The Expansion Era Ballot Part IV

November 20, 2013

The Last Five Candidates

 

8.  Dave Parker

                Like Steve Garvey only more so, Dave Parker balances above the line that defines where Hall of Fame selection should reasonably begin, with a career won-lost record of 300-201:

                 

Batting

Fielding

Total

Winning

YEAR

Team

AB

HR

RBI

AVG

OBA

SLG

OPS

W

L

W

L

Won

Lost

 Pct

1973

Pirates

139

4

14

.288

.308

.453

.761

3

2

1

1

4

3

.549

1974

Pirates

220

4

29

.282

.322

.409

.731

5

4

1

2

6

6

.523

1975

Pirates

558

25

101

.308

.357

.541

.898

19

4

6

2

25

7

.783

1976

Pirates

537

13

90

.313

.349

.475

.824

16

6

4

4

20

10

.680

1977

Pirates

637

21

88

.338

.397

.531

.927

21

4

7

1

28

5

.854

1978

Pirates

581

30

117

.334

.394

.585

.979

23

-1

4

4

27

3

.897

1979

Pirates

622

25

94

.310

.380

.526

.906

21

3

5

3

26

6

.806

1980

Pirates

518

17

79

.295

.327

.458

.785

13

8

4

3

17

12

.591

1981

Pirates

240

9

48

.258

.287

.454

.742

5

5

1

3

6

7

.471

1982

Pirates

244

6

29

.270

.330

.447

.776

5

6

1

2

7

7

.471

1983

Pirates

552

12

69

.279

.311

.411

.722

11

13

4

3

15

16

.483

1984

Reds

607

16

94

.285

.328

.410

.738

13

12

3

4

16

16

.494

1985

Reds

635

34

125

.312

.365

.551

.916

19

8

5

3

24

11

.697

1986

Reds

637

31

116

.273

.330

.477

.807

15

12

3

5

18

17

.523

1987

Reds

589

26

97

.253

.311

.433

.744

11

15

4

3

15

18

.441

1988

A's

377

12

55

.257

.319

.406

.725

9

6

1

3

10

9

.524

1989

A's

553

22

97

.264

.308

.432

.741

12

13

0

4

12

17

.407

1990

Brewers

610

21

92

.289

.330

.451

.781

16

11

0

4

16

15

.510

1991

Angels

466

11

56

.232

.279

.358

.638

8

13

0

3

8

16

.320

1991

BlueJays

36

0

3

.333

.400

.444

.844

1

0

0

0

1

1

.639

   

502

11

59

.239

.288

.365

.653

 

 

 

 

 

 

 
                 

246

144

54

57

300

201

.599

 

 

                Parker’s career winning percentage, .599, is just a little bit higher than Garvey’s, and in a slightly longer career.    Basically, they’re about the same. 

                I had an exchange of e-mails/posts/etc. with Tom Tango on the subject of Dave Parker’s Hall of Fame standing, and I wanted to review that with you because there is something of substance at issue.    It began with Tom posting this letter in the "Hey, Bill" section:

They may have their hearts in the right place, but this new HOF ballot is abysmal in its representation. First, Dwight Evans easily outclasses Dave Parker (matches him on peak, and destroys him on career). But then you have other outfielders from Parker's era that have as good or better case than he does (Bobby Bonds, Chet Lemon, Jose Cruz, Cesar Cedeno, Fred Lynn, Dale Murphy but still on normal ballot, Willie Wilson, George Foster, Amos Otis, Ken Singleton, depending on C/OF Brian Downing). Given your love for Otis, I'd like to see your comparison of Otis to Parker, especially with Win-Loss Shares.

 

                To which I replied. . .well, never mind with that; I wandered off topic.   The issue I am trying to reach here is, "How good was Dave Parker, really?"   I agree that Dwight Evans was a better player than Dave Parker, but I’m talking about Parker here, not Evans.   My methods don’t show Amos Otis as being on the same level as Parker, and I got interested in the question of why my methods are (apparently) much kinder to Parker than Tango’s methods are.    Tom’s next post on the subject was:

I agree that the "pile-up" of candidates, if the selection process is something like "vote for at most 4 of these 12" will simply split the vote, and no one gets in. But, if they are allowed to vote for as many as they want (or, similarly, go around the room, and vote on each player one at a time), then that's fine. I don't think people appreciate that the selection method controls the whole flow. Amos Otis is essentially tied with Parker in WAR. In my "Individualized Won-Loss" system, I have Parker at 83-42, and Otis at 79-36. At a 3x conversion rate, an 80-40 record converts to 240-120 Win Shares/Loss Shares.

 

                To which I responded:

 

Yes, precisely: People simply do not understand that the selection method controls the whole flow. Perfectly stated. The Hall of Fame itself historically has not understood this, and the public (mostly) cannot comprehend that the reason that one marginal candidate gets in and another is left out has nothing at all to do with how they are thought of by the baseball community, but rather, with the vicissitudes of the selection process. Some candidates find a gate wide open; others find a horde of good candidates trying to squeeze through a small opening.

 

<​i style="mso-bidi-font-style:normal">We appear to have several methodological disparities here, so let me see if I can nudge us toward understanding. You are attributing to Dave Parker only 125 games (Decisions) in his career, 83 wins and 42 losses.

Well, every 54 outs is a game, isn't it? If we assume that the responsibility for wins and losses is evenly shared between offense and defense (defined to include pitching), then a player has responsibility for one win or one loss with every 54 batting and baserunning outs (essentially. It might be 53.3 or something.)  Parker made 7,055 batting and baserunning outs in his career. That's 130 games--just based on his hitting and base running. I don't know what portion of the responsibility for run prevention you want to attribute to the fielders, but I use a figure around 12 or 13%. It varies. There is a spread around 50-38-12; the percentage for fielding used to be much higher.

If the hitters on a team are responsible for 50% of the wins and losses and the fielders for 12%, then there is a 50-12 ratio of batting to fielding "games responsible for" in the ordinary case. ..obviously some players have more defensive responsibility than others. But Parker was a right fielder, and right field is in the middle of the defensive spectrum, so he would appear to have a fairly normal ratio of offensive to fielding responsibility. This would suggest that he should be assigned responsibility for another 31 games as a fielder     { (130 * 12)/ 50 = 31}. That makes 161 games for which Parker is responsible. I actually have him with 167 games (501 Win and Loss Shares), probably because (a) the normal ratio of outs to games is slightly less than 54 to 1, and (b) the "fielders defensive responsibility percentage" on his teams may have been slightly higher than 12%.

Anyway, you have assigned him responsibility for only 125 games--not even enough to cover his performance as a batter. How do you figure? I guess what I really don't understand is, if you assign players responsibility for wins and losses at that low a rate, how can you account for all 162 games in a season?

 

 

                Tom responded (previously unpublished):

 

I should have noted that I removed his "below replacement level" seasons.  Those had a record of 7-24, and in my view, I'd rather zero those out, when interpreting his overall career. Anyway, including those seasons, he's 90-66.  I give out 57% "decision" to nonpitchers, and 43% to pitchers (which is also how many WAR I give out).  This is Parker's page: http://www.tangotiger.net/wonloss/index5.php?retroid=parkd001

 

                The page in question shows that Tango rates Parker’s 1977, 1978 and 1979 seasons as being even better than I have rated them, but that he rates several other Parker seasons far worse than I would rate them.    Dave Parker’s 1987 season, for example; playing for Cincinnati, Parker hit .253 with 26 homers, 97 RBI, 7-for-10 stolen bases, thirteen outfield assists, and finished 13th in the National League in putouts in the outfield.   His On Base Percentage was just .311, his OPS just .744 against a league norm of .732.   I have the season scored at 15-18, a .441 winning percentage.    Tom has it scored at 2-8, which would be about 6-24 in terms of Win Shares and Loss Shares.  

                We now understand why I had 167 "decisions" charged to Parker (501 Win Shares and Loss Shares), while Tom had 125.    It is because (1) he was ignoring the seasons in which Parker had no value, and (2) he is attributing 7% of decision to fielding, whereas I am at about 12%, actually 11% in the case of Parker.

                As to the policy of ignoring seasons in which a player has no value. . .that’s a good policy, and I do that as well, for example, when I figure WSAR (Win Shares Above Replacement).    I treat negative seasons as zero-value seasons, on the logic that, if a team chooses to play the player when he cannot perform well, that tells us more about the team than it does about the player.  

                But if you add those seasons back in, Tango has Parker at 90-66, which is 270 Win Shares, 198 Loss; I have him at 300-201.   I still see Parker as being substantially better than Tom does.   Why?

                There are two issues:

                1)   Of the 50% of the game that we allocate to pitching and fielding, what percentage goes to fielding?   I use a number that varies, but averages about 12%; Tango uses 7%.  That’s a huge difference.

                2)  How bad was Parker’s defense?

