estimating defensive value depending on the data available from a particular season. For recent seasons Defensive Runs Saved (DRS) is used. From 1989 to 2002 a version of Total Zone including zoning buckets used from project scoresheet is used. Retrosheet’s play-by-play data is used from 1950-1988, and a very basic form of Total Zone is used prior to that.

We’ve witnessed Pujols and Hudson at their peaks just in the last decade, while Pudge and Bonds wowed us in the ’90’s, and Jesse Barfield only a few years prior in 1987. That leaves only Belanger and Robinson to represent the ’60s and ’70s, while the entire first century of baseball is left wholly unrepresented in the all-star team of defensive peaks.

This might lead one to wonder: Are we seeing more defensive superstars as time goes on?

If we look at the number of players who managed a defensive runs/150 of at least 10+ runs in each season since 1920 requiring an appearance in at least 80 percent of their team’s games, the bias favoring recent eras becomes apparent:

image

Obviously teams did not just suddenly begin adding defensive wizards to their rosters at mid-century. In 1950, once Retrosheet’s play-by-play data becomes available, it is clear that our measurements of defensive performance become much more confident and much more willing to hand out excellent defensive ratings.

Would players like Frankie Frisch and Nap Lajoie put our modern defensive superstars like Chase Utley and Evan Longoria to shame with their glovework? Impossible to say, based on the limitations of the data available from that time period.

This is a rather bittersweet fact to accept. On one hand, we have to come to terms with the fact that there will never be a way to effectively compare fielders from different eras. But it also allows us the opportunity to engage in some wonderful arguments about which players were the greatest at fielding their position in the history of the game. If we did have these magical numbers assigned to each player’s value, we wouldn’t have much to talk about, now would we?


42 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Roger Munter
11 years ago

Very nice piece. Like James I’m a bit gobsmacked by Orlando Hudson’s presence here. And I think Pujols comes off better than I thought.  As for omissions, I think the most shocking to me is Dwight Evans who I saw quite a bit of during my college years. I’m not surprised particularly that Jesse Barfield is the #1 RF, but I’m quite surprised that Evans couldn’t make even a top 10 appearance.

Bill
11 years ago

Put in an adjustment factor to normalize the pre-1950 era.

That will give you an interesting type of look for an apples to apples comparison.

Jacob Rolling Rothberg
11 years ago

I will go to my grave convinced that Roberto Alomar was the greatest defensive second baseman ever. Having seen Hudson firsthand at his peak, I can say it’s absolutely no shock to see him graded so highly. That being said, neither he, nor anyone else, was Alomar’s equal.

Steve
11 years ago

What is causing Yadier Molina to not have higher Total Zone scores? I believe he peaks at 16. I know he’s considered one of the better defensive catchers.

James Gentile
11 years ago

Good catch on Evans. His best three year peak was about 22.1 runs/150 in 74-76 (this is quick math from my phone mind you), which would place him third below Barfield. But It looks as though he was disqualified for playing only 115 games in 1975, falling short of the 80% requirement. That he still amassed so many defensive runs in so few games is amazing!

He definitely deserved an honorable mention.

Tom
11 years ago

I’ll be honest… I only read the first couple of paragraphs to this article and then skimmed it to find Johnny Bench’s name. It was no where to be found, not even in the comments! I am no sabermetric expert but, I know about baseball after coaching it for years. When healthy, Johnny Bench was the best defensive catcher of all time. Ask any one who really knows the sport and the position. He changed the way the position was played. It looks like you were using just sabermetrics as the measure. I know you and most others would admit to the many weaknesses of this approach as your last paragraph implies—but to not even give a mention to those obvious names left off your list was a mistake.

Tom
11 years ago

By the way…most baseball people will tell you that the most important defensive positions are up the middle—C, SS, 2B, CF. A huge but little mentioned factor for the Big Red Machine’s success in the 70’s was their up the middle defense. I will put Bench, Concepcion, Morgan,and Cesar Geronimo up against any other group that ever played.

Rally
11 years ago

I think Yadier could match or surpass Carter if he could go back 30 years in time.  Carter in 1983 allowed 86 steals, and threw out 75 (47%).

Last year Yadier allowed 38, threw out 35 (48%).  The 35 runners thrown out is a career high.

Runners just don’t run as much as they did back in the 80’s, so a great throwing catcher has less impact.

David
11 years ago

Definitely some surprising names when you look at the entire list – Strawberry, Gedman, Jody Davis, Conine.  Other names you would expect to find included more than they are – Mazeroski (2), Grich (1), Dwight Evans (2).

