Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Hall of Merit > Discussion
Hall of Merit
— A Look at Baseball's All-Time Best

Saturday, September 29, 2007

2005 Results: Boggs Gets 100%, While Browning and Dawson Receive Hall of Merit Honors, Too!

In his first year of eligibility, legendary third baseman Wade Boggs received 100% of all possible points to become the 14th unanimous selection in Hall of Merit history (past unanimous selections include Hank Aaron, Joe DiMaggio, Lou Gehrig, Lefty Grove, Walter Johnson, Mickey Mantle, Stan Musial, Willie Mays, Babe Ruth, Mike Schmidt, Honus Wagner, Ted Williams and Cy Young).

It was a “slightly” longer wait for star batsman Pete Browning as he was inducted into the HoM in his 107th year on the ballot. He received 28% of all possible points.

Last but not least, All-Star outfielder Andre Dawson claimed the final spot for immortality in his 4th year of eligibility, narrowly besting fellow outfielders Bob Johnson and Alejandro Oms by only a handful of points (the latter two appear to be favorites to enter the HoM themselves in 2006). He earned 25% of all possible points.

Rounding out the top-ten were: Reggie Smith (huge jump!), Bucky Walters, Cannonball Dick Redding, Kirby Puckett (surprising finish after his 2004 showing) and Gavvy Cravath.

Thanks to OCF and Ron for their help with the tally.