                On the issue of how much "space" should be allocated to fielding in the split between pitching and fielding, there is as of yet no compelling logic or no convincing research on this issue.    It is, then, an issue not of what we know to be true, but of what we believe to be true.   Tango—if I am following his logic correctly, which is always a risk—Tango believes that 86% of pitching and fielding is pitching.   I place it about 76%.

                It is an issue of what we believe, rather than what we know, but. . .I don’t see how 86% can be the right answer.    John Dewan’s Defensive Runs Saved claims that the Kansas City Royals’ fielders in 2013 were 195 runs better than the Philadelphia Phillies fielders, the Royals being +92 runs, as a team, and the Phillies -103.    It seems improbable that this estimate is accurate, but let’s work with it for a second.    The 2013 standard deviation of runs saved, on a team level, is 51.34, according to John’s data.   The standard deviation of runs allowed by teams (2013) is 69.96.

                  If John’s estimates are correct, then, 54% of the difference between teams in terms of runs scored is accounted for by fielding, which means that no more than 46% can be accounted for by pitching, unless teams which have good fielding tend to have bad pitching and vice versa.    Some of the spread is accounted for by park effects; when you take out the park effects you’ll have less than 46% for pitching.

                OK, well. .. John’s numbers are obviously too large, then—but Fangraphs’ numbers aren’t appreciably smaller.   Dewan says that Andrelton the Great saved the Braves 41 runs; Fangraphs says 32.   Dewan says that Gerardo Parra saved the Diamondbacks 41 runs; Fangraphs says "No no; it was only 27."    Dewan says that Manny Machado saved the Orioles 35 runs this year; Fangraphs says it was only 34.   Baseball Reference has a third set of numbers, which are in the same range.   But my point is this:   in order to get the percentage of pitching-and-fielding combined that is "fielding" down to 12%, you have to assume that the number of runs saved by a good fielder is much, much smaller than any of those numbers.  With my number—12%--there is no room for good fielders saving 25 runs a year.   If the number is 7%, there is probably no room for good fielders saving 15 runs a year; you probably have to stop about 10.

                If my logic is correct, which. ..you know; it’s always a risk.     But if I understand this problem, then you can’t argue simultaneously that one fielder is 50 runs better than another fielder at the same position, but that fielding is only 7% of the game.   Either the fielding percentage has to be larger, or the number of runs saved by a good fielder has to be smaller; one or the other.    Even the number that I use—12%--is far too small to be logically consistent with the Runs Saved estimates of Dewan, Fangraphs, or Baseball Reference.   

                OK, the second issue here is, "How bad was Dave Parker’s defense?"    Fangraphs estimates that Dave Parker was 218 runs better than average as a hitter, but believes that he was 159 runs worse than average as a fielder.   In other words, they believe that Parker’s fielding was so bad that it offset 70% of his value as a hitter.   This 159 runs includes, I believe, a positional adjustment; in other words, they are saying (I think) not that Parker in 1984 was thirteen runs worse than an average right fielder, but that he was some number of runs worse than an average right fielder, and, when you include an adjustment for the fact that he was a right fielder, the total is -13.

                But my method includes a positional adjustment, as well, so that isn’t the difference between us.  There is actually a good deal that Fangraphs’ method and Win Shares/Loss Shares agree about, with regard to Dave Parker’s fielding.   Although Parker won three Gold Gloves in his career, both of us regard his fielding as a negative element of his overall game. .. a sub-.500 element.   Both of us agree that Parker’s best defensive season was 1977, and his second-best defensive season was 1975.   Our evaluations of his 1977 defensive work appear to be almost identical.   They have him at +17.2 runs for that season, and I have him with a defensive won-lost record of 7-1, which means that he was positive by about 18 runs.   Both of us agree that Parker’s 1977 Gold Glove was justified by his performance, but that his 1978 and 1979 Gold Gloves were not.   Both of us agree that Parker’s fielding performance over the second half of his career was very poor, and both of us agree that his best defensive season after 1980 was 1985.    I credit Parker with a defensive winning percentage of .608 through 1980, but .362 after 1980; Fangraphs has him at -13 runs through 1980, but -146 runs after 1980.

                But Fangraphs believes that Parker’s defensive work was so bad that it offsets 70% of his offensive value.    My estimate is 3%.    I credit Parker with a defensive won-lost record of 54-57, a .451winning percentage as a fielder.  

                                As was true of the other issue, I don’t believe that there is any compelling logic or any completely convincing research in this area.   It is, then, not an issue of what we know to be true, but of what we believe to be true.   So I will ask you then, what you believe:  do you believe that Dave Parker was such a dreadful defensive player that his defense offsets 70% of his work with the bat, or do you believe that he was a good fielder as a young player, a very poor fielder over the second half of his career, but on balance it’s not really much of a factor?   I’m not telling you what you should believe, but. . .my answer works a lot better for me.

(After finishing the article, I see that this is poorly stated.   The 3% vs 70% distinction is when measuring the player against an AVERAGE, not VALUE.    I list Parker at +102 Win Shares on offense, -3 on defense.  Fangraphs lists him at +218 runs as a hitter, -159 as a fielder.   But these measure the player against an average, and the "average" line is a mythical point of reference with no real-life consequences, whereas the replacement level is an actual floor with meaningful consequences.   Measured against the replacement level, Parker doesn’t offset anything like 70% of his value with his defense, in Tom’s system, and has positive value {above replacement level} in my system.)

                Among the 12 Hall of Fame candidates being considered here, I would list Parker 6th.

 

 

9.  Dan Quisenberry

                My good friend Joe Posnanski makes a passionate argument for Dan Quisenberry as a Hall of Fame candidate, the essence of which, I think, is Bruce Sutter.    Sutter and Quisenberry were both born in 1953, so they are exact contemporaries.  

                Sutter pitched 661 games in his career; Quisenberry, 674.

                Sutter pitched 1042 innings his career; Quisenberry, 1043.

                Sutter had a career ERA of 2.83; Quisenberry, 2.76.

                Sutter was an estimated 123 runs better-than-league in his career, park adjusted; Quisenberry was 148 runs better. 

                Sutter’s winning percentage was .489; Quisenberry’s was .549.  

                Sutter was the Closer for one team that won the World Series, and Quisenberry was the Closer for one team that won the World Series.   Quisenberry, however, was the Closer for four teams that made it to post-season play; Sutter, only the one. 

                The only significant stat in which Sutter has any advantage is Saves, and that is not a large advantage, 300 career Saves to 244.    Dan Quisenberry saved 45 games one year, which was a major league record at the time, and then saved 44 the next year.

                On what rational basis, then, can Bruce Sutter be put into the Hall of Fame, and Dan Quisenberry excluded?

                I will give you the answer that I think Joe would give you.   It’s not a rational basis; it’s image.   Sutter had more the image of a star, based on:

                1)  His famous splitter,

                2)  His strikeout rate,

                3)  His phenomenal first halves in 1978 and 1979 (he collapsed both times in the second half of the year),

                4)  Quisenberry’s lack of a fastball, and

                5)  Dan Quisenberry’s self-effacing manner.

                Quisenberry was most famous, while active, for his wit.  (Asked what happened to his sinker when it didn’t sink, Quisenberry said that "I still get a ground ball; it just takes a little longer to hit the ground.   Reggie Jackson hit one against me last week that is still burrowing its way to St. Louis.")

                And a sixth thing:   Sutter won a Cy Young Award; Quisenberry should have won at least one and perhaps two, but the voters shafted him.    But based on performance, Joe argues. ..if Sutter is in, Quisenberry should be in.

                I am. . how should I say. . .disinclined to argue with Joe Posnanski, and disinclined to argue against Dan Quisenberry.   My method does not show Quisenberry as a Hall of Famer, but then, my method doesn’t show any reliever as a Hall of Famer, except possibly Mariano. I don’t know how to make the Win Shares system show Closers as comparable in impact to starting pitchers or other position players.   

                Here are the Win Shares and Loss Shares for Quisenberry and Sutter, Quisenberry first:

 

                   

Total

YEAR

TEAM

W

-

L

G

Saves

SO

BB

ERA

Won

Lost

Pct.

1979

Royals

3

-

2

32

5

13

7

3.15

4

1

.716

1980

Royals

12

-

7

75

33

37

27

3.09

12

5

.701

1981

Royals

1

-

4

40

18

20

15

1.73

7

2

.771

1982

Royals

9

-

7

72

35

46

12

2.57

15

4

.777

1983

Royals

5

-

3

69

45

48

11

1.94

18

2

.902

1984

Royals

6

-

3

72

44

41

12

2.64

16

4

.796

1985

Royals

8

-

9

84

37

54

16

2.37

15

4

.777

1986

Royals

3

-

7

62

12

36

24

2.77

7

4

.651

1987

Royals

4

-

1

47

8

17

10

2.76

5

1

.851

1988

Royals

0

-

1

20

1

9

5

3.55

1

1

.532

 

Cardinals

2

-

0

33

0

19

6

6.16

1

3

.194

1989

Cardinals

3

-

1

63

6

37

14

2.64

6

2

.733

1990

Giants

0

-

1

5

0

2

3

13.5

0

1

.000

                   

 

 

 

                   

107

35

.752

 

 

                   

Total

 

YEAR

TEAM

W

-

L

G

Saves

SO

BB

ERA

Won

Lost

Pct.

