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

Sunday, September 04, 2005

1960 Ballot Discussion

1960 (September 19)—elect 2
WS W3 Rookie Name-Pos (Died)

264 103.5 1940 Hal Newhouser-P (1998)
237 61.3 1935 Phil Cavarretta-1B (living)
187 55.1 1942 Johnny Pesky-SS/3B (living)
170 55.1 1943 Allie Reynolds-P (1994)
137 55.2 1940 Ken Raffensberger-P (2002)
136 48.3 1944 Preacher Roe-P (living)
118 45.2 1947 Larry Jansen-P (living)
115 41.7 1943 Al Brazle-RP (1973)
103 47.7 1940 Sid Hudson-P (living)
121 35.6 1945 Cass Michaels-2B (1982)
118 27.6 1943 Johnny Wyrostek-RF (1986)
094 38.1 1942 Dave Koslo-P (1975)
044 16.0 19?? Connie Marrero-P (living)

1960 (September 11)—elect 2
HF% Career Name-pos (born) BJ – MVP - All-Star

HF 34-54 Leon Day-P (1916) 2.5 - 3*
00% 31-54 Silvio Garcia-SS/3B/P (1914) #9 ss 0 - 1*

Players Passing Away in 1959
HoMers
Age Elected

84 1922 Nap Lajoie-2b
78 1920 Ed Walsh-P

Candidates
Age Eligible

90 1911 Boileryard Clarke-C
85 1914 Ed McFarland-C
85 1917 Roy Thomas-CF
82 1917 Wid Conroy-3b
79 1921 Hooks Wiltse-P
77 1927 Dode Paskert-CF
76 1921 John Hummel-2b/UT
72 1929 Tilly Walker-LF/CF
68 1934 Joe Harris-1B
68 1935 Ken Williams-LF
67 1935 Jack Scott-P
66 1926 Lefty Williams-P
64 1935 Howard Ehmke-P
62 1949 Biz Mackey-C
59 1943 Jim Bottomley-1B
53 1950 Johnny Allen-P

Thanks to Dan and Chris!

John (You Can Call Me Grandma) Murphy Posted: September 04, 2005 at 08:06 PM | 228 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 2 of 3 pages  < 1 2 3 > 
   101. KJOK Posted: September 12, 2005 at 11:24 PM (#1614119)
He seems to indicate that AAA baseball is actually 75% Of MLB. I take it this is in wins or runs, not in AVG/SLG.

Most studies I remember came in around 80-82% using Runs/Wins for AAA to MLB....
   102. Howie Menckel Posted: September 13, 2005 at 01:24 AM (#1614354)
No changes here thru 1926..

HOMers by year (10 G min) thru 1959
1856-59 - 1
1860-65 - 2 (3 1864)
1866-67 - 4
1868-71 - 6/8/9/10
1872-76 - 12
1877-78 - 11
1879-80 - 16/17
1881-84 - 20 to 22
1885-89 - 23 to 25
1890-92 - 29 to 30
1893 ---- 26
1894-03 - 21 to 22 (20 1900)
1904-07 - 24 to 25
1908-15 - 26 to 28 (25 1912)
1916 ---- 31
1917 ---- 26
1918-21 - 22 to 25
1922 ---- 27
1923 ---- 31
1924-25 - 35/37

1926-27 - 40
1928-30 - 37/35/34
1931-32 - 38
1933-36 - 35 (33 1935)
1937 ---- 34
1938-40 - 28
and so on downward for now...


Those late 1920s and the 1930s are pretty popular.
   103. Dr. Chaleeko Posted: September 13, 2005 at 01:43 AM (#1614401)
Tom and Jschmeagol,

My bad on the AA citation. You're right I was thinking about the AVG/SLG conversion rates and not about win values when I said that.

But on the other hand, a pennant's a pennant's a pennant. And if Burns is timelined down by 12% or more, then what about Pete Browning? Is he to be timelined into oblivion?

Something doesn't seem right. (Although that something could be my logic!) And, of course, I'm not inspired when BP's website says one thing about Burns's career WARPs and Clay says something else. I'm not trying to bash BP or anything because they are one of my favorites, I'm just saying it's hard to believe from my POV.
   104. sunnyday2 Posted: September 13, 2005 at 01:52 AM (#1614434)
>Those late 1920s and the 1930s are pretty popular.

How many of those are NeLers? I would guess it is in part due to the earliest influx of NeL candidates, as compared to previous years.
   105. Mark Shirk (jsch) Posted: September 13, 2005 at 02:02 AM (#1614457)
I replied to Clay asking him about teh discrepancies. 73.4% seems way too low, where 88% seems to me to be on the low end of believable. It was nice to find out how he does it and how much the average at that time affects players.
   106. Howie Menckel Posted: September 13, 2005 at 12:51 PM (#1615052)
HOM's Negro Leaguers, 1920-37
1901----- 2
1902-09 - 3 or 4
1910-19 - 6 or 7 (8 in 1916)
1920-22 - 8 or 9
1923-30 - 12 or 13
1931 - 14
1932-37 - 9 to 11
   107. Howie Menckel Posted: September 13, 2005 at 12:52 PM (#1615053)
of course, that's "1901-37" in the header.
   108. DanG Posted: September 13, 2005 at 02:44 PM (#1615187)
HoMers debuting on each ballot, 114 total.

1898 14
1899 6
1900 2
1901 1
1902 3
1903 2
1904 1
1905 1
1906 0
1907 1
1908 0
1909 2
1910 0
1911 2
1912 0
1913 0
1914 3
1915 2
1916 2
1917 2
1918 0
1919 1
1920 2
1921 1
1922 3
1923 4
1924 0
1925 1
1926 1
1927 1
1928 1
1929 0
1930 0
1931 0
1932 1
1933 3
1934 7
1935 1
1936 2
1937 0
1938 0
1939 1
1940 2
1941 3
1942 1
1943 4
1944 2
1945 0
1946 3
1947 3
1948 2
1949 1
1950 3
1951 1
1952 3
1953 4
1954 1
1955 2
1956 1
1957 2
1958 0
1959 2
   109. Paul Wendt Posted: September 13, 2005 at 03:59 PM (#1615310)
1922 3
1923 4
1924 0
1925 1
1926 1
1927 1
1928 1
1929 0
1930 0
1931 0
1932 1
1933 3
1934 7

And Dan Greenia predicted it, fostering a new and then inconspicuous malady, looking ahead ;-)

The letter from Clay Davenport really belongs elsewhere. Is there any way follow up can be directed elsewhere? Nah.
   110. sunnyday2 Posted: September 13, 2005 at 04:31 PM (#1615353)
Summary:

1898-1899 (retire by 1894, primes 1865-1890)--20
1900-1909 (retirements 1895-1905, primes 1885-1900)--13
1910-1919 (retire 1905-1915, primes 1895-1910)--12
1920-1929 (retire 1915-1925, primes 1905-1920)--14
1930-1939 (retire 1925-1935, primes 1915-1930)--15
1940-1949 (retire 1935-1945, primes 1925-1940)--21
1950-1959 (retire 1945-1955, primes 1935-1950)--19

Of course we are not done with the 1950s eligibles, nor any other cohort, though I would guess we will elect more backlog from the '50s than the '40s, '40s than '30s, etc. etc. etc.