Dennis Bedard
11 years ago

Belanger’s longevity is a tribute strictly to his defensive skills.  He was a horrible hitter, always skirting with the Mendoza line.  Kudos to Earl Weaver for putting him in the lineup every day and realizing his contribution was not reflected in the common statistical data of that era.  Belanger and Weaver had something in common:  both were change smokers.

Guy
11 years ago

Nice work pulling this together.  The graph tells two important stories, I think:

1) it would be best to exclude pre-1950 players, or do a separate analysis for them.  The pre-1950 TZ system regresses performance so heavily, the results just aren’t comparable at all. (And it suggests the handful of pre-1950 players who managed to make your lists may have been defensive monsters, though pre-1950 TZ is a crude instrument.)

2) After 1950, it appears the number of great fielders is steady.  But that’s not what we should see:  over time, as talent rises, variance should decline—as we know it has on offense. Plus, K rates have steadily risen, making it harder to put up huge fielding totals.  I think this decline is masked by the sequence of metrics used:  TZ in 1950-1988, TZ+ for 1989-2002, and DRA for 2003+.  The problem is that each successive change in metric tends to increase variance, i.e. increase the proportion of fielders rated as great (or terrible).  This is a function purely of methodology, apart from any real changes in talent.  As it happens, these metric shifts create an illusion of stability, where we should see a steady decline. What this means in practice is that there should almost certainly be more players from 1950-1989 on your lists, and fewer 1990+ players.

One way to correct for this problem would be to use Rally’s TZ data for all years, which he posted here at THT:  http://www.hardballtimes.com/main/article/measuring-defense-for-players-back-to-1956-part-2.  Then at least you are doing an apples-to-apples comparison.  I suspect the resulting lists would favor pre-1990 players a bit more.  Just one example:  the peaks for Pujols and Keith Hernandez are about equal, rather than Pujols appearing to be twice as valuable. (And I wonder whether TZ captures Hernandez’s handling of bunts?)

Guy
11 years ago

James, you might also be interested in looking at Michael Humphreys 2012 SABR presentation, that looked at how fielding metrics (including TZ) systematically underestimate great fielders.  He compares WOWY results to the metrics for great fielders, and finds the average WOWY rating is more than twice the TZ rating.  He argues that Buddy Bell and Keith Hernandez, for example, had defensive value about twice as large as their TZ rating.

James Gentile
11 years ago

Thanks for the link, Guy. And I’ll definitely have to get ahold of the WOWY/TZ comparison. This is all very fascinating stuff. Thanks again!

John Northey
11 years ago

Barfield is no shock to anyone who was in Toronto in the mid-80’s.  His arm was a legend around here.  You wanted a runner on 3rd with less than two out just to see him have a chance to throw the guy out.  He was the one guy you’d pay to watch play defense.  A shame his career ended as quickly as it did.

Guy
11 years ago

James:  You’re welcome.  I think you’ll find Rally’s data very interesting.  Maybe someone can bring it up to date (in ends with 2007 season). I looked at a few 2B, and Hudson’s peak is matched by Frank White and exceeded by Mazeroski.

Rally:  Very interesting point about Yadier vs. Carter.  Does TZ give any credit to a catcher like Yadier who suppresses the running game (SBAs are down about 25% since 1983, but 55% fewer runners are going vs. Carter).

Mike Magie
11 years ago

Dang no Frank White.

Marc Schneider
11 years ago

Your point about underepresenting players from earlier eras is well taken.  When I see players’ gloves from the 10s/20/30s, I always wonder, how the hell could they catch anything?  Even during 40s and 50s, the equipment and field conditions were nothing like what they are now.  This is not to take anything away from current players-some of the difference probably represents increased athleticism, but defense is something that seemingly wouldn’t be affected as much by players being stronger and faster.  If you can catch, you can catch.  It’s a shame, as you mention, that it’s impossible to fairly evaluate defensive players from long ago.  Was Marty Marion really as good as reputed?  Was he even better given the differences in equipment and field conditions?

Greg Simons
11 years ago

Ozzie Smith’s best three-year stretch defensively was during his age-33 to age-35 seasons?  That seems…odd.

Sure, he probably learned how to position himself better with experience, but I’d think he would have paired that knowledge with his athleticism in such a way that his peak would have been several years earlier.

studes
11 years ago

James, love that graph.  Deeply awesome.

James Gentile
11 years ago

RE Ozzie Smith: I thought the same thing. Incidentally, Baseball Prospectus’s FRAA has Smith’s 1979-82 periods as his strongest, which would make more sense. 

FRAA also gives the best three year peak for a shortstop to Rey Sanchez 1999-2001, and doesn’t rank Belanger nearly as high.