1976

Cubs

6

-

3

52

10

73

26

2.7

9

3

.742

1977

Cubs

7

-

3

62

31

129

23

1.34

21

0

.990

1978

Cubs

8

-

10

64

27

106

34

3.18

12

8

.593

1979

Cubs

6

-

6

62

37

110

32

2.22

17

3

.837

1980

Cubs

5

-

8

60

28

76

34

2.64

12

6

.682

1981

Cardinals

3

-

5

48

25

57

24

2.62

10

5

.674

1982

Cardinals

9

-

8

70

36

61

34

2.9

13

7

.653

1983

Cardinals

9

-

10

60

21

64

30

4.23

6

10

.387

1984

Cardinals

5

-

7

71

45

77

23

1.54

19

3

.845

1985

Braves

7

-

7

58

23

52

29

4.48

7

9

.444

1986

Braves

2

-

0

16

3

16

9

4.34

2

2

.519

1988

Braves

1

-

4

38

14

40

11

4.76

2

6

.284

                   

 

 

 

                   

130

62

.677

 

 

                The gross weight of Sutter’s career is greater than the gross weight of Quisenberry’s career, because

(1) Sutter had more strikeouts and walks,

(2)  the DH Rule in the American League removes one avenue by which pitchers can contribute to their teams, and

(3) he had more Saves, which indicates that he was used more times in high-leverage situations.  

However, Quisenberry’s winning percentage is 75 points higher than Sutter’s, so Quisenberry is credited with 72 more Win Shares than Loss Shares, whereas Sutter has only 68—consistent with Joe’s pro-Quisenberry argument.  

                Among the 12 candidates listed here, I would rank Quisenberry 5th.  

 

 

10.  Ted Simmons

                 

Batting

Fielding

Total

Winning

YEAR

Team

AB

HR

RBI

AVG

OBA

SLG

OPS

W

L

W

L

Won

Lost

 Pct

1968

Cardinals

3

0

0

.333

.500

.333

.833

0

0

0

0

0

0

1.000

1969

Cardinals

14

0

3

.214

.250

.357

.607

0

0

0

0

0

1

.378

1970

Cardinals

284

3

24

.243

.333

.317

.650

4

8

3

1

7

9

.429

1971

Cardinals

510

7

77

.304

.347

.424

.771

13

9

5

2

17

11

.607

1972

Cardinals

594

16

96

.303

.336

.465

.801

15

9

6

2

21

11

.654

1973

Cardinals

619

13

91

.310

.370

.438

.808

19

7

6

3

25

10

.720

1974

Cardinals

599

20

103

.272

.327

.447

.774

14

12

6

3

19

15

.571

1975

Cardinals

581

18

100

.332

.396

.491

.887

19

4

5

3

24

7

.764

1976

Cardinals

546

5

75

.291

.371

.394

.765

14

8

6

2

20

11

.652

1977

Cardinals

516

21

95

.318

.408

.500

.908

18

3

4

3

22

6

.789

1978

Cardinals

516

22

80

.287

.377

.512

.889

19

3

4

4

22

7

.757

1979

Cardinals

448

26

87

.283

.369

.507

.875

13

5

4

3

17

8

.689

1980

Cardinals

495

21

98

.303

.375

.505

.880

16

4

3

4

20

8

.709

1981

Brewers

380

14

61

.216

.262

.376

.638

6

11

2

3

9

14

.380

1982

Brewers

539

23

97

.269

.309

.451

.759

13

11

3

3

16

14

.532

1983

Brewers

600

13

108

.308

.351

.448

.799

17

8

1

5

18

13

.591

1984

Brewers

497

4

52

.221

.269

.300

.569

5

18

1

4

6

22

.199

1985

Brewers

528

12

76

.273

.342

.402

.743

11

12

1

3

12

15

.443

1986

Braves

127

4

25

.252

.313

.386

.698

3

3

0

1

3

4

.411

1987

Braves

177

4

30

.277

.350

.390

.740

3

4

1

1

4

6

.416

1988

Braves

107

2

11

.196

.293

.308

.601

1

4

0

1

1

5

.240

                 

 

 

 

 

 

 

 
                 

225

142

60

54

285

196

 

 

                With regard to Ted Simmons, it appears that I may need to revise my previously expressed opinion.   With regard to Simmons and the Hall of Fame, the argument I have long made is this:

                1)  Simmons actual defensive performance is much better than his defensive reputation, and

                2)  Simmons’ is one of the best-hitting catchers of all time.   

                As to Simmons’ defense. .. .it is hard to remember that in the early 1970s there were no publicly available statistics about runners thrown out or stolen bases allowed by catchers, and this allowed catchers to be judged by reputation and appearance.    When you play for a non-contending team, it hurts your reputation—particularly at catcher.   When Simmons was a young player the Cardinals were not competitive.   They lost 90 games in 1976, 93 games in 1978, and played about .500 baseball most of the 1970s.  

                Simmons’ defense was often compared to Johnny Bench, two years older than Simmons, which, you know. . .it’s like comparing me to Shakespeare.   There were no stats about Catcher’s ERA in that era, and none of the other pretty good methods we have now to understand what catchers are doing.   Catchers were being "seen" in the dark, in a sense.    Those who played for championship teams were assumed to be good catchers; those who didn’t, weren’t.  

                Simmons was in fact a far better defensive catcher, as a young player, than he was given credit for being.   Simmons threw out 39% of potential base stealers in 1970, 42% in 1971, 44% in 1973, 44% in 1976, and 36% in 1977, 1981 and 1982.   These are not only "adequate" numbers; they’re very good numbers.   They’re not Johnny Bench numbers, but they’re really good.  

                Simmons was very much underrated as a defensive player, and he was one of the better-hitting catchers of all time.    That’s all true.    But my thinking about Hall of Fame standards is better organized now than it was ten years ago, or five years ago, and I am no longer certain that Simmons meets a Hall of Fame standard.     While his defensive contribution as a young player was very good, his defensive contribution over the second half of his career—as is true of many players—was very limited.     Simmons was in the majors for 21 years.   I credit him with a defensive winning percentage of .676 for the first seven seasons, .558 for the second seven seasons, .271 for the last seven seasons.    That’s a common pattern.   For his career as a whole, his defensive winning percentage is .528.   

                As to his offense. . .well, his impact with the bat is similar to that of Dave Parker or Steve Garvey, a little less.   The critical issue is whether catchers have to be treated differently, in Hall of Fame evaluation of their won-lost impact, than other position players.

                It is certainly a reasonable argument that catchers do have to be treated differently than other players, because catchers get their bodies beat up so much by foul tips and squatting and carrying around 20 pounds of gear.    It is not fair to look at catchers as if these things didn’t happen. 

                Ted Simmons was a better ballplayer than Ray Schalk, Rick Ferrell or Ernie Lombardi, all of whom are in the Hall of Fame.     But one doesn’t need to make special rules for Bill Dickey or Gabby Hartnett or Yogi Berra or Johnny Bench or Ivan Rodriguez to make him a Hall of Famer.    Their careers are short, yes, but they still manage to qualify easily by the standards applied to other positions—300 career Win Shares, or 100 more Win Shares than Loss Shares.   Ted Simmons was a significantly better player than Ernie Lombardi, but I wouldn’t have voted for Ernie Lombardi.    It is not clear to me at this time that Ted Simmons’ career meets a Hall of Fame standard, and I would rank him 7th out of these 12 candidates.  

 

11.  George Steinbrenner

                I don’t really have any organized way of thinking about whether an owner should be in the Hall of Fame.   Steinbrenner probably had more impact on the game than any other owner since Walter O’Malley, perhaps more than O’Malley.   He did what all fans want their owners to do:  he committed himself absolutely to the success of his team.  He did that for an extremely long time.   Intuitively, I think he is an obvious Hall of Famer, and I would rank him 3rd among these 12 candidates.