Nice symmetry of the first four groups representing 1865-1935. Too many from 1925-1950? Or about the way we wanted it?
   111. Mike Webber Posted: September 13, 2005 at 04:43 PM (#1615376)
As best friend of Roush, I figure it is my duty to do due diligence on whether I am way out of line. I stole Chris Cobb's Charts and added Roush, career only.

Medwick   Averill   Johnson   Sisler    Roush
8142 PA   7215 PA   8047 PA   9013 PA   8309 PA
134 OPS
+  133 OPS+  138 OPS+  124 OPS+  126 OPS+
1984 g    1668 g    1863 g    2055 g    2002 g
.300 EQA  .303 EQA  .308 EQA  .291 EQA  .294 EQA
1 FRAA    7 FRAA    14 FRAA   2 FRAA    7 FRAA
254 FRAR  317 FRAR  256 FRAR  215 FRAR  430 FRAR
--        --        --        33 PRAR   --
624 BRAR  560 BRAR  677 BRAR  596 BRAR  605 BRAR
267.4 bws 223.9 bws 250.8 bws 260.2 bws 260 bws
44.4 fws  54.5 fws  36.6 fws  24.6 fws  61 fws
1899 adj
g#1761 adj. g1848 adj. g2075 adj. g
311.8 ws  308.4 adjws287.4 ws 300 adjws 321 adjws
(add 100 RAR)                 (add 25 RAR)(add 25 RAR

22.99     25.61     21.40     19.97     23.18  WS/600 PA 


so some disclaimers:
I am not sure I got all the changes exactly the way Chris did them, and the adjusted games I just skipped because I knew I didn't understand that one.

Why just career and not 5 and 10 year? Well I worked thru the 5 year, and what is showed is on rate metrics he is right there, but because he played in 40 to 90 games less than the other guys, even with a WWI adj, his totals don't stack up.

Plus I feel a little guilty using Bpro metrics since I don't really believe in them.

That being said, I think Roush is still the player I'd rather have over Averill. Though it is paper thin close. I think Medwick's peak probably slides him in front of the centerfielders.

Also, while phantom credit is spread out all over the electorate, Roush doen't get any for the times he held out. I'm not really advocating he should, but hey some guys give it due to late career starts, or being dead young, or being chucked out of civilized baseball as a cheat. Him staying home until June to get a good contract doesn't mean he was any less of a player.

One last note to those that use WARP3, in his five year peak, when the NL was supposedly very weak, just like George Burns, Roush get hit with a 26% dicount from WARP1 to WARP3. Wow that is pretty steep.
   112. Chris Cobb Posted: September 13, 2005 at 05:04 PM (#1615429)
Mike,

Could you give us a summary of the playing time that Roush missed for holding out? I'm not sure what I will do with that info, but it would at least help clarify my image of him if I could distinguish time missed due to injury from time missed due to contract disputes.
   113. TomH Posted: September 13, 2005 at 05:26 PM (#1615486)
I don't think using a % discount from WARP1 to WARP3 is the best measure.

A great hitter like Dan Brouthers only loses 24%, while if Ken (24 career WARP) Reitz had put up his numbers in 1900, he probably woudl have lsot 100% or more of his WARP value.

I'd suggest 'WARP (wins) per year of fulltime play' would more descriptive.

Here are some WARP1-to-WARP3 conversions of EqA, essentially changes in batting average, for some players from 1880 to 1970:

Anson -17 (Cap loses 17 pts of EqA going from raw to league-strength adjusted)
Browning -31
Van Haltren -12
Wagner -11
Cobb -9
Hornsby -11
Gehrig -7
Appling -3
Vaughan 0
Gordon -1
Kiner 0
Mantle -6
Mays +1

Browning isn't really being 'timelined' as much as he is being 'American Association adjusted'

Are these numbers too big? You tell me. The diff from Mantle's 1950s to the early 1900s sure ain't much at all!
   114. sunnyday2 Posted: September 13, 2005 at 05:34 PM (#1615500)
So the AL in Appling's time was better than the AL of Mantle's time?

Kiner's NL was better Mantle's AL by the same margin that Wagner's or Hornsby's NL was better than Anson's?

I been tellin' you guys to vote for Kiner!
   115. Mike Webber Posted: September 13, 2005 at 05:47 PM (#1615520)
Roush out of the lineup:

1916 – July 20, Traded to Reds. Had played full time in Fed Lg year before, bought along with Kauff and 3 others by NYG. Quoted as hating McGraw, 71 PA before trade 300 after. Cincy was pretty awful though, and NYG had Kauff in CF.

1921 - ?? played just 112 games. Reds fan out there know?

1923 – Held out until July 23, age 29 season. Played 49 games, 9 win shares.

1928 – Tore abdominal muscles, played 46 games.

1930 – Held out entire season rather than play for McGraw after a $7,500 pay cut. Age 37 season, 15 win shares the year before, 5 the next year.

Plus the Reds missed about 20% of 1918 and 10% of 1919, though I think everyone here makes some allowance for that.

Apparently rarely came to spring training on time, which isn’t a good thing. Maybe he didn’t need it, but if he did he didn’t give himself the opportunity.
   116. Chris Cobb Posted: September 13, 2005 at 05:56 PM (#1615541)
Thanks, Mike!

I hope other info about 1921 will turn up.
   117. Michael Bass Posted: September 13, 2005 at 06:21 PM (#1615597)
That Mantle's AL was poor should not come as any sort of surprise. The AL was much slower to integrate than the NL, and suffered greatly for it. It would not surprise me to find that 50s/early 60s AL was the worst AL/NL league of the 20th century in comparison to its partner for that reason.

Also, while I have absolutely no numbers to back this up, it's my impression that the AL of this era had more bottom-feeder franchises than the NL. This contributes to the issue.
   118. sunnyday2 Posted: September 13, 2005 at 06:28 PM (#1615609)
But the comparison was Appling's AL and Mantle's AL. Appling's was integrated?
   119. DanG Posted: September 13, 2005 at 06:31 PM (#1615614)
From the SABR Bioproject entry for Roush:

During the 1920-1921 off-seaons Edd was injured when his brother Fred accidentally plunked Edd with birdshot, one hit each to the lip, cheek and thumb. That December Baseball Magazine stated that Roush was the greatest outfielder in the National League: "In ground covering, he has no superiors and few approximate equals, while he was a fine a base runner as ever and hit for the grand average of .352."
Edd Roush Bioproject
   120. Michael Bass Posted: September 13, 2005 at 06:31 PM (#1615617)
Oh sure...I can definately see the issues with that. Just saying that Mantle's AL certainly is worthy of bottom of the barrel treatment.