Greg Simons
11 years ago

I grew up watching The Wizard, and he’s my favorite player ever, so any metric that doesn’t show him as the greatest defensive shortstop ever must be flawed.  wink

mark
11 years ago

James

nice angle.  Was the 80% filter based on 130 + games per season or 390 over the three years?

Mark

James Gentile
11 years ago

Thanks Mark.

I required 80% from each of the three seasons separately. So, yes, 130 games for seasons after 1960 assuming there were no canceled games, excluding strike seasons.

SB McManus
11 years ago

If the objective is to filter out sampling error over a consecutive 3 year period, why wouldn’t you use a threshold of, say, 80% of games played over the three year period?  That should get rid of players where there isn’t enough data to make a reliable 3-year statistical observation but wouldn’t accidentally eliminate someone who plays 100%-79%-100% of a team’s games.

Maybe it’s just an academic observation and doesn’t matter in practice.

James Gentile
11 years ago

80% of games over the three year peak is probably the better way to go, yes. The only advantage of having requisites at each individual season would be query speed. It should be no big deal to write in the 3 year requirement. I’ll report back any significant changes.

Bill Rubinstein
11 years ago

I agree with some of the comments above. There has been a constant narrowing of the gap between the best hitters, league average hitters, and replacement level hitters, and it is difficult to believe that this hasn’t occurred with fielders, rather than the opposite, as this study seems to show. Another factor is the vast increase in the number of strikeouts, meaning that there are fewer fielder’s chances than in the past. It is not clear how this study takes this factor into account. But it is an interesting study!

gdc
11 years ago

It would be more surprising to see Belanger overlap Brooks than the opposite-after all, there are only so many balls hit in the infield and presumably the peak Robby cut off a few chances for his SS.
I would guess that the old timers didn’t dive for balls as much between the injury risk and the general two-handedness of fielding.  But if they grew up seeing people dive for the ball and get recognition, the competitive nature would make them try the same.

Duane Sjoberg
11 years ago

I think your increase in defensive superstars is simply due to expansion of baseball from traditional 8 teams per league starting in the early 60’s…simply more potentials to choose from.

Ron
11 years ago

Kaline came down from a higher league to play rightfield. His game was legendary from the beginning; coming in, going back, east-west, he caught everything. Not only a cannon for an arm, the ‘most accurate’ throwing arm to second, third, and home I ever saw. After the 23 assist season iirc in 1958, the league mostly stopped running on him. That didn’t stop Kaline from making the throw to the bag or home plate with runner on anyways. I used to marvel and love that, runner tags up takes a step or two, and the ball is a perfect strike from right field. smart move runner.

Jon Roegele
11 years ago

Nice research, James. I don’t think I truly appreciated how good a defender Barfield was in that stretch. Always saw his arm, but he obviously had more going for him than that. Thanks for the article!

chuck
11 years ago

surprised too that omar vizquel isnt on the list.

Michael A. Humphreys
11 years ago

James,

Nice article. Just wanted to let you know there is another method for valuing fielders throughout MLB history: Defensive Regression Analysis (“DRA”). When Guy mentioned “DRA” being used post-2002, I think he meant “DRS” based on proprietary BIS batted ball data.

The real DRA is the system used in the 2011 book, published by Oxford University Press, Wizardry: Baseball’s All-Time Greatest Fielders Revealed. Tim Marchman, reviewing in the Wall Street Journal, kindly called it “the book of the season.”

DRA is the first and still only fully disclosed comprehensive statistical model of baseball defense (pitching and fielding) that can be applied to whatever open-source public data exists. I fully disclosed the methodology for all the years reported in the book (1893-2009).

Since the book came out, I’ve done some more research, which as Guy mentioned, I presented at last year’s SABR conference. What Guy didn’t mention is that the presentation showed why and how DRA is more accurate than the Baseball-Reference system, certainly before 1989, possibly through 2002, and maybe even now, given the persistent biases in the way that Retrosheet and BIS batted ball data has been collected.

DRA also resolves some of the anomalies noticed by your readers. Ozzie Smith’s peak occurred all before he hurt his rotator cuff circa 1985. His high ratings in later years are run savings by Pendleton misattributed to him.

Fielders before 1950 actually ‘saved’ more runs because the variance in ability was much higher.

Wizardry has a unique way of adjusting historical ratings for the growth in talent pools over the years, so that the best fielders are not all from before 1920. The system “Talent Pool Adjusted Runs” (“TPAR”) is explained in Chapter Four, which also includes plots of top fielder career runs savings over history, so the trends can be seen.

If you or anybody else reading this thread is in New York City the first Friday of baseball season, April 5th, I’ll be doing a presentation of DRA at The Harvard Club of New York City. All are welcome.