 

12.  Joe Torre

                 

Batting

Fielding

Total

Winning

YEAR

Team

AB

HR

RBI

AVG

OBA

SLG

OPS

W

L

W

L

Won

Lost

 Pct

1960

Braves

2

0

0

.500

.500

.500

1.000

0

0

0

0

0

0

1.000

1961

Braves

406

10

42

.278

.330

.424

.754

9

8

3

2

13

10

.565

1962

Braves

220

5

26

.282

.355

.395

.750

5

4

3

0

8

4

.655

1963

Braves

501

14

71

.293

.350

.431

.781

13

8

4

2

17

10

.631

1964

Braves

601

20

109

.321

.365

.498

.863

19

6

3

4

22

10

.684

1965

Braves

523

27

80

.291

.372

.489

.862

16

6

5

2

21

8

.709

1966

Braves

546

36

101

.315

.382

.560

.943

20

2

4

3

25

5

.832

1967

Braves

477

20

68

.277

.345

.444

.790

12

9

4

3

16

12

.580

1968

Braves

424

10

55

.271

.332

.377

.709

12

6

3

4

15

9

.622

1969

Cardinals

602

18

101

.289

.361

.447

.808

18

6

6

3

24

9

.732

1970

Cardinals

624

21

100

.325

.398

.498

.896

18

7

5

3

22

10

.686

1971

Cardinals

634

24

137

.363

.421

.555

.976

27

-3

3

6

30

2

.929

1972

Cardinals

544

11

81

.289

.357

.419

.776

14

8

4

4

18

12

.599

1973

Cardinals

519

13

69

.287

.376

.403

.779

16

6

2

5

18

11

.627

1974

Cardinals

529

11

70

.282

.371

.401

.772

13

8

4

4

17

12

.579

1975

Mets

361

6

35

.247

.317

.357

.674

6

10

2

2

8

13

.402

1976

Mets

310

5

31

.306

.358

.406

.764

9

4

2

3

10

7

.597

1977

Mets

51

1

9

.176

.204

.294

.498

0

3

0

0

0

3

.056

                 

 

 

 

 

 

 

 
                 

228

97

57

51

285

148

.658

 

                It is my opinion that Joe Torre had a Hall of Fame career as a player, and that he had a Hall of Fame career as a manager.   Among the 12 candidates on the Expansion-era ballot, I would rank Torre #1.

                As a player, Torre’s career is very similar to Ted Simmons’, almost certainly more similar to Simmons than to any other player in history.   As a catcher he threw extremely well.   He threw out more than 40% of base stealers six times, as a regular, including 49% in 1961 and 1966.   He did, as Simmons did, have some trouble catching the ball. ..that is, some trouble dealing with potential wild pitches/passed balls.    Torre and Simmons were teammates from 1969 to 1974, and Torre moved away from the catcher’s position so that Simmons could catch.

                The difference between Torre’s playing career and Simmons’ career is this:  that after he gave up catching, Torre had an MVP season at third base.   Simmons gave up catching at about the same age, but he never had a great season after he stopped catching.   If Simmons had had an MVP season as a first baseman or DH in 1984, he would be a Hall of Famer, but he never had that season, never came close to it.  

                As a manager, Torre’s record exceeds the minimal standards of a Hall of Fame career by about 77%.

 

                My ballot:   1. Joe Torre (yes), 2.  Bobby Cox (yes), 3. George Steinbrenner (yes), 4.  Tony La Russa (yes), 5.  Dan Quisenberry (yes), 6.  Dave Parker (maybe), 7.  Ted Simmons (maybe not),  8. Billy Martin (maybe not right now), 9. Steve Garvey (probably not), 10. Tommy John (I’m afraid not), 11. Dave Concepcion (no).  Marvin Miller. ..certainly not right now; we can talk about it in a few years.

 
 

COMMENTS (58 Comments, most recent shown first)

1000ringsofsaturn
I'm surprised that your list of position players for the Hall consist of Torre, whose candidacy as a player is muddled by his success as a manager, and Quiz, who is a reliever, an no one else. I'm also surprised you like Torre as a HOFer: He must be like the 15th best C of all time.
12:51 AM Jul 1st
 
bjjp2
So, looks like they did a pretty good job.
12:08 PM Dec 9th
 
tangotiger
The offense/defense relationship can be expressed like this:

80^2 = 70^2 + 40^2 (rounding error)

That means that one standard deviation in runs at the team level (per 162 games) is 80 runs for hitting, 70 for pitching and 40 for fielding.

And if you try to suggest that pitching is 70/(80+70+40), you'd get 37%.

By the same rationale, fielding is 40/(80+70+40) = 21%.

And offense would be 80 / (80+70+40) = 42%.

But we started off with the idea that offense=defense. Ultimately, the problem is between the adding of standard deviations (which you can't do), but maintainting the relationship at the standard deviation (which you do want).

This inherent conflict essentially rears its ugly head at some point, if you start with a top-down approach of trying to do the splitting off the bat.

In WAR, I do the reverse, starting with the idea that the spread is 80, 70, 40, so that at the player level, I'll see that kind of relationship as well. The spread in player hitting is twice that of player fielding. The spread in hitting is a bit more than the spread in pitching. That gets maintained.

And when you get a player's total rating (off+def), you see that the spread will be sqrt(80^2 + 40^2) ~ 90 runs.

So at the nonpitcher level, the spread at the team level is 90 runs, and at the pitcher level is 70 runs. We can then allocate wins and losses at that relationship 90/(90+70) = 56% nonpitchers, and 44% pitchers (rounding error).

But, you can't then take that split, and try to reverse it back into separate components of offense, pitching, and fielding. It won't work so cleanly.

8:59 AM Nov 29th
 
cderosa
Ha, well it seems I am in the habit of seeing parts of Win Shares as strengths that others see as weaknesses. I shall ponder it.

Chris
8:27 AM Nov 29th
 
tangotiger
"But before the tools are even applied, the assumption Win Shares makes about who gets responsibility for making outs on balls in play creates a narrower range for fielders than DRS and similar systems. So while the breakdown of hitting/pitching/fielding may be similar in win shares and other systems on average, win shares isn't going to have the same extremes"

All of this is correct and well-stated, and I believe this is an inherent limitation with Win Shares.

As I stated elsewhere, the spread in hitting is about double the spread in fielding. At the team level basically, one standard deviation is about 80 runs for hitting, while it's about 40 runs for fielding.

Win Shares however forces that relationship to be 4:1.
7:31 AM Nov 29th
 
cderosa
Hi Tango, sorry if the moment has passed here (and hope everyone on this thread had a nice Thanksgiving), but in reference to what you wrote:

>we give the pitcher credit for .95 outs or .14 outs or .53 outs or what have you. (The overall average is around .7 outs.) The fielder is given credit for the gap between that number and either 1 (if he makes the out) or 0 (if he doesn't).

I think I am trying to say the same thing in regard to DRS: 100% of the deviation from that overall average, i.e. the gap, goes to the fielder. What I am suggesting is that the equivalent step in the Win Shares system proceeds from a different assumption: the degree to which the team is above or below average at turning balls in play into outs figures equally (50/50) in the pitchers' and fielders' claims on the team's wins.
The +/- tool that lies at the base of DRS is a sophisticated tool (as your description indicates), and the Win Shares equivalent is quite crude in comparison (was team's park-adjusted Defensive Efficiency Record above or below average?). But before the tools are even applied, the assumption Win Shares makes about who gets responsibility for making outs on balls in play creates a narrower range for fielders than DRS and similar systems. So while the breakdown of hitting/pitching/fielding may be similar in win shares and other systems on average, win shares isn't going to have the same extremes, like a +64 season for Frankie Frisch.
I see in one comment below, I stated incorrectly that the fielder "gets the average too." I should have said the fielder also gets the positional baseline, which is in fact a result of a division of responsibility between pitcher and fielder. It is the deviations from the baseline that are the subject of the differing 100/0 or 50/50 interpretations.

Chris DeRosa

5:55 AM Nov 29th
 
tangotiger
You'll have to ask Dewan or Ben how they capture all that. Sportivsion captures launch parameters with HITf/x, as does Trackman.
8:21 AM Nov 28th
 
KaiserD2
Briefly:

Riceman, I think you wildly overstated your case. Frisch's result, we all seem to agree, was probably anomalous, and yes, an average second baseman (which we agree he was not) would probably have made more outs than expected in his place. But the recognition of some anomalies certainly doesn't invalidate the whole metric, and Humphreys's measurements show pretty consistent patterns year-to-year and over the course of people's careers. There's nothing unique to fielding about all this; I'm sure many high batting averages have benefited from the lucky placement of poorly hit balls.

Cderosa, the pitcher does play some role. How much is still, as far as I can tell, a huge matter for debate. You seem to be saying that the pitcher deserves credit for getting the ball over the plate (or near it) so that the batter would swing and put it into play. I suppose that's true, but Voros McCracken and others have argued that no pitcher seems to have a measurable advantage in turning balls in play into outs. Others still disagree.

But more to the point--Humphreys is assuming that two things are important about pitchers: their handedness (and its effect on whom they face), and their ground ball/fly ball tendency. Those two things have substantial effects, on the average, on how many balls any position player except a catcher has to field. In effect, yes, he is assuming that there are no pitchers who are especially good at generating easy grounders to short or fly balls to right. Or that if there are, those tendencies won't show up significantly in the course of a season. But he's comparing fielders to each other.

As I indicated--at one point in his book Humphreys identifies the 1973 Baltimore Orioles as the best defensive team in history and tells how many runs he thinks their defense saved. (Sorry but I don't have time to look it up--I'm supposed to be editing a non-baseball book at the moment. . ) That means he has reached a conclusion on how many runs the pitching staff saved, too. But he doesn't address that issue systematically.