I wonder if there isn't something to be said for my other point though. The 50s seemed to be something of a nadir for competitiveness in the AL. Is it possible the non-Yankee teams just gave up for the most part?
   121. TomH Posted: September 13, 2005 at 06:40 PM (#1615634)
The numbers surprise me too. The barely-integrated NL of the late 40s superior to the somewhat-intgerated AL of the late 50s? Does not seem right. We'll have to deal with that in another month or two.

There is some downturn for the Korean War years, during which some guys played. By the way, anyone know why Hank Aaron played through the Korean conflict?

Accounting for the annual Yankee drubbing of the NL in October would change the 1950s ##s a bit, no?
   122. Dr. Chaleeko Posted: September 13, 2005 at 07:41 PM (#1615746)
GVH's -12 league-quality rating should be sending a number of voters into howling fits. Different members of the electorate have at various times opined that the 1890s NL is underrepresented and was superior in quality to those NLs and ALs (and FLs!) that followed them due to being a smaller, condensed league, drawing on roughly the same talent pool.

WARPs timeline/quality adjustment is not consistent in many respects. Expanding on the Appling/Mantle example above, Arky Vaughan's NL is a 0 league. But he played before integration too. Were I doing any kind of reasonable thought experiment, would I really believe that the quality of play from 1930-1945 was consistent with today's quality of play? Of course not, but that's apparently not what WARP is measuring.

If I get the drift of Clay's email to jschmeagol, then WARP is directly measuring run/win values, not quality of play. However, the underlying assumption appears to be by directly measuring R/Wins that WARP is indirectly measuring quality of play.

But as we're seeing this assumption can't possibly be accurate, and as someone here pointed out some time ago, you can't consider questions of quality of play without considering timeline because the two form a gordion knot of league-quality issues.

So to me the answer seems kind of simple: look at WS or WARP1, cross-check with the other, and timeline judiciously.
   123. Mike Webber Posted: September 13, 2005 at 08:49 PM (#1615888)
Edd Roush 1921

I asked Greg Rhodes, a fellow SABR member if he knew... here is a little more info. Below is his reply.

"Mike: According to my book Redleg Journal, Roush held out until May 1. I don't know how soon he actually started playing since he only missed 3 weeks with the holdout, which would have been about 15 games. 112 games played means he missed 40 games, I don't know how to account for the rest."

Greg Rhodes
Executive Director Hall of Fame and Museum
Cincinnati Reds
   124. DavidFoss Posted: September 13, 2005 at 09:40 PM (#1616022)
There is some downturn for the Korean War years, during which some guys played. By the way, anyone know why Hank Aaron played through the Korean conflict?

He was only 19 when the war ended. How young were they drafting? He's about three years younger than Mays and four years younger than Ford.
   125. DavidFoss Posted: September 13, 2005 at 09:49 PM (#1616052)
I wonder if there isn't something to be said for my other point though. The 50s seemed to be something of a nadir for competitiveness in the AL. Is it possible the non-Yankee teams just gave up for the most part?

Chicago and Cleveland had solid teams for most of the decade. I think the lower division teams were quite weak, though. Particularly the A's, Senators and Browns/Orioles. The 1954 AL was particularly non-competitive after the top three.
   126. Al Peterson Posted: September 13, 2005 at 09:53 PM (#1616067)
Doc stated:

Were I doing any kind of reasonable thought experiment, would I really believe that the quality of play from 1930-1945 was consistent with today's quality of play? Of course not, but that's apparently not what WARP is measuring.

I wouldn't be so fast to dismiss the quality of ML baseball the time up til integration when compared to to say only a few years later, the 1950s. Major advancements in 1946 in pro sports was the addition of two leagues - AAFC (football) and NBA. Adding these leagues to the NFL, and the increased popularity of these alternatives to playing baseball, was the beginning of the end as "Baseball is King".

And don't even get me started about NASCAR :)
   127. TomH Posted: September 13, 2005 at 11:15 PM (#1616229)
a better way to word the Aaron question would have been "does anyone know why Hank never missed any time for military service?" Was he 4-F for some odd reason? Did he serve at an extremely young age (I doubt it, since he played ball before entering the majors)?
   128. Chris Cobb Posted: September 14, 2005 at 01:57 AM (#1616598)
On Aaron and military service: He was definitely not 4-F, and he did not serve in the military.

I don't recall the source (and I'm sure there are folks on this site who know the details better than I), but I remember reading the Aaron received some communication from the draft board while he was playing for Eau Claire, to which he did not respond, putting himself in some legal jeopardy. When the team discovered this, they asked him what he was doing, and he basically said that it wasn't his business to be involved with that. The team gave him guidance on his legal obligations, and he was (iirc) simply not drafted. There was a bit more to the story on how the whole situation was handled, but I just can't recall the details. Someone who's reading this thread surely knows the whole, true story! . . .
   129. karlmagnus Posted: September 14, 2005 at 12:09 PM (#1616872)
Tom H, interesting data. Where does Beckley fall? He played through three very separate eras, generally on one of the weaker teams, thus improving the quality of his opposition.
   130. TomH Posted: September 14, 2005 at 01:21 PM (#1616911)
adding Beckley and a few other top players on our ballot to the timeline:

Pike -26
Anson -17
Browning -31
....Beckley -11
Van Haltren -12
....Jennings -11
Wagner -11
Cobb -9
....Sisler -7
Hornsby -11
Gehrig -7
....Sewell -8
Appling -3
....Averill -6
Vaughan 0
....Medwick 0
Gordon -1
Kiner 0
Mantle -6
Mays +1
Bonds 0
Jeter +11 (!)
apparently the AL is perceived as much stronger than the NL recently!

By this measure, there isn't much timelining going on between our current top players, except that Medwick gets a WARP3 boost from the 'perceived as strong' NL of the 1930s. As others have pointed out, the farm system concept (B Rickey) took hold earlier in the NL, lending credence to this notion.
   131. karlmagnus Posted: September 14, 2005 at 01:44 PM (#1616943)
Jeter gets +11 for "clutchness" and willingess to appear in commercials with George Steinbrenner! Apart from that, interesting data indeed.
   132. sunnyday2 Posted: September 14, 2005 at 01:57 PM (#1616961)
Let's see:

Hornsby's competition was 4 points weaker than Sisler's and Gehrig's.