RK   LY  Player                   PTS  Bal   1  2  3  4  5  6  7  8  9 10 11 12 13 14 15
----------------------------------------------------------------------------------------
 1  n/e  Wade Boggs              1296   54  54                                          
 2    4  Pete Browning            364   22      6  2  2     3  2  1  3     2           1
 3    7  Andre Dawson             326   23      3  2  2  1  2  1     2  4  2  2     1  1
----------------------------------------------------------------------------------------
 4    6  Bob Johnson              322   25         2  5     2  2  4     2  2     2  1  3
 5    8  Alejandro Oms            316   25      3  2  1  1     1  2  3  2     1  4  1  4
 6   14  Reggie Smith             279   20      1  2  2  2  3     1  3  1  2     1  1  1
 7   10  Bucky Walters            278   18      1  3  3  2  1  2     2  1  1  2         
 8    9  Cannonball Dick Redding  273   15      5  3  1  1     1     2  1  1            
 9    5  Kirby Puckett            269   21      2  1  3  1  1  1     3  2        2  1  4
10   13  Gavvy Cravath            261   21      2  1        2  1  3  2     4  2  1  2  1
----------------------------------------------------------------------------------------
11   12  Tony Perez               261   17      1  5  1     3     1        2  1  3      
12   11  Hugh Duffy               241   16      1  3  1  1  2  2     1  1  3     1      
13   15  Tommy Leach              239   16      1  2  1  3     3  1  2     1  2         
14  n/e  Bret Saberhagen          237   18      2  1  1  2  2     1  1  2     1  1  2  2
15   17  Luis Tiant               232   20      2     1     1  1  2  4        2  1  4  2
16   18  Graig Nettles            232   19         1     1  2  4  3  2        2     3  1
17   19  Phil Rizzuto             211   15         2  2  2  1     2  1  2  1  1     1   
18   22  Ken Singleton            207   18         2  2     1  1        1  2  3  2  2  2
19   21  George Van Haltren       191   12      2  2     1  2  1  1     1     1  1      
20   23  Bus Clarkson             189   14      2     1  4     1  1           1  1     3
21   20  Dizzy Dean               189   12      3  1  1  1  1        1     2  2         
22   16  John McGraw              187   11      3  1  2  2  1                 1        1
23   26  Tommy Bridges            159   10      1  1  4     1     1  1                 1
24   24  Mickey Welch             158   11      1  1  1  1     3  1        1     1  1   
25   29T Burleigh Grimes          153   13      1     1     3              2  2  2  2   
26   28  Dave Concepción          140   11      1     1  1     1  2     1  1  1  1     1
27   32  Larry Doyle              137   10      2     2     1        1        1     3   
28   27  Vic Willis               135    9      1  1     1  1  1  1     2  1            
29   31  Dale Murphy              134   12               2        3  2  1     1     1  2
30   29T Orlando Cepeda           133   11            3     2     1           1  2  2   
31   34  Elston Howard            130   12               3              2  2  3     1  1
32   25  Lou Brock                130    9      1  1  1        1  2     1     1  1      
33   33  Rusty Staub              126   10            1     2     1  2  3     1         
34   37  Tommy John               125    8         3  1        1        2              1
35   36  Bobby Bonds              121   10         1     2           1  1  2  1  1  1   
36   38  Bob Elliott              114   10               1     2     1  2  1  2  1      
37   35  Norm Cash                113    9            1     1  3     1     2        1   
38   39  Ben Taylor               102    8         1        1  2     1     1     1  1   
39   48  Pie Traynor               98   10            1  1     1              1  1  4  1
40   41  Carl Mays                 90    8               2     1        1  1  1  1     1
41   42  Wally Schang              87    6      1     1        1     1  1  1            
42   45  Don Newcombe              86    8                  2     1     1     1  2  1   
43   50T Lee Smith                 82    5         2     1  1                       1   
44   40  Dave Bancroft             77    7               2     1              1  2     1
45   47  Vern Stephens             70    6                  1  1  2           1        1
46T  46  Chuck Klein               67    5         1     1                 2  1         
46T  49  Rick Reuschel             67    5      1           1           2           1   
48   44  Bill Monroe               65    5               2     1  1                    1
49   95T Bert Campaneris           64    5               2  1              1        1   
50   54  Ed Williamson             64    4         1  1           1  1                  
51   52  Sal Bando                 62    5                     2  2              1      
52   50T Frank Tanana              61    4      1        2                             1
53   62T Urban Shocker             60    6                     1     2           1  2   
54   43  Don Mattingly             59    5      1                 1              2  1   
55   61  Johnny Pesky              56    6                           1  1  1     2  1   
56   55  Addie Joss                52    4      1                 1           1     1   
57   53  Thurman Munson            49    5                  1           2              2
58   69T Leroy Matlock             49    4                     2  1              1      
59   71  Jack Quinn                48    4         1                       1     2      
60   60  Ernie Lombardi            46    4               1        1        1        1   
61   59  Wilbur Cooper             46    3         1        1                 1         
62   62T George J. Burns           45    4                        1     2  1            
63   58  Tony Oliva                45    3      1           1                       1   
64   65T Lance Parrish             44    4                     1     1  1           1   
65T  56T Frank Chance              40    4                     1     1           1     1
65T  72T Al Rosen                  40    4                              2     2         
67   67  Tony Mullane              38    3                  1  1              1         
68T  56T Buddy Bell                36    3               1     1                       1
68T  69T Rabbit Maranville         36    3            1                 1        1      
70   64  Lefty Gomez               35    4                  1                    1     2
71   81  Fred Dunlap               33    4                              1  1           2
72   75  Bruce Sutter              32    3                        1  1              1   
73   74  Frank Howard              30    3                           1  1           1   
74T  65T Ed Cicotte                30    2            1           1                     
74T  76  Jimmy Ryan                30    2               1     1                        
76   77  Bobby Veach               29    3                                 2  1         
77   68  Ron Cey                   28    4                                 1           3
78   78  Jim Kaat                  25    2                        1  1                  
79   84T Jim Rice                  24    3                                 1     1     1
80   72T Jack Clark                22    2               1                             1
81   82T Luke Easter               22    1         1                                    
82   80  Sam Rice                  21    2                        1              1      
83   84T Dave Parker               18    2                                    2         
84   82T Brian Downing             17    1            1                                 
85   86  Luis Aparicio             16    1               1                              
86   79  Bill Mazeroski            15    1                  1                           
87T  93T Tommy Bond                13    1                        1                     
87T  87  Sam Leever                13    1                        1                     
87T  88T Carlos Morán              13    1                        1                     
90T  91T Tony Lazzeri              12    1                           1                  
90T n/e  Virgil Trucks             12    1                           1                  
92T  91T Fielder Jones             11    1                              1               
92T  88T Hack Wilson               11    1                              1               
92T  88T Dizzy Trout               11    1                              1               
95T 100  Dick Lundy                10    1                                 1            
95T  95T Jack Morris               10    1                                 1            
97T  93T Brett Butler               9    1                                    1         
97T  95T Mickey Vernon              9    1                                    1         
99T  99  Elmer Smith                8    1                                       1      
99T  95T Jim Fregosi                8    1                                       1      
99T 100  George Kell                8    1                                       1      
99T n/e  Dolf Luque                 8    1                                       1      
103T 100  Charlie Hough              7    1                                          1   
103T 104  Bill Madlock               7    1                                          1   
103T 100  Gene Tenace                7    1                                          1   
103T n/e  Jim Whitney                7    1                                          1   
107T 104  Dutch Leonard              6    1                                             1
107T 104  Dennis Martinez            6    1                                             1
107T 104  Al Oliver                  6    1                                             1
Ballots Cast: 54
John (You Can Call Me Grandma) Murphy Posted: September 29, 2007 at 10:34 PM | 231 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