DK
12:02 PM Nov 27th
 
Riceman1974
Tango:

Are the angles, exit speed, and other bated ball characteristics determined by dedicated cameras, stringers in the press box, or a combination of both? And isn't it true that UZR and DRS often have widely different results for the same player, even though they share the same BIS data?


8:59 AM Nov 27th
 
tangotiger
I presume if you had three tremendous outfielders and a terrible player at third base, that a GM would prefer to flip his 3Bman for Cabrera, rather than try to trade a McCutchen for a Longoria or Beltre, and then bring in Trout.
8:49 AM Nov 27th
 
steve161
Riceman1974, for me at least, that final "one more point" clouded the issue. If you had ended your paragraph before typing that, I would have agreed 100%: the difference in defensive value between Trout and Cabrera is not remotely close to 30 runs or more and in no way justifies preferring Trout for the MVP.

In effect, you're arguing that defensive runs, as computed by current metrics (whether Runs Saved or any other), are not the same as offensive runs (e.g. Runs Created). Bill James has made a similar argument, and I agree with him. These are apples and oranges.

However: I don't agree that Cabrera's offensive superiority is as great as you seem to suggest. The OPS difference is 1.078-.988 (amusingly, when Trout played CF his OPS was 1.078). Throw in baserunning and you get a Runs Created difference of 146-141 favoring Cabrera. They're pretty close. An MVP case can be made for either and I personally would have given Cabrera the nod by a hair.

But take their respective ages into account. and I think Tom Tango's over/under of 5 is plausible. Indeed, if he could tell me who those 5 GMs are, I'd strongly recommend firing them.
5:58 AM Nov 27th
 
tangotiger
At least you went from the impossible position of "guarantee all 30", to a reasonable-sounding over/under of 20.

Anyway, given "current performance level", which I guess you mean last two years or something, and if they could get that for the next 10 years, I'd think the over/under for Cabrera would be closer to 5.
10:01 PM Nov 26th
 
tangotiger
Chris: it works as I describe it. This is true for UZR, this is true for Dewan, this is true for every single zone system. You figure out how often a ball with the particular characteristics is normally turned into an out. And you give the difference of that and 1 or 0 to the fielder, depending on whether there is an out or not.

The difference between UZR and Dewan is in determining the "characterstics" of each batted ball. Sure, you start with the spray angle and distance, but then you look at hang time or its proxy, you look at launch angle (proxied by GB, FB, LD, Pop), you look at the exit speed (proxied by observer), you look at whether the pitcher is a GB pitcher or not, the handedness of the batter, pitcher, the park, the base-out configuration, and so on.

Once you have all that information, you make a best estimate as to how often a ball with those characteristics is turned into an out.
9:57 PM Nov 26th
 
Riceman1974
Regarding Cabrera v Trout, the best way to define my thought process is imagine if you were guaranteed 10 years of current performance level from each player, and a GM was forced to pick between the two. I say in that context, Cabrera is the choice of over 20 at least.
7:20 PM Nov 26th
 
cderosa
Hi Tango,
I'm not seeing that, for the marginal runs, in DRS. From my reading of the Fielding Bibles, I get that their is a division between fielders and pitcher to establish the positional baselines, but it looks to me like all the observed marginal gains and losses go in the fielder's column. No?

Thanks for weighing in,
Chris DeRosa
4:52 PM Nov 26th
 
tangotiger
"you asked all 30 GMs whom would you rather have, I guarantee you all 30 would say Cabrera"

This is summary opinion with no evidence.

If you want to suggest that the over/under of number of GMs that prefer Cabrera to Trout (at same age) might be 20 or 21, that's at least something reasonable enough. But to put the over/under at 29.5? How can we even have a conversation?

I'd set the over/under at 10 for Cabrera.
3:08 PM Nov 26th
 
steve161
"One final point: Let’s assume that Trout and Cabrera were the same age (let’s split the difference and say 25), and you asked all 30 GMs whom would you rather have, I guarantee you all 30 would say Cabrera."

Well, we don't know what Trout will do at 25, but he reached 40 Win Shares at 22. Cabrera didn't reach 30 until he was about 27.

Cabrera might turn out to have been the better hitter when their careers are over but if so it won't be by much. In every other aspect of the game Trout will most probably have been Cabrera's superior, unless he gets hit by a bus.​
8:39 AM Nov 26th
 
tangotiger
"I think it is a better idea to credit Frisch with *part* of the run value for those plays"

This IS how it works. If a pitcher puts a ball in play, basically, he gets credit for 0.7 outs. If a fielder makes an out, the fielder get +.30 outs. If he doesn't, the fielder gets -.70 outs.

All we are doing is being more particular about each ball in play. Some we give the pitcher credit for .95 outs or .14 outs or .53 outs or what have you. (The overall average is around .7 outs.) The fielder is given credit for the gap between that number and either 1 (if he makes the out) or 0 (if he doesn't).
7:22 AM Nov 26th
 
bjames
I not only think that the "guarantee" offered on behalf of GMs about Cabrera and Trout is incorrect, I think it is likely that most GMs would prefer Trout to Cabrera at the same age.
3:44 PM Nov 25th
 
rgregory1956
I really enjoy these types of discussions. They are interesting, entertaining and informative. I just wish more of you were active in the Reader Posts, where we discuss things like this, and many, many more topics.
1:10 PM Nov 25th
 
cderosa
Hi KaiserD2, thanks for the description of the Humphreys system. I will have to read up on it more someday! I’d love to spar more about the Pythagorean too, but I’m going to limit myself to the hotter question here. You wrote:

>I suppose what you are saying, to go back to Frankie Frisch, is that it's quite possible that his pitchers generated a wildly above average number of ground balls in his direction.

What I’m saying is that even if we assume Humphreys (or Sean Smith’s Total Zone, or DRS for a recent second baseman) is in the correct range for the number of balls-in-play above average Frisch turned into outs, I think it is a better idea to credit Frisch with *part* of the run value for those plays, rather than all of it, and to credit part of it to the pitching staff, not for inducing balls to be hit to Frankie Frisch or near Frankie Frisch, but merely for putting the ball in play somewhere in the ballpark, where fielders have a chance to do something about it. The logic being, every time you resolve a batter-pitcher confrontation with a ball in play, the batter joins the general category of someone who is two-thirds out.

“Two-thirds out” was how Bill put it; I just find the argument persuasive.

Chris DeRosa

12:51 PM Nov 25th
 
Riceman1974
KaiserD2 and cderosa:

Regarding Mr. Frisch again. I think it's safe to say that the reasons why his defensive numbers were so outrageous that year were due to A) Frisch was an excellent 2nd basemen with good range, and B) Frisch had an inordinate number of balls hit to him that year, even accounting for pitching handedness and everything else Humphrey's system and other defensive metrics account for.

Now, doesn't reason B inherently mean that if you replaced Frisch with "Joe Average" second baseman, playing behind the exact same pitchers in the exact same innings as Frisch did, then Joe Average's numbers would also show him to be an above average player, with multiple runs saved above average, simply due to the fact that his pitchers are giving up a massive amount of ground balls to second? For all we know, Joe Average might have saved 30-40 runs above average for the Cardinals that year, with Frisch's own excellence adding another 20-30. In other words, Frisch may have been only 20-30 runs above average, because even an average 2nd baseman would have the skills to handle all those ground balls to second for the Cards that year.

If the above statement is true, then ANY defensive metric which states that player X saved Y numbers of runs above the level of a mythical average player, is essentially worthless. We have no idea what an average player would have done instead of Frisch, or Andrelton, or anybody, because the only player on the field at that moment getting to that batted ball was that specific player. I realize that BIS data says that only x% of balls were turned into outs in that specific zone, but is that zone league-wide? Is it specific to that park? What about the speed of the ball, the wind and humidity? There are so many variables. These new defensive metrics appear to be state that ONLY Andrelton, or Frisch, or whomever, could have made that play on that ball. I believe, more often than not, more than half the position players in the league could have made that play, because the true talent discrepancy in defensive skill is not that vast.

Most sabermetricians and GMs would agree that finding players with average pitching and hitting ability is rare, but average fielding ability is not. This fact has been proven throughout the history of the game. Great hitting and pitching are highly valued and have been highly compensated. Good defense is cheap. There are 100 Greg Luzinsky-type players (all hit no mitt), in baseball history for every Bill Bergen (all mitt no hit). I believe these metrics are useful because they identify who is a consistently good fielder, but to say that a player was worth X number of runs defensively is a worthless statement. And not to open Pandora’s Box here, but isn’t the defensive evaluation the key difference between Trout and Cabrera’s WAR numbers? I won’t deny that Trout is a good centerfielder and Cabrera is a bad third basemen, nor will I deny that centerfield is more important defensively than 3rd base, but I refuse to accept that Trout is 30 , or 40, or 50 runs better than Cabrera defensively and hence is the clear choice for MVP. One final point: Let’s assume that Trout and Cabrera were the same age (let’s split the difference and say 25), and you asked all 30 GMs whom would you rather have, I guarantee you all 30 would say Cabrera.
11:37 AM Nov 25th
 
KaiserD2
cderosa,

I didn't realize win shares had been changed in that respect and I'd like to take a detailed look at how some time. I don't agree with you about the Pythagorean formula vs. real won/loss records. I do think chance is by far the biggest reason for deviation from the Pythagorean formula in real records. If it wasn't, we would observe teams or managers that consistently missed or beat their Pythagorean formula. Many people, including myself, have tried to do that, and no one has been successful.