Mantle's was 7 points weaker than Mays' and 6 points weaker than Kiner's.

Vaughan's was 8 points better than Sewell's and 3 points better than Appling's.

And Jeter's is 11 points better than Bonds'? My god.

Not to mention, Jeter's is 22 points better than Honus Wagner's. I guess the AL in the 21st century pretty much represents the greatest baseball ever played on planet earth. Ptolemy would be thrilled.
   133. Mark Shirk (jsch) Posted: September 14, 2005 at 02:05 PM (#1616977)
A response from Davenport about the mix up concerning George Burns WARP3:

Ack. No, the problem was on my end; rather than looking on the web, I
ran a new card for him, not realizing I had turned off part of the
adjustments. So, from the top:

Burns' EQR total drops from 1188 to 1083 due to the difficulty
adjustment, a drop of 9%. His RAR drops from +491 to +386 (-21%),
because the RAR baseline does not move with the adjustments. By either
measure, he lost 105 runs.

His fielding RAR drops from 384 to 229, primarily because the total
fielding RAR for an average team drops from ~300 in the late teens to
about 190 in the standardized output, although the difficulty
adjustment also applies here and takes out about 15 runs.

Clay
   134. TomH Posted: September 14, 2005 at 02:10 PM (#1616986)
re: Jeter/Bonds competition, I could play the skeptic as well, but

a. I haven't studied the ##s as BP has
b. given the AL dominance in W.S. play (14 of last 21 winners) and all-star games (15 of 21), and the tendnecy for big-sending AL teams to possibly garner the top free agents,

it is not implausible that the AL has been the better league in recent times.

Anyone have the interleguae play W-L records? While I'm not a big fan of that system, it probably does at least give the best insights about relative league strength.
   135. Chris Cobb Posted: September 14, 2005 at 02:12 PM (#1616990)
Burns' EQR total drops from 1188 to 1083 due to the difficulty adjustment, a drop of 9%. His RAR drops from +491 to +386 (-21%), because the RAR baseline does not move with the adjustments. By either measure, he lost 105 runs.

This is a really important thing to notice in understanding the WARP1-WARP2 adjustments, which I have only been grasping myself lately as I've compared the 1910s outfield cohort to the 1930s cohort.

There's 1) a difficulty adjustment and 2) an adjustment of replacment level. The first moves the baseline of what is average. The second moves the distance between average and replacment level. 1910s outfielders like Veach and Burns have lower peak EQAs than 1930s outfielders like Medwick and Averill, but they earn as many or more BRAR because of the lower replacement level in their era in the WARP system.
   136. Mark Shirk (jsch) Posted: September 14, 2005 at 02:14 PM (#1616993)
One thing that struck me from Davenport's first email was that he makes an adjustment based on the average player from that position. For instance, the average LFer during Kiner's career was 17.1 runs above replacement in a given season, this seems to have been lower than in some eras. For instance, George Burns' contemporaries were 27 runs above replacement.

Davenport uses 15 runs above replacement as his average and adjusts everyone to that average. You can see how this would affect Burns more than Kiner. Is this why Kiner's Eqa or EQR aren't hurt as bad as some others? Because the replacement level for LFers was higher during this time?

Could this affect Jeter in comparison to Bonds? Jeter may very well have played his career in a golden age for SS's, where the replacement level is higher forcing the average for the position down when compared to replacement.

We think of Mantle playing in an era with a lot of god CFers, but with the exception of Doby , weren't they all in the NL? Mays, Snider, Ashburn, etc. Could the replacement level for CFers during Mantle's time been pretty lower for the AL?
   137. sunnyday2 Posted: September 14, 2005 at 02:40 PM (#1617042)
>it is not implausible that the AL has been the better league in recent times.

I don't think anybody would say that the AL isn't better, much less that such a statement would be implausible.

The question is: 11 points? When the ML in Mays' day was 7 points better and in Cobb's day 2 points better? This just seems a bit extreme.

Of course the real question now is how much extra credit to give Ralph Kiner!?
   138. Thane of Bagarth Posted: September 14, 2005 at 02:56 PM (#1617074)
And Jeter's is 11 points better than Bonds'? My god.

A large part of the difference between the AL and NL post-1972 is due to the DH. Probably something to do with the DH changing the quality of the average (and replacement) hitter in the AL.

The AL "difficulty rating" jumps by 0.104 from '72 to '73. If you subtract that amount from the ratings around the turn of the millenium, the leagues are much closer.

Here are the ratings for both leagues from 1972-1974 and 2000-2002.
AL     NL
1971 0.939  0.974
1972 0.943  0.984
1973 1.047  0.976
1974 1.043  0.991
.
.
.
2000 1.115  0.988
2001 1.097  0.999
2002 1.087  1.010 
   139. sunnyday2 Posted: September 14, 2005 at 03:17 PM (#1617123)
Thane, great catch. The AL went from being about .04 weaker than the NL in 1972 to .07 better in 1973. The AL's rating compared to the NL:

1971 -.035
1972 -.041
1973 +.071
1974 +.052

2000 +.127
2001 +.098
2002 +.077

The changes from one year to the next are .006, .019, .029 and .021. Oh and then there's that change of .112.

I think you are right. The formula has responded to a change in the structure of the game rather than to a change in actual difficulty or competition level, and it has never been corrected. If the change in competition could be isolated from the change in structure, it would probably be a little different than .104 or .112. The average change in competition in the other 4 years that are available in the above data is about .018. So the structural change is likely to be somewhere in the .086 - .130 range (we don't know if it moved with or counter to the competition change).

So IOW all competition adjustments since 1973 ought to be questioned.

But the larger question is if this sort of (alleged) error can creep in, how credible are all the other competition adjustments pre-1973?
   140. Mike Webber Posted: September 14, 2005 at 03:35 PM (#1617168)
Does anyone else ever get the feeling when they looks too closely at the BPro stuff that they are seconds away from hearing,

"I am the great and powerful Oz. Pay no attention to that man behind the curtain!"
   141. Chris Cobb Posted: September 14, 2005 at 03:52 PM (#1617193)
One thing that struck me from Davenport's first email was that he makes an adjustment based on the average player from that position. For instance, the average LFer during Kiner's career was 17.1 runs above replacement in a given season, this seems to have been lower than in some eras. For instance, George Burns' contemporaries were 27 runs above replacement.

Davenport uses 15 runs above replacement as his average and adjusts everyone to that average. You can see how this would affect Burns more than Kiner. Is this why Kiner's Eqa or EQR aren't hurt as bad as some others? Because the replacement level for LFers was higher during this time?