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

Page 3 of 3 pages  < 1 2 3
   201. karlmagnus Posted: October 13, 2007 at 03:05 PM (#2574508)
I have to say I regard heavily quantitative methods such as DanR's with considerable skepticism, for two reasons. First, they don't take account of the random element, which can easily skew reaults among closely equivalent columns of figures. Second, they tend to reward players who have closely related skill sets; either "hitters" or "pitchers."

The attempt to reach consensus tends to produce a "consensus backlog" dominated by good-but-not-great players with a 1-dimensional feel to them. I would suggest that the last few HOM members should be players who combined skills, without necessarily floating to the top in heavily quantitative systems that measure 1 dimenion. Thus hitters at glove positions (Lombardi, Schang, Stephens)players who hit and pitched (Smith, Van Haltren, maybe someone off our radar screens like Kid Gleason), pitchers who could hit (Mays) or players whose career was cut short by FATE (Joss, Cicotte, Leever, FATE in three different guises) are the ones we have overlooked, because their abilities don't pop out in statisitical studies focused on one dimesnsion. As Bill James said, low similarity scores are an indication of quality.
   202. John (You Can Call Me Grandma) Murphy Posted: October 13, 2007 at 03:12 PM (#2574511)
players whose career was cut short by FATE


Fate dealt me a stinky hand by holding back the talent that I needed. John Murphy for the Hall of Merit!
   203. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 13, 2007 at 04:03 PM (#2574538)
karlmagnus, I don't know why you would say that, since my system doesn't distinguish between hitting and fielding value...I have Rogers Hornsby as one of the ten greatest MLB position players since 1893, Piazza in my top 25, Jeter in my top 100...I've even voted for Toby Harrah! Harrah seems much stronger than Stephens to me once you deflate Vern's war years.

As for pitchers who could hit, I tossed Newcombe a vote last election thinking precisely of his superb hitting (along with the low pitching stdev of the 50s NL and his being blocked, which you could consider "fate" as well).

Cicotte's case wasn't fate--he chose to throw the World Series! (That said, I've voted for Cicotte). How do you determine when to invoke your "fate" clause and when not to? John Murphy is right--fate determined that none of us were major league baseball players; that doesn't make us Meritorious. Even limiting it to guys who did reach the majors, guys' careers have been cut short for God knows how many reasons...why Sam Leever and not Tony Conigliaro? His most similar player at age 21 was Mickey Mantle and at 22 was Frank Robinson...
   204. John (You Can Call Me Grandma) Murphy Posted: October 13, 2007 at 04:59 PM (#2574559)
As for Joss, he had a bum arm. Even if he had lived another 10 years, he may not have have produced anything of quality in the majors anyway.
   205. 'zop sympathizes with the wrong ####### people Posted: October 13, 2007 at 05:07 PM (#2574565)
Two studies with similar results makes the estimate "settled law"? In the social sciences, that would qualify it as "an interesting hypothesis with some empirical support."

It should be pointed out that, from the perspective of real science (formally known as the natural sciences), social science as a whole is an interesting hypothesis with some empirical support.
   206. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 13, 2007 at 06:13 PM (#2574604)
Speaking of Conigliaro, here's a fun list: Players with at least 10 WARP2 by age 22, MLB position players from 1893-2005. Non-HoM'ers (or guys who are not a lock for the HoM) in bold:

1. Ty Cobb, 29.2
2. Mel Ott, 25.0
3. Ted Williams, 24.2
4. Arky Vaughan, 23.0
5. Joe Jackson, 21.7
6. Mickey Mantle, 20.7
7. Alex Rodríguez, 20.5
8. Jimmie Foxx, 20.3
9. Sherry Magee, 19.4
10. Ken Griffey, Jr., 18.2
11. Rogers Hornsby, 17.5
12. Eddie Mathews, 17.4
13. Stan Musial, 17.2
14. Tris Speaker, 17.1
15. Vada Pinson, 16.9
16. Al Kaline, 16.6
17. Jimmy Sheckard, 15.9
18. Rickey Henderson, 15.7
19. César Cedeño, 15.5
20. John McGraw, 15.5 (would be higher if his 1891 and 92 were included)
21. Cal Ripken, 15.1
22. Johnny Bench, 14.8
23. Frank Robinson, 14.7
24. Fred Lindstrom, 14.5
25. Albert Pujols, 13.8
26. Donie Bush, 13.7
27. Andruw Jones, 13.6
28. Stuffy McInnis, 13.5
29. Robin Yount, 13.4
30. Hank Aaron, 13.0
31. Jim Fregosi, 12.5
32. Joe DiMaggio, 12.3
33. Eddie Collins, 12.2
34. Sam Crawford, 12.2
35. Tony Conigliaro, 12.0
36. Joe Kelley, 11.8
37. Buddy Lewis, 10.7
38. Chris Speier, 10.6
39. Alan Trammell, 10.4
40. Bob Horner, 10.3
41. Joe Morgan, 10.3
42. Ross Youngs, 10.0

Some interesting names. Wow, through age 26 (1965) Pinson sure looked like HoM'er...what happened to him? He fell off a cliff just as he should have hit his peak. Cedeño's case is famous--shooting someone, even accidentally, is bad for your HoM prospects. Did McGraw quit by age 30 just so he could dedicate himself to managing full-time, or could his body just not take it anymore? As far as I'm concerned, he'd have been an *inner circle* HoM'er if he had played into his 30s at anything like the level of his 20s. Lindstrom seems to have been very up-and-down, two big years in '28 and '30 and then not much more to write home about. Donie Bush seems like a dead ringer for many great-glove, average-bat shortstops I think are Meritorious really throughout his 20s, but he fell apart with the live ball era--was it injuries or could he not adjust to the new game?

Conigliaro had a great start to his career, no doubt, but it's not like he was a LOCK for a superstar career based on what he had accomplished up to that point.

I had never heard of Buddy Lewis before he popped up on this list.
   207. Paul Wendt Posted: October 13, 2007 at 07:31 PM (#2574664)
Did McGraw quit by age 30 just so he could dedicate himself to managing full-time, or could his body just not take it anymore?

After major injuries in 1900 and 1901 his legs were in bad shape. In some respects he couldn't have been the same player; a clear case where Addie Joss is murky.
   208. EricC Posted: October 13, 2007 at 08:27 PM (#2574690)
But it can't be irrelevant--you simply have to use SOME replacement level, whether it's absolute zero (counting stats), 50% of average (Win Shares), 80% (me), or 100% (BRAA, RCAP, etc.).

I rate players in a top-down way. The "replacement level" concept plays no role in my system.
   209. rawagman Posted: October 13, 2007 at 08:50 PM (#2574702)
Me too. And I would bet that most, if not all GMs throughout history build closer to that than the reverse as well.
   210. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 13, 2007 at 09:34 PM (#2574712)
EricC, I'd appreciate it if you could explain what you mean by "a top-down way." What is your ranking methodology (assuming it is quantitative and consistent)?
   211. EricC Posted: October 13, 2007 at 09:44 PM (#2574715)
By top-down, I mean to equate (after smoothing) the observed distribution of value from the top down such that the M'th best player of year X to the N'th best player of year Y, and the 2xM'th best player to the 2xN'th best player, etc. Those who say that they aren't timeliners will probably always have M = N. I believe in timelining, because of increases in the population and the number of teams, so I have M getting larger as the years go by. I use the tail-end of a Gaussian distribution as the idealized distribution, based on a Bill James Abstract essay on how talent in baseball is not normally distributed.
   212. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 13, 2007 at 09:50 PM (#2574717)
OK, but what criteria are you using to establish that initial ranking (M'th best player of year X), before translating from one season/era to another?
   213. EricC Posted: October 15, 2007 at 10:33 AM (#2576850)
but what criteria are you using to establish that initial ranking ?

I use Win Shares per plate appearance. The system needs to use a comprehensive rate stat. In spite of the flaws of WS, at least the methodology is published in full, unlike that of WARP, which may be superior, but I can't tell.