Now try to bear with me as I try to explain what I consider to be Humphrey's approach to the batted ball problem. While he didn't publish an answer to the question, how much is pitching and how much is fielding, it's inherent in his approach, because he's basing all the individual ratings on the average number of balls in play turned into outs by fielders at a particular position, after correcting balls in play for the handedness of batters and ground ball tendencies. I think he would argue (and I don't know the man, not yet) that the average for centerfielders reflects two things: the average skill of pitchers at generating catchable fly balls and the average skill of center fielders.

Having said that, I suppose what you are saying, to go back to Frankie Frisch, is that it's quite possible that his pitchers generated a wildly above average number of ground balls in his direction. I had suggested on the other hand that chance had generated a lot of ground balls just within his reach. Both could be true, but either, it seems to me, would be random chance deviations that did not reflect superior skill either on the part of the pitchers or on Frisch's part. And such things do happen and are taken into account today by sabermetricians working for teams and monitoring fielding plays on video. One such told me that an outfielder for his team had had a remarkable number of balls hit just within his reach this season.

I hope that was reasonably clear.

DK
8:00 AM Nov 25th
 
cderosa
And hi again. Regarding win shares being based on the team record rather than the Pythagorean projection: I actually consider this a virtue of the original system.

Sabermetrics erred when it started treating teams’ won-lost records as nothing more than lucky or unlucky deviations from their runs scored/runs allowed “true quality.” It took something that was useful to know (runs are somewhat more stable year to year than decisions) and fetishized it to the point of obscuring rather than illuminating baseball’s complexity.

But if I remember correctly from earlier pieces on this site, the new win shares/loss shares do not break down from the team’s record, but build up from the player’s record, so Bill’s current system may be more to your liking.

7:27 AM Nov 25th
 
cderosa
Hi KaiserD2, what I was trying to get at was that win shares is unlikely ever to attribute that much of an area (64 runs above average) of responsibility to a fielder. What the original win shares did was to start with the runs saved by the team, and divide credit between the fielders and the pitchers. The more the team did the things done mostly by pitchers (strikeouts, walks, homers) the greater share of responsibility attributed to the pitchers. The more the team did the things done mostly by fielders (errors, double plays, etc.) the greater the share of responsibility attributed to the fielders. Once you had the fielders’ runs, you broke them down to individuals based on indicators derived from commonly available data. The part I want to emphasize, though, is that the system treats turning balls in play into outs as a thing that is done together by fielders and pitchers, 50/50.

In Defensive Runs Saved, it is 100/0 in favor of the fielder. If Frankie Frisch the Fordham Flash catches a ball that 85% of the time a fielder doesn’t catch, he gets 85% of the run value of the play. For the pitcher, it’s a non-event. None of the marginal credit goes to the pitcher for putting the ball in play where fielders have a chance to contest the outcome. In WAR on Baseball Reference and in Total Runs from this site, the fielder gets all the marginal gains for Defensive Runs Saved above average, and then gets the average too in the form of a positional baseline.

If I have this right, this assumption about batted balls goes a long way toward explaining why the range of fielding credit in Win Shares is much tighter than in WAR, Total Runs, or the Humphrey’s system as you describe it. And because WAR treats pitching as what is left over after you deal with fielding, the WAR results for pitchers are broader than Win Shares’ as well.

You know how a guy Win Shares seems to characterize as “not so hot in the outfield anymore,” has other systems screaming “he’s a total catastrophe”? This is the main reason. Not because Win shares doesn’t take advantage of modern batted ball tracking, but because it treats turning balls into outs as a team effort to begin with.

Bill says there is no compelling logic on the issue and asks the reader: what do you believe? I believe that while 50/50 is a leap of faith, it makes better sense than 100/0. Voros McCracken demonstrated that pitchers do not consistently create outs on balls in play, but it seems to me it is in their area of responsibility to create opportunities from which outs may be made.

7:24 AM Nov 25th
 
KaiserD2
cderosa--I can't see how it really matters how you make the calculation. My hypothesis is that, among other things, Frisch simply had an extraordinary number of balls hit just within reach of him in that year and turned them into outs. That was reflected in his assists total (Humphreys doesn't count putouts for infielders, which I agree with) and thus it will pop up in any system of evaluation. Humphreys does his best to control for the number of right- and left-handed batters the team faced and the ground ball tendencies of the pitching staff. Having done that, his regressions estimate how many plays an average second baseman would make. Evidently Frisch's total was WAY in excess of what the model predicted. I don't know exactly how win shares does fielding but there must be some similarities.

I have always felt, by the way, that win shares is subject to its own anomaly because it's based on actual team wins and losses, not runs scored and runs allowed. If a team has performed substantially above or below its pythagorean percentage, its players, as I see it, will earn either more or fewer win shares than they actually deserved based on their individual performance.​
7:37 PM Nov 24th
 
cderosa
KaiserD2 wrote: "I'm sure any system would produce some anomaly along the lines of Frankie Frisch in that one year."

But Win Shares would not, I think. The original win shares gave half the responsibility for the quality of the team's Defensive Efficiency Record (turning balls in play into outs) to the pitchers. Defensive Runs Saved, and therefore the version of WAR that incorporates it, gives the fielders all the responsibility for turning balls in play into outs. It seems to me that this is the biggest reason for the systems' major disagreements, and the heart of the questions people have raised on this site about the scale of fielding.

Bill says in the article that there is no compelling logic on the issue, but I do find somewhat compelling something he wrote in another piece, which is that if the pitcher gets the batter to put the ball in play, the batter is, on average, two-thirds out. What strikes me about this is that it is true even for a team of lousy fielders, and therefore pitchers must share in the responsibility for what happens on those balls.

This is the most important reason why I prefer win shares to WAR, even though my preference makes me feel, in the general sabermetric conversation, like the guy who bought Betamax.


5:42 PM Nov 24th
 
hotstatrat
That's what I thought: SLIGHTLY the other way - meaning the left fielder has to cover more ground because center-fielder is generally shaded slightly the other way - and it is the CF that gets those, not the RF. But, when a player really gets a hold of a ball, he pulls it. I would think that happens more than it gets GREATLY sliced to the opposite field. However, a look at range factors among leading outfielders this year shows qualifying right fielders with a median of 2.08 and qualifying left fielders at only 1.86. Ouch.

Is this something new or has this always been the case? Going back to 2008: LF: 1.88, RF: 2.03. Hmm. OK, 2003? Well, that's much closer: LF: 2.06, RF: 2.07. But was that a fluke? 2002 is as far back as ESPN goes: LF: 1.89, RF: 2.02. I surrender.

If this has been a trend over a longer period of time, it is more evidence that players are getting better and better. At my softball playing level almost everyone is a pull hitter.
4:04 PM Nov 24th
 
bjames
To Bill James (and everyone else): the standard deviation among individual batters in a league is surely much, much higher than the standard deviation of park-adjusted runs scored among teams. No team has ever been as far above the league average in run production as Mike Trout is.


This is complete nonsense. Teams have been three times as far above league as Trout has been. ​
3:53 PM Nov 24th
 
tangotiger
"so more balls will be hit to left, no?"

No. These days especially, there are as many or more balls hit to RF. The basic rule is that groundballs are pulled, and flyballs go SLIGHTLY the other way (on average).
10:27 AM Nov 24th
 
hotstatrat
re: left fielders vs. right fielders - yes, a right fielder is expected to have a better arm than a left fielder, but a left fielder is expected to have more range. There are more right-handed batters than left-handed, so more balls will be hit to left, no? That's why Mike Trout played left field when Peter Bourjos was healthy and why Ricky Henderson, Tim Raines, and Lou Brock were left-fielders, no?

re: Steinbrenner - "character" is part of the Hall of Fame quotient. It is so undefinable that it means whatever we want it to mean. Go ahead and endorse him, but we don't have to and I wouldn't.

re: % of defense that is fielding. Wouldn't a middle infielder have a much greater share of a team's defense than a corner infielder; likewise a CF have a greater share than a RF/LF? . . . just as the overall total would depend whether the pitcher was a finesse pitcher (John) or a strikeout king (Ryan).
10:02 AM Nov 24th
 
KaiserD2
To Riceman: I'm sure any system would produce some anomaly along the lines of Frankie Frisch in that one year. And clearly, it's possible for a single player who have a very unusually large number of balls hit just within his reach in a single year. It certainly doesn't invalidate the system. Humphreys managed to explain a couple of other anomalies, such as Joe DiMaggio's terrible defensive stats in 1947, and the numerous plays made by first basemen in the Polo Grounds. I did think one thing was missing from Humphreys's model: he didn't use pitchers' assists as a factor in his regressions, and they are, I found many years ago, a good predictor of ground ball tendencies. But I still think he did very well. And I have no trouble believing that GMs and managers could fail to appreciate the significance of defensive statistics. The Yankees cost themselves a lot of games by not putting ARod at shortstop and Jeter at third.