These RAR refer to defensive value, not offensive value. They don't have anything to do with EQA, which is not calculated by position, but for all hitters together.

Here's how value in WARP1-2 is calculated, as I understand it.

1) Player's performance relative to league average as a hitter and, at his position, as a fielder, is calculated. This is expressed for batting as EQA and for fielding as Rate1.

2) When applied to playing time, these become BRAA and FRAA.

3) batting and fielding replacement levels for that season are calculated, and the BRAR and FRAR values for a player's BRAA and FRAA are assigned. We now have the player's WARP1.

4) The relative value of average for that season in an all-time context is calculated, and BRAA and FRAA totals are adjusted accordingly to get an all-time figure.

5) A fixed "all-time replacement level" is set, and all-time BRAR and FRAR are calculated off of the player's all-time BRAA and FRAA. For fielding, this "all-time replacement" is not only establishing quality of play, but the relative defensive importance of the position. We now have the player's WARP2.

6) Adjust for season length. We now have the player's WARP3.

To go back to the example of Burns, taking 1917 as the year, we have the following adjustments:

Batting WARP1
EQA .314 in 152 g --> 46 BRAA --> 65 BRAR

to WARP2
all-time adjustment of average value
.314 EQA --> .303 EQA. 46 BRAA-->35 BRAA
all-time average is 3.5% higher than average in NL 1917

all-time adjustment of replacment level
65 BRAR-->55 BRAR
Virtually no change: the distance between an average hitter and a replacement level hitter in NL 1917 is about the same as the all-time replacement level. The drop in Burns' BRAR is due entirely to his drop in BRAA for this year.

A quick examination of a variety of records suggests that the distance between brar and braa is historically constant: I don't see any examples of players who lose runs here when adjusted for all time as well as in the adjustment of the value of average itself.

Fielding
Rate1 108 in 152 g --> 12 FRAA --> 40 FRAR

To WARP2
All-time adustment of average value
108 Rate1 --> 107 Rate 2; 12 FRAA --> 11 FRAA
all-time average is 1% higher than 1917 average for LF

All-time adjustment of replacement level
40 FRAR--> 24 FRAR
This is the doozy. Fielding replacement level is 46% higher for LF all time than in NL 1917. How much of this adjustment has to do with ability and how much this has to do with the defensive importance of the position is not distinguished by the WARP system, but both factors could be in play. However, since the BRAR-BRAA distance seems fixed, I suspect that most of the difference has to do with the changing defensive importance of the position.

It's worthwhile thinking through all this partly because WARP makes all of these adjustments explicitly and separately, whereas win shares does not. In fact, I would describe win shares as sort of a hybrid of WARP1 and WARP2, with its fielding rates more-or-less pegged to an all-time context of positional value, but with some give in them to reflect changing circumstances.
   142. Dr. Chaleeko Posted: September 14, 2005 at 03:57 PM (#1617200)
Mike,

Despite my recent heavy criticism of BP's WARP2/3 adjustments, I feel like you might be overstating the case a little bit. I agree that there are very large chinks in their armor, but they are honest and reasonable attempts to answer questions of league quality. I don't think that they're misrepresenting themselves.

And they do change and tweak the formulae quite frequently, so it's not like a "Brownie, you're doin' a heckuva' job" kind of thing where they ignore deficiences to put a good face on things.

Similarly, it's not as though WS doesn't have it's issues, nor again that James hasn't said as much nor claimed there are fixes he'd like to make. The biggest (and most frustrating) difference between James and BP is that BP doesn't publish out its methodology. That said, Clay Davenport is responsive to questions (as we've seen) and doesn't gloss anything over, so I can't fault him, even if I disagree with some of the outcomes of WARP2/3.
   143. Chris Cobb Posted: September 14, 2005 at 04:01 PM (#1617207)
I think you are right. The formula has responded to a change in the structure of the game rather than to a change in actual difficulty or competition level, and it has never been corrected. If the change in competition could be isolated from the change in structure, it would probably be a little different than .104 or .112.

I think this "change in structure" vs. "change in competition level" is a false distinction. If your assessment of value starts with an assessment of a player in relation to league average, removing pitchers from the batting order and replacing them with DHs is going to have a _permanent_ effect on the level of a league average hitter. Therefore, the change in struccture produces a permanent change in competition level.

We've already discussed the potential impact of the DH on batting win shares, and we have been concerned about whether batting win shares in the DH-AL and the non-DH NL will mean the same thing. What's the evidence that WARP's adjustment for the DH is incorrect? Wouldn't trading in 10% of the league's at bats taken by sub-replacement-level hitters with hitters who were formerly league average raise the level of average by at least 10%?
   144. Mark Shirk (jsch) Posted: September 14, 2005 at 04:03 PM (#1617210)
Okay, I thought he was taling about RARP, or teh EQR stat used to show offense above replacement level.

The DH thing is a good catch. However, how should this be adjusted? Aren't Jeter, ARod, Manny, and Vlad getting hurt a little by playing with a DH when there is no adjustment? Should they be penalized for this as they are in Win Shares?

I would say the best thing to do would be to use not plain offensive levels for the NL, but non pitcher offense. At least when moving form WARP1 to the higher WARP's.

Though at the same time when there are only eight real hitters, each hitter is more valuable. On the other hand, players shouldn't get penalized for playing in one league or the other.

Any thoughts?
   145. karlmagnus Posted: September 14, 2005 at 04:16 PM (#1617247)
It would seem to me that the replacement value should be the same between a DH and a non-DH league. A Bonds whose fielding becomes too atrocious can move from the NL to the AL. If hitting replacement value was higher in the AL, then a light hitting player who was below replacement value in the AL could catch on in the NL. In such a case, there would since 1973 be numerous players who washed out of the AL and made lengthy if undistinguished careers in the NL, and none (other than the odd non-fielding Ortiz) the other way round. If this is not the case, then hitting replacement values must be the same in both leagues.

The W/L in interleague play is an important data point on league quality, but not definitive, becasue it is likely to be more difficult for AL teams to make pitchers bat than for NL teams to insert their best pinch hitter or 4th outfielder as DH.The worst hitting puitchers should naturally congregate in the AL, and even if they don't, lack of practice would atrophy their skills.
   146. Dr. Chaleeko Posted: September 14, 2005 at 04:24 PM (#1617269)
Any interleague W/L records would tend to be small in sample size and would also tend to be influenced mightily by scheduling and "natural rivalries" like NY/NY and CHW/CHC, etc....
   147. sunnyday2 Posted: September 14, 2005 at 04:28 PM (#1617275)
>Therefore, the change in struccture produces a permanent change in competition level.

Yes but the question is how to compare the competition level of one league with one structure vs. another league with a different structure.