Guy
11 years ago

Michael:  Yes, I meant to type “DRS” for post-2002 RField.

BTW, do you have any plan to publish your WOWY results for the 46 players you analyzed?  That would be very interesting to see.

Michael A. Humphreys
11 years ago

Guy,

I’m actually thinking of expanding the WOWY study!

If you have more questions about DRA, please don’t hesitate to email me at

Moeball
11 years ago

Michael:

I’ve thought it has been interesting watching the evolution of fielding analysis over the years. It seemed to me the first big breakthrough was the LWTS from Pete Palmer in Hidden Game of Baseball and the various Total Baseball encyclopedias since then.  But there was a lot of concern about over-crediting some players – Lajoie’s putouts totals, Frisch (1927 in particular), Mazeroski maybe getting double-credited on DPs?

So we have FRAA now, TZ, DRA, etc. One question I have, though – it was mentioned above about the possible narrowing of the gap between best and worst fielders, which would make sense since there has been an overall narrowing of the talent spectrum in recent decades (not just in baseball, but in all sports pretty much). Do you think we are headed to a point where the best fielders are only 5-10 runs better than average and the worst fielders are also only about 5-10 runs per season worse than average? If the range becomes that narrow, it would seem to me there would be a huge impact on lineup selection for managers. If the worst fielders are only 10-20 runs (at most) below the best fielders, then wouldn’t it just about always be better for the team to have the better hitter playing the position as opposed to the “good glove, little hit” player? Seems to me the Mark Belanger, Orlando Hudson type player would get phased out because they are just costing teams too much offensively and, as good as their gloves are, they cannot begin to make up that difference defensively. Mark Belanger is valuable to your team if he is 30+ runs per season better than average defensively; if he’s only 10-15 runs better than average, he’s not making up for what he’s costing you offensively.

Michael A. Humphreys
11 years ago

Moeball,

First to the question at hand—

While I do believe the ‘spread’ between the best and worst fielders is narrowing, open-source data and some fairly simple math suggests that the very best fielders, during their three-to-five-year peak, are probably more like 15-20 runs better than average. The worst fielders are about as bad in the opposite direction.

The spread is narrowing, however. I think that is due to (1) more proprietary batted ball information being bought or compiled by teams, so fielders can be better positioned, and (2) the phenomenal physical condition of current players. Mark Belanger was a heavy smoker.

* * *

Now some important background regarding other fielding systems—

Pete Palmer’s Offensive Linear Weights was by far the single most useful sabermetric formula ever created. And it was based on sound statistical methodologies.

However, his Fielding Linear Weights were literally just made up—there was absolutely no statistical modeling behind them. And the outfielder formulas were never disclosed. And as you point out (and as I may have mentioned in Wizardry), each double play by Maz gave him .4runs for the assist, .2 runs for the putout, and another .4runs or so for the DP. In fact, the extra out on a GDP is worth something between a walk and a hit, or about .4 runs total. The version of Fielding Linear Weights that was around when Maz got elected to the HoF gave him something like an extra 200 runs of career value he didn’t really have.

Lajoie was actually genuinely dominant in his time. See my essay about him in Wizardry.

Frisch’s 1927 was just a fluke year. I think that was the record for assists in a year, maybe just at second, but maybe for any position. There was no evidence of a ground ball pitching staff. Just a fluke.

FRAA has never been disclosed. The vague descriptions of the system never indicated a reliance on statistical modeling, but more like ad hoc adjustments based on ideas published by Charles Saeger.

There is strong evidence that BP has copied from DRA in the past. In 2007 I wrote an article here at THT in which I disclosed for the first time just the outfielder formulas for DRA. Shortly thereafter, the FRAA ratings for the sample of outfielders in my article went from have only a .70 correlation with DRA to a .87 correlation. (Infielder ratings did not converge.) Perhaps even more significantly, the ‘spread’ of ratings (std) went from 9 runs/season to 12 runs/season (it was 13 for DRA). Now that DRA has been fully disclosed in Wizardry I have no way of knowing the extent to which BP has copied it.

And as I said before TZ is not fully disclosed and, as explained in Wizardry and again in my 2012 SABR presentation, biased in current years and both biased and systematically understated for years before we have play-by-play data (circa 1950, though the data is very spotty before 1957 and circa 1969-70.)

Again, DRA is the only comprehensive open-source statistical model of baseball defense (fielding and pitching). It is the first and only model that reveals the statistically significant relationships between and among objective, publicly available pitching and fielding statistics and actual team runs allowed.

DRA is designed to adapt (and is adapted to) whatever public, objective information is available.