To Bill James (and everyone else): the standard deviation among individual batters in a league is surely much, much higher than the standard deviation of park-adjusted runs scored among teams. No team has ever been as far above the league average in run production as Mike Trout is. Thus I see no reason to expect that things would not be different with respect to individual fielding on the one hand and team fielding on the other. The SD among individuals would be higher.

DK
9:33 AM Nov 24th
 
Riceman1974
KaiserD2 (great name by the way):

I read (and reread) Humphrey's Wizardry, and even he is not completely confident in some of his conclusions. In some seasons he has September call-ups leading the league in runs saved at their position because they had 2 good weeks with a lot of balls hit to them (e.g., Dave Philly, 1946 White Sox). He also had Frankie Frisch saving 64!! runs above average in 1927! There are too many contemporary accounts to deny that Frisch was a great second baseman, but 64 runs above average? Were NL 2nd basemen so collectively terrible in 1927, that the average level of play was so poor compared to the Fordham Flash? I have much nothing but respect for Michael Humphrey's, John Dewan, Sean Smith, and Tango, and I find all their work fascinating if not illuminating, but I just question their run values when it comes to assessing defensive play. If these run values were undeniably true, why would Brendan Ryan be out of a starters job, when according to WAR, he's more valuable than most ML shortstops, including the man he's backing up? Our GMs just stupid, or could it be that defense is not as valuable, run-wise, as some people think.
8:59 AM Nov 24th
 
OldBackstop
I think in terms of comparing Ryan to John on preventing steals and holding runners, if you have ever played the game and been a baseruner, lefties and righties should be on a whole separate spreadsheet from each other. I'd be interested to see numbers there...but, sh1t, lefties are just staring you down. Huge advantage.
11:05 PM Nov 23rd
 
bjames
It is theoretically possible for the standard deviation of performance at some positions to equal the standard deviation for teams only if they are NOT randomly distributed. If they are randomly distributed, it is impossible for the standard deviation for a position to be as large as the standard deviation for a team, or anything approaching as large.
1:32 PM Nov 23rd
 
KaiserD2
A couple of points about fielding. First, I think that Michael Humphreys in Wizardry did the best job of creating a valid set of players' fielding statistics. It's clear, too, that his data would provide an answer to the how much is pitching/how much is fielding question, but he doesn't publish any systematic answers.

My second comment is that, speaking as a layman, it wouldn't be all that surprising for me if the SD for individual positions--especially the most important ones, SS and CF--were as high as the SD for teams. The reason is that fielding skill is in practice pretty randomly distributed, and it would be very unusual for a team to have almost all good or almost all bad fielders. In addition, a lot of high numbers for outfielders may be directly connected to low numbers for outfielders next door, who yield marginal fly balls to the better man.

DK​
12:07 PM Nov 23rd
 
CWright
I pretty much agree. I have strongly supported Joe Torre for the Hall of Fame. Cox and LaRussa are clear HOF managers. Because of the limited value of relievers, I'm on the fence at times on Quisenberry while also arguing I would cast a vote for him before I would for Sutter. But I think relievers should be represented in the HOF, and bottomline, if you give me a ballot, I would vote for Quisenberry.

The decision on Steinbrenner will be an interesting one. I'm not wild about putting executives into the HOF, in part because it is so hard to weigh accurately their individual contributions to the game. For me, such a candidate has to have a really exceptional case. I'd personally never vote for Steinbrenner. I know of too many examples where he would not play within the rules, and that interfers with my accepting that he made a Hall of Fame caliber contribution to the game. I'm even kind of surprised he is on the list given that going back to July 30, 1990 he was banned permanently from day-to-day management of the Yankees. I have to acknowledge that part of my reluctance to support Steinbrenner for the HOF comes from knowledge shared by front office members of the Yankees and is not in the public domain. But what is in the public domain is indictative of a pattern that is not consistent in my eyes with a contribution to the game worthy of the Hall of Fame. I also note that the Yankees were a much more successful team after George was banned from being involved in the day-to-day management of the team. It just doesn't add up to Hall of Fame for me.
9:38 AM Nov 23rd
 
Tubbs44
What was Howser and the Kansas City mgmt thinking in '86 by demoting Quiz from closer to just a part of a bullpen by committee? This cost him a ton of saves & likely the HOF.

The players chosen by the Historical Overview Committee left something to be desried. The Committee played it safe by selecting holdover players from the previous Exp Era ballot such as Garvey, Concepcion, and Tommy John who last 15 yrs on the BBWAA ballot. This year they added Parker who also lasted 15 yrs. It would be nice to see the Committee put Dwight Evans, Keith Hernandez, Bobby Grich and others who the BBWAA snubbed but are much stronger sabermetric candidates
10:51 PM Nov 22nd
 
wovenstrap
I think it's curious that you agree there is an HOF logjam and that you like this approach to the problem by the Hall, but you basically didn't vote for any of the players. OK, Torre and Quiz. But to my mind, this suggests either that there is NOT a logjam or that the Hall's approach won't do much to change anything. But I could be misunderstanding what will happen in future years, i.e. when guys like Raines get on the ballot.
9:13 PM Nov 22nd
 
KaiserD2
If you're going to argue about who should be in the Hall of Fame, you have to take some account, obviously, of who's already in it. This means that there's no way for me to argue based on my own standard of what makes a great player because at least half the position players in the Hall wouldn't come close to making it. So I won't argue about how many of Bill's candidates are great players. Basically, my standards rely much more on how often a guy met a very high single-season standard, and I hope to develop them at length in the future.

But having said that, I would argue very strongly against George Steinbrenner as a Hall of Fame candidate. He certainly wanted to win as badly as any owner in history, and he was certainly willing to spend as much money as anyone to do so, but the record shows, I think, that he had no idea how, and the record of the Yankees under his ownership proves it. He was not responsible for acquiring most of the key players on the mid-1970s team. For about 15 years, beginning in 1982, his leadership was quite disastrous on the field. It is my understanding that in the late 1990s he realized he had to back off, and he had also finally allowed the Yankees to develop prospects. Meanwhile, he did a number of things to bring himself and the game into disrepute, such as his absurdly co-dependent relationship with Billy Martin and his illegal pursuit of Dave Winfield. I don't see why he should be in the Hall.

DK
3:22 PM Nov 22nd
 
Steven Goldleaf
Don't get Steinbrenner at all. Awful human being, convicted, suspended from baseball, intemperate, bullying of his players, media, managers. There are so many owners who are modest and decent--can't we elect some of them first?
3:04 PM Nov 22nd
 
Riceman1974
Jwilt:

You make a good point. The spread in plays made looks legit, since the LHP for the Orioles was about the same for 2013 and 2011 (the year Reynolds averaged 2.00 RF), and the GB ratios were the same as well.

However, you are comparing the absolute extremes, aren't you? Reynolds fielded .897, he may have been the worst 3rd baseman since Butch Hobson, if the not the worst of all time. The Orioles had somebody else play 1/3 of 3B innings to make up for Reynolds's atrocious play. Machado is an excellent third basemen, who basically played every inning at third in 2013. My point was that the "average" level of play should be high enough to offset extreme play at most defensive positions, and I would expect at some positions (SS and 3B for example) the best payer may be +15-20 while the worst may be -15-20. But I find it difficult that a single player can be 40 runs better than average.
2:07 PM Nov 22nd
 
wovenstrap
Yes, we neophytes may need an extra article to sort out the defensive metrics issue. There was an article on here in August about Andruw Jones in which I got into it with Tango briefly about whether Andruw or a hypothetical equivalent was saving his teams 50 outs a year. I said he wasn't -- I don't think 162 games has nearly enough plays for that. Tango pointed to Dewan as (part of his) evidence that he might have the right numbers. Based on what I'm reading in the article above, I might be seeing indirect support from Bill. I'm also curious whether that's the same conversation or a different conversation, and if so, how.
2:04 PM Nov 22nd
 
tangotiger
If the standard deviation in team fielding runs is 40 per season, this would imply that the standard deviation at the positional level is around 13 runs.

If the standard deviation in team pitching runs is 69 per season, this would imply that the standard deviation at the pitching level is around... well, a bit more complicated because you have 200-inning pitchers and 70-inning pitchers... but, if you assume you have 9 pitchers at 162IP each, then at the pitcher level, one standard deviation is 23 runs.