The implication of Bonds 0 and Jeter +11 is (not that Jeter is a better hitter, but that) the AL is a better league, just as Cobb played in a better league than Wagner and Mays played in a better league than Mantle.

Either 1) I am confused and am drawing an inference that everybody else understands should not be drawn, or 2) several of us are confused, or 3) there is an adjustment for structure that should be made when drawing comparisons with leagues (contemporary NL, AL before 1973) with a different structure.
   148. sunnyday2 Posted: September 14, 2005 at 04:30 PM (#1617278)
Interleague play is still a better data point that World Series and All-Star games, no?

Or is Bonds 0 Jeter +11 a better data point than all of the above combined?
   149. Mark Shirk (jsch) Posted: September 14, 2005 at 05:18 PM (#1617384)
I don't like interleague play because the schedules always have some matchups and the sample size is small enough that if say, the Yankees dominated the Mets for five years it would throw things off.

I wonder if the difference isn't so much that Jeter gets getting aboost that Bonds is not but that Jeter, ARod, et all have their Eqa's artificially low compared to their NL contemporaries. The Jeter +11 thing may just be evening things out.

Karl,

You would admit that there is a higher hitting replacement level right? Eqa, the stats where Jeter is +11 and Bonds is 0, is an offense only stat. Therefore the AL would have a higher replacement level in this metric. It would be similar is OPS+ had league strength adjustments as well.

Your question about does teh NL have a higher or lower replacement level is tricky. The DH brings up the offensive replacement level without lowering the defensive one. Maybe the replacement level for pitching is lower since pitchers aren't hitting. Does that balance out the higher hitting replacement level? I don't know. If I had to guess I would say that the AL has the higher replacement, slightly, due to the DH.
   150. KJOK Posted: September 14, 2005 at 05:38 PM (#1617437)
The DH thing is a good catch. However, how should this be adjusted? Aren't Jeter, ARod, Manny, and Vlad getting hurt a little by playing with a DH when there is no adjustment? Should they be penalized for this as they are in Win Shares?

Offensively, the replacement level will be a little higher in the AL, but in theory this would be EXACTLY offset by the defensive replacement level being the corresponding amount of wins/runs/something lower in the AL.
   151. karlmagnus Posted: September 14, 2005 at 05:45 PM (#1617457)
Why does the DH raise the offensive replacement level in the AL and not the NL? There are more jobs for hitters in the AL, but if the hitter replacement level was higher there would be a flow of hitters from AL AAA to NL majors, where they could get jobs. The AL "bucket" of hitters is 1/8 bigger, but the levels should be the same. The flows of hitters who can't field and pitchers who can't hit from the NL to the AL should be balanced by a flow of banjo hitters and medicore pitchers who hit well from the AL to the NL. This would equalise the replacement level, even though the (non-pitching) average hitting level in the AL was higher, as was the average pitching level.

I think I come down once again to scrapping fancy metrics and sticking with simple ones, as I'm not sure the problem with WS/WARP in this case is correctible.
   152. sunnyday2 Posted: September 14, 2005 at 06:01 PM (#1617523)
The implication of Bonds 0 Jeter +11 is (thanks, again, to Thane's interesting discovery) is that the AL became a better league when it adopted the DH.

Not only did it become a better league that it previously had been, it also became a better league than the NL. Just because it adopted to DH. This does not stand up to common sense.

As I said before, either 1) I am confused and this is not what the numbers mean (though I think they have been widely taken to mean as much). Or, 2), this is a really flawed metric. One or the other.

I understand that you add 11 points to Jeter's BA to get a fair comparison to Bonds' BA. But if Jeter moved to the NL, would he actually hit for 11 extra BA points? Is he facing tougher pitching in the AL? That has nothing to do with the DH.

Do I add X percent to his raw WS or WARP totals because of this, in order to get a fair comparison with Bonds' WS or WARP totals? This is in fact what WARP2 does, no?

Like I said, this conceptually makes no sense to me. Maybe that's just me. Or maybe it is in fact nonsense. None of the above has clarified it to me in the least. I don't care about replacement level, I care about comparing Derek Jeter to Barry Bonds (or whatever) when it comes time to cast a vote. How does this data inform that choice?
   153. Michael Bass Posted: September 14, 2005 at 06:36 PM (#1617601)
The DH, as mentioned before, is going to present some tough issues with the uberstats.

As Marc reasonably points out, Jeter is not actually going to hit any better* in the NL than the AL. So WARP3's adjustment seems bad. However, raw WARP1/WS do not work with the DH either, because replacement level is higher in a DH league than a non-DH league. For example, in Jeter's case, if he performs exactly the same in the NL as in the AL, he will get more WS/WARP1 in the NL case. This seems unfair, no?

Perhaps WARP3's adjustment isn't downgrading Bonds in relation to Jeter, it's just putting Jeter on equal footing replacement level-wise with Bonds. Most likely, WARP2 goes too far with this, as it does with timelining.
   154. Dr. Chaleeko Posted: September 14, 2005 at 06:54 PM (#1617648)
in Jeter's case, if he performs exactly the same in the NL as in the AL, he will get more WS/WARP1 in the NL case. This seems unfair, no?

If I understand WS correctly, it's basis for comparing players is to tally up RC and allocate credit for a team's total RC (or RS) based on how big the RC pie is and how big a piece the hitter is responsible for. (With adjustments for park and whatnot, of course.)

So for WS, Jeter's league is less important than his team. For example, if Jeter moves from the Yankees to the Giants, and the teams score the same number of runs, WS will credit him with more WS because the park effects in SF are pretty offense-suppressive. Now if Jeter puts up the same numbers, goes to the Giants, AND they score fewer runs, his WS will shoot upward because he's now responsible for a greater share of the team's offensive production. On the other hand, if he goes to Colorado and puts up the same numbers, WS will whack him down to size through the combination of a nice fat park adjustment and a bigger piece of pie to slice from, for which Jeter is a smaller piece than in either NY or in our fiction Giants example.

If I understand WS (a big if), this is its primary innovation, the conversion of RC to wins on a team basis, not at a league-wide basis.

Or maybe I've got it all wrong. Either way, if I understand WARP1, it is taking the approach that it is the league, not the team, against which a player must be compared, thereby using a mythical replacement and average player as its yardsticks for determing R/Win, instead of team performance.

It's a pretty big difference in approach when you think about it, and it gives you two different looks at the same guy.
   155. Michael Bass Posted: September 14, 2005 at 07:00 PM (#1617664)
This one I can actually speak with authority on. :)

The marginal runs (replacement level) are assigned on the team basis, yes. You won't really see much difference between NL and AL teams here.