Anyway, #211 is an attempt to put rates of performance across seasons on the same scale. To compare careers, consider a plot where the x-axis is playing time in units of full seasons, and the y-axis is the average performance over time as rated by the rate of performance metric. All consecutive season streaks of all players can be represented as points on this plot. Careers are compared, finally, by compressing the two-dimensional plot to a one-dimensional career rating, and taking the optimum result for each player.

The advantages of the system are that the units of the y-axis (i.e. replacement level) don't matter, that bad years at the beginning or end of a career won't hurt a player, that the system could be applied to other sports (think for example of golf: to compile a quantitative list of greatest golfers, one would probably not try to define a "replacement level golfer"), and that the system, imo, captures the way in which HoF voters "think".

The disadvantage, aside from the numerical uncertainties, etc,, that apply to all systems, is that the way the two-dimensional career data strength and length is projected to a single rating is arbitrary. I tried to see if HoF voting patterns defined a sharp in/out line, but, not surprisingly, the HoF doesn't define a clear "peak vs. career" criterion. An empirical but perhaps not arbitrary criterion that I could use would be to set the rating so that, e.g., the 100th best performance of all time over each length interval get the same career rating. The problem with this is that it leads to results that appear much too peak-extreme and subject to fluke excellent seasons. In practice, I'm using a compromise between the voting patterns as best defined by the HoF and the rarity of performance rate over the length interval.
   214. KJOK Posted: October 15, 2007 at 09:02 PM (#2577481)
Also, I'd like to remind people that Bill James devoted a 3-page section of his Win Shares book to arguing that win shares are not based on value above a replacement level. Since I believe he introduced the concept of replacement level to sabermetrics, I have to believe he knew what he was talking about. Therefore, I think it's misleading to keep saying that WS uses a replacement level of 50%.


Bill James is very smart, but advanced math is not one of his strong points. If Win Shares did not have a replacement level, it would be IMPOSSIBLE to get negative Win Shares (well, it is impossible, but only because James ROUNDS THEM UP TO ZERO). I think it's been mathematically shown that James used MARGINAL Wins, which by definition will have some replacement level.
   215. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 15, 2007 at 09:20 PM (#2577493)
Yeah, Brent. Win Shares sets its offensive zero point at 52% of league average. Everything above that counts towards Batting Win Shares, everything below that is given 0 (see Bill Bergen, who of course should have negative Batting WS but doesn't, leading his teammates to be incorrectly penalized). As the great Bill James himself would say, if that's not a replacement level, I'm a lug nut.
   216. jimd Posted: October 16, 2007 at 12:13 AM (#2577841)
Also, I'd like to remind people that Bill James devoted a 3-page section of his Win Shares book to arguing that win shares are not based on value above a replacement level. Since I believe he introduced the concept of replacement level to sabermetrics, I have to believe he knew what he was talking about. Therefore, I think it's misleading to keep saying that WS uses a replacement level of 50%.

Win Shares asserts that replacement players have value. That value is determined by however many wins a team of replacement players would win in a given era (it might be the same across baseball eras, it might not; James makes no comment about that). This "value below replacement" is awarded to the players who actually played; those who didn't play can't receive that value, though they could have if only the manager had played them instead.

If you believe that a replacement team should be able to play .250 ball, then that has a consequence in Win Shares. It implies that 50% of the league total Win Shares are awarded based on playing time, the other 50% is awarded based on value added over and above replacement.
   217. Brent Posted: October 16, 2007 at 05:51 PM (#2579569)
Bill James is very smart, but advanced math is not one of his strong points. If Win Shares did not have a replacement level, it would be IMPOSSIBLE to get negative Win Shares (well, it is impossible, but only because James ROUNDS THEM UP TO ZERO). I think it's been mathematically shown that James used MARGINAL Wins, which by definition will have some replacement level.

Let's see:
a) Replacement level is the zero level in some baseball measurement systems, and value below replacement level is possible
b) Win shares also has a zero level, and negative win shares are also possible (before adjustment).
c) Therefore, the zero level of win shares is replacement level.

I don't know much about advanced math, but something seems to be missing in this "proof."

Yeah, Brent. Win Shares sets its offensive zero point at 52% of league average. Everything above that counts towards Batting Win Shares, everything below that is given 0 (see Bill Bergen, who of course should have negative Batting WS but doesn't, leading his teammates to be incorrectly penalized). As the great Bill James himself would say, if that's not a replacement level, I'm a lug nut.