The standard deviation in team offensive runs is 80 runs per season, which would imply a standard deviation at the hitter level of 27 runs per fulltime hitter.

27^2 = 23^2 + 13^2 (rounding error)

So, at the nonpitcher level, one standard deviation is sqrt(27^2+13^2) = 30 runs.

That's 30 for nonpitchers and 23 for pitchers, for the spread in observed runs at those levels.

We choose to use that spread in terms of allocation of value, or 30/(30+23) = 57%

12:58 PM Nov 22nd
 
tangotiger
pack: I responded in the comments below.
12:52 PM Nov 22nd
 
rtayatay
Regarding the splits - it looks like Bill likes 50/38/12.

I'm unclear on Tom's split.
12:31 PM Nov 22nd
 
jwilt
Here's another way of looking at the fielding spread:

We know there are fielders who have a spread of approximately one play per game between them. For example, that's about the difference between the Orioles' last two third basemen (Manny Machado and Mark Reynolds). Manny's old fashioned range factor is just over 3.00, Reynolds from a couple years ago on the Orioles was just over 2.00.

That's about one play per game, or maybe 150 per full season. If a single is worth .46 runs, and we assume those plays turned singles into outs, that's a 69 run spread between Manny and Mark Reynolds.

If the spread is as small as Bill suggests, +/- 10 or 15 runs, then how can you have players with a spread in plays made of one per game?
12:21 PM Nov 22nd
 
packbringley
"With my number—12%--there is no room for good fielders saving 25 runs a year. If the number is 7%, there is probably no room for good fielders saving 15 runs a year; you probably have to stop about 10."

I'm not very savvy. Does Tom think Bill is wrong about this?
12:00 PM Nov 22nd
 
bjames
Responding to Chuck. . .the last significant revision of the Win Shares system was designed to make the system distinguish between right fielders and left fielders. Not sure when exactly that was done; probably about 2009.

If you look at the defensive statistics of right fielders and left fielders, they're basically the same. . .very minor differences. I don't believe that's an accurate reflection of reality. The defensive responsibilities of left fielders and right fielders are essentially the same EXCEPT THAT THE RIGHT FIELDER IS EXPECTED TO THROW WELL ENOUGH TO KEEP RUNNERS FROM GOING 1ST to 3RD AND TO KEEP BATTERS WHO HIT A DOUBLE FROM TAKING THIRD. But in the statistical image of the game based on traditional statistics, this difference essentially disappears. I adjusted the system about 2009 to give more credit to right fielders, by robbing just a tiny bit of credit from all other fielders.
10:44 AM Nov 22nd
 
chuck
Bill, when you compared Parker and Evans back in 2009, Parker had a W-L record of 289-212, which is short of both HOF standards; here he gets to the 300 standard and almost to the +100 standard. It appears that the difference is in the defensive numbers, as previously he was 43-68. What changed in your method that boosted him up to 54-57?
10:33 AM Nov 22nd
 
tangotiger
For my database at the office, I only have it updated through 2011. Anyway, since 1947, one standard deviation in runs allowed at the team level is 77 runs (per 162 games). Since 1969, it's 75. Since 1993, it's 83. Since 2001, it's 79.

For fielding runs (at Baseball Reference), one standard deviation is 38-40 runs, regardless of the time period.

So, if we go with the typical spread in defense (one SD = 80 runs) and the typical spread in fielding (one SD = 40 runs), then we'd assume that the typical spread in pitching is sqrt(80^2-40^2) = 69 runs.

Therefore, the allocation between pitching and fielding would be 69/(69+40), or 63/37 split.

2013 would seem to have been anomolous in comparison to the historical numbers (through 2011).

9:44 AM Nov 22nd
 
tangotiger
As to the relief issue: I also only have Mariano Rivera as someone who would qualify as a HOF. I think it's easy enough to prove: if every player was a free agent, how much would Mo have earned in his career, how much would David Wells, and how much would Trevor Hoffman have earned? If you would pay more for Wells' career than Hoffman's, then it's hard to then suggest that Hoffman accomplished more than Wells.
9:18 AM Nov 22nd
 
tangotiger
Bill: the 2013 example is not representative of what has happened over a period if years. Otherwise, I agree that you did in fact calculate the pitching and fielding numbers appropriately. I'll see if I can come up with historical numbers.

***

To (try to) explain further, in my framework (WAR), each component (hitting, running, fielding, pitching) is compared against the average, and a final step turns that into a replacement level number (such that 0 has no value). That final step is set to that 57% of value goes to nonpitchers and 43% goes to pitchers.

Baseball Reference makes that last step so that the split is 59/41. Everyone here can make their own choice as to how to do it.

In Win Shares, Bill starts with the split first (50/38/12, or what have you, for offense/pitching/fielding), and then tries to allocate on that basis. In this approach, the "spread" in offense will be 4x the spread in fielding.

This is why we have the differences we do. In Win Shares, the spread of terrible and great fielders is much tighter than what is used in WAR.
9:14 AM Nov 22nd
 
Riceman1974
Great work once again Bill.

I would put Simmons in definitely because I believe you do have to consider the "tools of ignorance" aspect. I would also put in Quiz because I miss him, even though I'm a Red Sox fan and I used to hate his submarine delivery striking out Jim Rice and Dewey in the late innings in the mid-80s. I think he exceeds the HOF standards for a reliever, and he was one of the nicest and most erudite men in the history of the game, and yes that counts for me.

No on Garvey and Parker, sorry guys.

As for the pitching-defense debate, your skepticism about Dewan's and B-Ball Reference's runs saved numbers matches my own. I don't think defense rates anywhere near as high as their metrics indicate, mainly because if true, then player valuation (in terms of salary) has been horribly wrong for the past 130 years.

I also don't believe there is that great a discrepancy between average defensive ability and great defensive ability, as Andrelton's and Parra's runs saved numbers seem to imply. Only players who can reasonably play their position get significant innings at that position, so what is defined as "average" is determined by a handful of starters who presumably play that position well enough to not cause their pitchers to revolt. Logically, this would presumably preclude any player being that much better or worse than the average. If a defensive metric told me Andrelton saved 15 runs, I would by that; but 41? No way.

Thanks again Bill.

-Christopher


9:13 AM Nov 22nd
 
jwilt
Tango... thank you. I was so interested in Bill's take on this and your response I went back and did a search on your old Book blog to see what you'd said on the topic (prior to your comments). And despite being an engineer who's been reading sabermetrics for nearly 30 years I think I broke my brain.

Hopefully you're right, and by the 5th or 10th time reading this it'll sink in.
8:56 AM Nov 22nd
 
bjames
But I didn't use the standard deviations. I used the variances. This is actually clear if you read the subtext of the article. Read the Dewan example. . .you'll see that the conclusion is based on the variances, not the standard deviations.

I'll post on this later. I don't like to comment on my own articles until other people have a chance to do so, because I think it discourages comment. ..just trying to stave off additional comments based on a misunderstanding.
8:52 AM Nov 22nd
 
tangotiger
Correction:

"This is a simplified example. The answer is closer to 57%"

That should say 43%.
8:37 AM Nov 22nd
 
tangotiger
The split between offense and defense, or nonpitchers and pitchers, is not as clear as the "backdoor" method that Bill is trying to apply (in my system).

In my system, it works something like this:
one standard deviation in offensive runs is something like 15 runs
one standard deviation in fielding runs is something like 9 runs
one standard deviation in pitching runs is something like 12 runs

Now, here's where it gets scary: the standard deviations don't add up, but the variances (or the square of standard deviations) do.

So:
off^2 = pitch^2 + field^2
15^2 = 12^2 + 9^2

How about the split between nonpitchers and pitchers?

15^2 + 9^2 = 17.5^2

So, the nonpitchers have one standard deviation of 17.5 runs. Pitchers is 12 runs.

Hence, the "share" of value for pitchers is 12/(17.5+12) = 41%

(This is a simplified example. The answer is closer to 57%.)

What about the share between pitchers and fielders? The share of defense going to pitchers is 12/(12+9) = 57% (in this illustration).

The problem is if you try to use these splits, by adding the standard deviations (and not the variances), things don't add up. You might say that we have 15+12+9, and so pitching is 12/(15+12+9) = 33%. But, this is not true, since I just finished showing in this illustration that it was 41%.

The key point is that offense and fielding is not independent at the player level. When you add up the distributions, you have to add up the variances, not the standard deviations.

This is where the great divergence comes between my system and Bill's. It's why I give much more WAR to pitchers than Bill does.

And this is why Parker comes out as he does. It's not that I give a very small share, but rather, a very large share. And since he's negative, that large share works against him.

If whoever is reading this didn't understand it: don't worry, as I've been talking about this for ten years, and most people don't follow it the first, second, and third time. Maybe by the tenth time of re-reading it and working it out, maybe you might be swayed.


8:36 AM Nov 22nd
 
 
©2019 Be Jolly, Inc. All Rights Reserved.|Web site design and development by Americaneagle.com|Terms & Conditions|Privacy Policy