Where the DH has a huge effect is in the dividing up of those marginal runs. Where NL teams basically divide up the runs 8 ways, AL teams divide them up 9 ways. About the same marginal runs + 1 extra regular to get a piece of them = AL players hosed.
   156. Chris Cobb Posted: September 14, 2005 at 07:22 PM (#1617708)
Not only did it become a better league that it previously had been, it also became a better league than the NL. Just because it adopted to DH. This does not stand up to common sense.

As I said before, either 1) I am confused and this is not what the numbers mean (though I think they have been widely taken to mean as much). Or, 2), this is a really flawed metric. One or the other.

I understand that you add 11 points to Jeter's BA to get a fair comparison to Bonds' BA. But if Jeter moved to the NL, would he actually hit for 11 extra BA points? Is he facing tougher pitching in the AL? That has nothing to do with the DH.


His batting average would not simply rise by 11 points, because it's a different set of factors that establish the overall offensive level of a league--parks, pitchers, weather, etc. But his OPS+, say, _if it were calculated without removing pitchers from the hitting pool_, would certainly rise significantly if he were to move from the AL to the NL. EQA, like OPS+, is a metric that is normalized to league-average batting value.

It's my belief, at this point, that the WARP2 increase for the DH-era AL is the necessary correction for their _not_ removing pitchers from the batting data in all the rest of baseball history. They provide EQA and BRAA and BRAR for all pitchers in baseball history, so my guess is that their basic formulas don't drop pitchers out of their calculation pool for batting value, so that they have to make a corrrection to those values in a league in which pitchers don't hit, just as we would have to make some adjustment when comparing AL to NL OPS+ if the baseline for NL OPS+ included pitchers hitting.

We've seen that OPS+ values jump even back in the 1920s and 1930s if we calculate them from complete league OBP and SA values, rather than from pitchers-removed OBP and SA values.

Offensive metrics that remove pitchers from the offensive pool will be increasingly important to us when we reach the DH years, esp. if Karlmagnus is correct (and I think he probably is), that batting replacement level for position players is the same in both leagues.


Where the DH has a huge effect is in the dividing up of those marginal runs. Where NL teams basically divide up the runs 8 ways, AL teams divide them up 9 ways. About the same marginal runs + 1 extra regular to get a piece of them = AL players hosed.

I think this is right, but I think that the effect is mitigated by James' decision to remove negative batting value from pitchers' pitching win shares, rather than adding it into the team's batting win shares. This is another example, I think, of the win shares system's hybrid approach. Rather than going for a technically precise solution that would lead to major divergence between the meaning of batting win shares in the AL and the NL post-dh and then attempting to correct for that (as WARP does) James gives win shares a mixed approach that doesn't make the problem go away but that diminishes its significance.
   157. Chris Cobb Posted: September 14, 2005 at 07:25 PM (#1617711)
Offensively, the replacement level will be a little higher in the AL, but in theory this would be EXACTLY offset by the defensive replacement level being the corresponding amount of wins/runs/something lower in the AL.

KJOK, could you explain why this would in theory be true? It's not immediately obvious to me.
   158. Dr. Chaleeko Posted: September 14, 2005 at 07:30 PM (#1617723)
AL players hosed.
I'm not sure I'd say AL players are hosed by any means. They get their fair share of the offensive pie. As BJ would probably say it, WS is an accounting system, it allocates credits.

There's 9-15 guys contributing to the offense, each guy gets his share of the credit based on how much offense is produced.

In the NL there's 8-15 guys, they get their share of the credit based on how much offense is produced.

Because NL teams have a pitcher batting, they will likely produce less park-adjusted offense en toto than an AL team. So NL teams have a smaller pool of offense to divide, and a smaller number of regulars to divide it among, which should offset a certain amount of any hosing that might be going on.

Take a look at the last ten years
YEAR AL R/G  NL R/G  DIFF/162G
------------------------------
2004 5.01    4.64    59.9
2003 4.86    4.61    40.5
2002 4.81    4.45    58.3
2001 4.86    4.70    25.9
2000 5.30    5.00    48.6
1999 5.18    5.00    29.2
1998 5.01    4.60    66.4
1997 4.93    4.60    53.5
1996 5.93    4.68   115.5
1995 5.06    4.63    66.7

Roughly 50 runs per team difference over 162 games.

If a replacement hitter is somewhere around 50-60 R per year, then the net effect is to give AL team an additional replacement-level hitter's worth of credit to divvy up (which is OK in WS because it doesn't have a replacement level). This makes sense from an NL point of view in that pitchers, who are considerably below replacement, do not accrue a full season's worth of #9-slot PAs because of frequent pinch-hitting and double-switching. And, unless Dusty runs your team, then most PHs will be above replacement to balance out the pitchers somewhat.

So a smaller pool of offense for the NLs, but fewer guys to give credit to, and a bigger pool of offense and more guys to give credit to for the ALs.
   159. Michael Bass Posted: September 14, 2005 at 07:39 PM (#1617743)
That's an inaccurate description, Doc. The total amount of WS given for the average AL team and the average NL team are essentially equal. The pool of offensive WS is the same for both leagues, because different offensive replacement levels are calculated between leagues (this is necessary so as not to punish AL pitchers, who obviously will have higher ERAs than their NL counterparts).

As one example, using 2003 Wins Shares, NL teams averaged 118 offensive WS/team, while AL teams averaged 115 offensive WS/team (normal variations mean they won't be exactly equal, so don't take the NL averaging more to indicate anything).
   160. karlmagnus Posted: September 14, 2005 at 07:40 PM (#1617745)
Supposing the Red Sox kidnapped Selig, and arranged for the Yankees to be transferred to the National League East. Jeter would put up stats that were close to the same, or very marginally better because he wasn't facing all the great non-hitting AL pitchers, or very marginally worse because that damn Pedro would be able to remain in games for 8 innings due to being able to coast against pitchers batting.

Which? And what would happen to his WS?
   161. Michael Bass Posted: September 14, 2005 at 07:43 PM (#1617756)
Actually, I'm not making my point well enough I think.

Win Shares absolutely has a replacement level. It's just really low. Teams are given credit for marginal runs above 52% of "expected", and for marginal runs below 152% of "expected". And the reason that AL hitters suffer is that "expected" is calculated using the individual league environment, not major league baseball as a whole. Thus the baseline for AL hitting credit is higher than that for NL hitting credit (while the baseline for pitching AL credit is lower).
   162. Chris Cobb Posted: September 14, 2005 at 07:43 PM (#1617757)
A higher scoring league does not mean more batting win shares to go around, because win shares _does_ have a zero point: 1/2 the league average of runs scored. Batters on a team get win shares for creating marginal runs created above this zero point. Their share of team win shares depends on how much they exceed league marginal runs scored versus how much the defense beats league marginal runs allowed (1 1/2x league avg. runs allowed). As the league rs rises, the zero point rises with it.
   163. Michael Bass Posted: September 14, 2005 at 07:45 PM (#1617762)
Supposing the Red Sox kidnapped Selig, and arranged for the Yankees to be transferred to the National League East. Jeter would put up stats that were close to the same, or very marginally better because he wasn't facing all the great non-hitting AL pitchers, or very marginally worse because that damn Pedro would be able to remain in games for 8 innings due to being able to coast against pitchers batting.