Rather than tell us what the great Bill James would say, why don't we read what he actually did say (p. 107 of Win Shares):

"For many years, I have argued that 'value' for a baseball player consists of being better than a replacement-level player. In fact...I hope I am not confusing the record here, but I think this was one of my ideas, this 'replacement level' concept for measuring a player's value. If somebody else was writing about this first, I apologize, but I think it was mine.

"Anyway, the Win Shares system operates on an entirelydifferent theory of value—that value consists in winning games. Two distinct theories:

"1. That value consists in being better than replacement level,

"2. That value consists in winning games for your team.

"Because I have in the past advocated theory (1), or because there is a 'marginal' concept within Win Shares, or for some other reason, some people have assumed, incorrectly, that Win Shares are a way of measuring marginal win contributions, or contributions above replacement level. Not isolated people—lots of people. There has been discussion of Win Shares on bulletin boards and in similar forums, much of which assumes that Win Shares represent marginal win contributions, and debates the system based on that assumption.

"No, no, no...this doesn't have anything to do with marginal win contributions. There is no 'replacement level' contemplated within this system. Our theory here is that value consists in winning games."

And on p. 109:

"Look, I acknowledge that there is some difference between "Wins" (or Win Shares) and "Value." In the other book, the New Historical Abstract, I didn't always rate the players according to their Career Win Shares. Win Shares are like Wins. Paul Splittorff won more games in his career than Sandy Koufax. Nobody believes that this makes Paul Splittorff a greater pitcher than Sandy Koufax. You have to consider Win Shares, Loss Shares, big-impact seasons, impact on pennant races, special contributions, etc."

There's more, but it's clear to me that (a) James understands replacement level, (b) is aware that it's possible to interpret WS as a value-above-replacement-level system with extremely low replacement level, (c) rejects that interpretation because he did not intend for replacement level to be the zero point in his system, and (d) recognizes that simply using raw WS would overstate the career value of long-career, low-value players and makes adjustments for this factor in actually evaluating players.
   218. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 16, 2007 at 06:03 PM (#2579581)
Well, if a nonzero percentage of league average offense below which no credit is given isn't a replacement level, what on earth would you prefer to call it? James's semantics make no sense here.
   219. Brent Posted: October 16, 2007 at 06:11 PM (#2579591)
Different measurement system. Different units of measurement. Why does this seem so obvious to me and yet so many smart people don't get it?
   220. Chris Cobb Posted: October 16, 2007 at 06:39 PM (#2579626)
I take Dan's claim to be that, although Bill James describes wins shares as being based on a conception of value in terms of "winning games for your team," it doesn't actually fit that description, because a team made up of players that would earn zero win shares would still win a certain amount of ball games. Win shares thus assigns a value of zero wins to players who would accumulate wins.

James sets out to justify the discrepancy between the concept (value consists in winning games) and the execution (winning games below a certain rate doesn't count) by admitting the system breaks down at the extremes but that it does a good-enough job of allocating responsibility for wins in the vast majority of actual cases.

The more I look at how win shares has worked in a lot of actual cases, the less I am convinced that it does a "good-enough" job for our purposes.
   221. sunnyday2 Posted: October 16, 2007 at 09:26 PM (#2579866)
If anyubody is voting straight WS, I would agree with that. But I doubt if anybody is doing that. Bill James wouldn't do it. He admits to using a bullshirt dump to get from the WS numbers to a ranking. Any of us is smart to do the same.

The catch is that there are systems out there that claim greater perfection, such that in theory one can vote the numbers straight up. This is where I get off. I think one needs a bullshirt dump no matter what statistical system you use as your main statistical tool and the vast majority of us have one. My sense is those who claim not to have a bullshirt dump have just buried it in the numbers in the form of this assumption or that.

Now I think James himself could do better. We've improved on WS by the simple expedient of adjusting for season length and filling in missing seasons for certain circumstances. But. The fact that James' bullshirt dump is out there in the open rather than hidden like the lunatic uncle is a good thing. It is a strength rather than a weakness. How one constructs one's own bullshirt dump of course can be a different strength or a different weakness. But when James calls attention to the fact that there is a bullshirt dump, he is doing us a favor.

And of course the actual math is transparent and the numbers don't change every quarter.