Furthermore, I’m sure if whatever undisclosed systems there are were fully disclosed, you would find that the DRA formulas are actually _simpler_ than the formulas for the other systems.

Again, if you or any other readers have questions, feel free to email me at

Thanks!

John Northey
11 years ago

When it comes to spreads between fielders I wonder if the spread isn’t more a moving in-and-out range. By that I mean that it grows for a period, then compresses.  The reason for this thinking is teams feeling a need for more offense leading to sub-optimal defense (such as Hubie Brooks at SS for the Expos in the mid-80’s) then shifting the other way where teams have super defense and horrid offense (Expos shifting to Luis Rivera who had a 66 OPS+ after moving Brooks to RF). 

Look at Detroit and how they gave up a lot defensively at 3B vs teams like the Jays and Rays who have amazing defense at 3B – I’d have to think the spread is pretty big there this year vs many other seasons.  Different philosophies can lead to big spreads one would think.

Michael A. Humphreys
11 years ago

John,

There is a very interesting historical example of a spread growing and then narrowing that I mentioned in Wizardry, pp. 129-130.

“A funny thing happened when integration arrived: the standard deviation [in runs saved at third base] exploded.

“I believe there were two forces at work. Slow [white] outfielders who could not compete with [black and Latin American] outfielders, but who could hit, were tried at third, where they *under*-performed, thus creating *negative* variance. Someone like Killebrew might have played more in the outfield in the 1950s. And slow middle infielders who could not compete with [black and Latin American] middle infielders were tried at third, where they *out*-performed, thus creating *positive* variance. Brooks Robinson might have played short if he had played when Lou Boudreaux did. And Lou Boudreaux, who admitted he wasn’t the quickest on his feet, would probably have played third if he had played in the 1970s.”

Moeball
11 years ago

Michael, John – thanks for the comments. I guess I’m just fascinated how various managers view lineup structure depending as much on views about defensive performance as offense. Back in ‘86, for example, Mets manager Dave Johnson gave considerable playing time at SS to rookie Kevin Mitchell. Clearly Johnson’s thinking was that it was more important to have a guy at the position who could actually hit and that he couldn’t be costing the team that many runs defensively even as a poor fielder.

At any rate, one other question I have is about the various “Zone Rating” methods to evaluate defensive performance. In theory, just being able to measure what % of time a fielder turns balls in play into outs seems like it would be a pretty accurate measurement of performance.  No need to know whether there is a groundball staff or flyball staff, righty pitchers or lefties – it wouldn’t matter how often the ball was hit into the player’s area, what would matter is what % of time these opportunities were converted into outs. Yet I don’t see consistent agreement on this. Is it because of variations in official scorers and possible bias in how they are scoring plays? Are there arguments as to whether this ball was really in Player “A”‘s zone or should it have been allocated to Player “B”?

Michael A. Humphreys
11 years ago

Moeball,

Funny you mentioned Zone Rating. You should read my THT articles from 2007 “Ghosts in the Outfield” about the shocking inconsistency in “zone data,” which I call “batted ball data” (because it gives trajectory and location data about each batted ball) at least from 2003-05.

After that article came out, others started doing studies showing all sorts of biases. See my discussion in Wizardry, pp. 66-70.

See in particular the following blog thread:

http://www.insidethebook.com/ee/index.php/site/article/suzr_v_buzr/

And this one about biases in collecting batted ball data (start with comment #89):
http://www.insidethebook.com/ee/index.php/site/comments/how_much_random_variation_in_fielding_on_bip/

Guy
11 years ago

Michael:  Interesting observation about third base in the 1960s.  I suspect that increasing speed on the part of baserunners is part of this story as well.  The level of offense at SS (vs. other positions) fell dramatically in the 1960s, I suspect because a lot of older SSs just couldn’t throw out a new wave of faster hitters, and teams were so desperate to get a good glove there they didn’t care if you could hit (see: Dal Maxvill, Mark Belanger).  It appears that teams tried to keep decent bats at 3B for a few years longer—presumably the price of bad defense there is not quite as high—but then starting in the late 60s offensive levels at 3B dropped too.

Michael A. Humphreys
11 years ago

Guy,

That is very interesting. So what was needed at short was not what we commonly think of as speed/range but arm strength and accuracy. Did people at the time notice this, the way that people noticed in the 1980s that Ripken could play short because of his strong arm, notwithstanding his lack of foot speed?

By the way, Cal Ripken is _wildly_ overrated at B-R.com due to some systematic biases in the system and the batted ball data from 1989-99. On a career basis, I have him as only mildly above average, about +50 defensive runs.