So WS may not be "good enough," but it is "as good as."
   222. KJOK Posted: October 16, 2007 at 10:10 PM (#2579929)
The more I look at how win shares has worked in a lot of actual cases, the less I am convinced that it does a "good-enough" job for our purposes.


Which is one reason why James was at one point working on "NEW" Win Shares, which were to include LOSS Shares. Not sure if he's still working on it though.
   223. KJOK Posted: October 16, 2007 at 10:18 PM (#2579936)
Different measurement system. Different units of measurement. Why does this seem so obvious to me and yet so many smart people don't get it?



If you want an answer, you'll have to read here:

Tango Tiger & Wood Win Shares Critique

and here:

Patriot's Win Shares Walk Thru
   224. Rick A. Posted: October 17, 2007 at 04:00 AM (#2580791)
BClarkson 1489

Sorry this is so late.

I've got Clarkson with 1589
   225. Rick A. Posted: October 17, 2007 at 04:01 AM (#2580792)
bump
   226. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 17, 2007 at 08:27 PM (#2581685)
Brent, if Win Shares doesn't have a replacement level, how do you explain the following phenomenon?

Career offensive totals for William Aloysius Bergen

Hits: 516
Doubles: 45
Triples: 21
Home runs: 2
Bases on balls: 88
Stolen bases: 23
Runs scored: 138
Runs batted in: 193
Batting Win Shares: 0.0
   227. John (You Can Call Me Grandma) Murphy Posted: October 17, 2007 at 08:39 PM (#2581705)
Batting Win Shares: 0.0


It's actually 0.1, Dan. :-)
   228. KJOK Posted: October 17, 2007 at 10:27 PM (#2581818)
Batting Win Shares: 0.0


It's actually 0.1, Dan. :-)


Well, without the 'elimination' of negatives, it's actually negative something or other using the Win Shares calculation.
   229. Brent Posted: October 18, 2007 at 12:50 AM (#2581882)
Brent, if Win Shares doesn't have a replacement level, how do you explain the following phenomenon?

Obviously there's a "zero point" for batting win shares (and also for pitching and fielding win shares--though one would have to be a pretty awful fielder to get zero fWS). James, however, didn't set the zero points at replacement level. They were set so that the team's win shares would add up to 3 times its wins. James recognizes that if the system were measuring value above replacement level, the zero point would need to be set higher and a team's wins above replacement would add up to fewer than 3 times its wins. He chose not to try to set the zero point at replacement level.

Here's a slightly absurd analogy--would Celsius criticize Fahrenheit for choosing a ridiculously low value for the melting point of ice because the zero point in Fahrenheit's scale is well below the zero point in the Celsius system? Fahrenheit would have simply responded that the zero point in his system was intended for another purpose. (According to Wikipedia, Fahrenheit's zero point may have been determined by the lowest point in a mixture of water, ice, and ammonium chloride.)

As folks over at the Hardball Times have shown, it may be possible to rescale win shares to get "win shares above bench," i.e., a replacement-type concept. (I'm not defending their replacement levels--just mentioning that such a rescaling has been done.)

Every time win shares comes up around here, it seems like we get a data dump of all of the criticism that's ever been written on the subject. I'm not trying to defend win shares, but when Bill James devotes 3 pages to explaining that he didn't design the system as a measure of value above replacement level, it seems to me like an unfair criticism to say "yes he did, and moreover, he blew it in picking the replacement value." Sure, one could interpret win shares as a replacement-value system with an unrealistically low replacement value, but that's not how James designed it and there is no reason to necessarily interpret it that way.

I hope that we don't need to continue this debate, because I think it must fundamentally just be a semantic misunderstanding of some sort. Apparently James (and I) think it's possible to have a system with a zero point that isn't a replacement value and you don't.
   230. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 18, 2007 at 01:04 AM (#2581890)
OK. Yes indeed, this is semantic. You are using "replacement level" to mean "the actual MLB replacement level," whereas I am using it to mean "the zero point above which a system recognizes player value," which can either be set (correctly) to roughly 80% of positional average or (incorrectly) to any other value. Case closed.
   231. DavidFoss Posted: October 22, 2007 at 05:24 PM (#2588911)
bump
Page 3 of 3 pages  < 1 2 3

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
A triple short of the cycle
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 0.3362 seconds
49 querie(s) executed