Which? And what would happen to his WS?


No idea, though I'm sure the effect is small either way.

And absolutely, if the Yankees were transferred to the NL, and performed exactly the same, with the exception of no DH, the remianing Yankee hitter WS would rise.
   164. sunnyday2 Posted: September 14, 2005 at 07:45 PM (#1617763)
Chris and all, I am struggling to understand all of this, but I think my original comment is true.

When I see Wagner -11 and Cobb -9, I see that Cobb played in a little bit better league.

When I see Bonds 0 Jeter +11 I am seeing something different. I am seeing the effect of the DH. I am probably also seeing that the AL is a better league, but because there is noise in the equation, I am left to say, well, yes, the AL is a better league and the World Series and All-Star games prove it. OK. But we set out to "prove" league competitiveness with Wagner -11 Cobb -9 Bonds 0 Jeter +11, I think we have admitted that we can't. There's a very big noise in there.
   165. Dr. Chaleeko Posted: September 14, 2005 at 07:53 PM (#1617778)
SD2,

Agreed, and so I fall back to the premise I wrote dozens of posts ago (and who surely many others have posted long before I did): use WARP1, use WS, don't use WARP2/3.
   166. Chris Cobb Posted: September 14, 2005 at 08:03 PM (#1617807)
But we set out to "prove" league competitiveness with Wagner -11 Cobb -9 Bonds 0 Jeter +11, I think we have admitted that we can't. There's a very big noise in there.

Agreed. What we're seeing is primarily the effect of the DH, not an effect of league quality. I think that the DH adjustment is necessary, but the WARP system unfortunately does not enable us to distinguish a league quality adjustment from a structural adjustment in this case.
   167. sunnyday2 Posted: September 14, 2005 at 08:11 PM (#1617817)
Doc, I used WS.

I used to use WARP1, never WARP3 for obvious reasons. Also, as someone said somewhere up above (can't find it now), WARP is always fiddling with and changing the numbers. Except it was said above as a positive attribute. Sorry, if it's a worthy system, you'd think it would be at least moderatly stable instead of wildly unstable. So I don't use it at all.

I rather like ERA+, OPS+ and probably some of the numbers yest and karl use to supplement. I'm gonna have to go tall all lower case.
   168. Mark Shirk (jsch) Posted: September 14, 2005 at 09:18 PM (#1617955)
Karl,

WARP doesn't take pitcher hitting out of teh hitting pool for the NL, giving that league a lower replacement level because the pitchers are involved. Pitcher dont' usually hit better than replacement so it drags down the replacement level. Thi is how the AL has the higher replacement level.

I guess you could also argue that DH's are most likely to be above replacement level hitters, which would still give the AL a slight advantage, but that is more theory than fact.

Everyone else,

The run scoring environment in the AL is higher than that of the NL correct? If the run scoring environment is higher than each run scored or created is worth less than than in the NL. This is because of the DH.

WS does distinguish between leagues so if the NL run enviroment were say 4.6 runs per game and the AL 4.8, it doesn't matter. What matters is the league scoring, not the MLB scoring. That evens things out a bit between teh leagues, making the potential offensive WS equal in both leagues. Now the AL players are splitting WS 9 different ways, the NL players are splitting it 8 ways (more like 8.2 or something with pitchers involved). This is a disadvantage for AL hitters.

From the team perspective this evens out, there are 162 games in both leagues and the average team will still get 243 WS in a season. However, the AL practicaly has a full-time player getting a lot of at bats added to the pool, so the individual position player most likely loses a little bit.

I don't know if I agree with this or not. One the one hand, I think AL players get screwed a bit. On the other, James has a point, if each run is worth less in a certain league than it is, no extra credit should be given.

WS is about the actual, which is why it can be 'distorted' by pythag, the DH, etc. WARP is about the theoretical.

sorry about the typing mistakes, but I don't have time to edit this.
   169. Mark Shirk (jsch) Posted: September 14, 2005 at 09:22 PM (#1617968)
I also have no problem with what WARP does. It is not like Davenport is pulling things out of his ass.

They have made the decision to count pitcher hitting in their replacement level evaluations throughout baseball history. This benefits NL players in WARP1. In order to make this up, AL players get a boost. Their average player is not as far above replacement level b/c that level is higher.

So when these players numbers are translated into the replacement level that Clay is using for WARP2/3 the NL players get hurt, or more accurately the AL players are recieveing a boost.

I think that using WARP1 on AL players form the modern era would be misleading since they are getting hurt by WARP's higher replacement level for offense. WARP3 corrects for this.
   170. sunnyday2 Posted: September 14, 2005 at 09:26 PM (#1617976)
Well, WARP3 injects other stuff in there, too.
   171. sunnyday2 Posted: September 14, 2005 at 11:25 PM (#1618200)
Well, we've got a Ralph Kiner discussion going here a year early. Why not Vern Stephens.

http://www.baseballthinkfactory.org/files/primer/
   172. sunnyday2 Posted: September 14, 2005 at 11:26 PM (#1618204)
Or you can go right direct to the source

http://www2.presstelegram.com/sports/ci_3026148
   173. jimd Posted: September 15, 2005 at 01:22 AM (#1618603)
The Jeter +11 thing is being over-interpreted by some. (long)

The implication of Bonds 0 and Jeter +11 is (not that Jeter is a better hitter, but that) the AL is a better league

Not precisely. The implication is that it is more difficult to be an above-league-average hitter in the AL than in the NL. When both leagues had similar structures, the inference that the AL was better than the NL (or vice-versa) appears to be acceptable from that implication. That inference is no longer valid post-DH, until one knows how much of the adjustment accounts for the DH.

To establish that, take a particular (neutral park) batting line and convert it to NL EQA and to AL EQA. (EQA uses league averages in its calculations.) The difference will give an idea of how much of the adjustment is "structural" (DH) and how much is due to "league quality".

The implication of Bonds 0 Jeter +11 is (thanks, again, to Thane's interesting discovery) is that the AL became a better league when it adopted the DH.

Not really. The implication is that it became more difficult to be an above-league-average hitter. When the pitchers stopped taking their AB's, they were replaced with real hitters, which caused the (WARP-1) EQA's of the established players to go down. IOW, the overall quality of league hitting went up.

Also, the league comparison became more difficult. Removing the pitching AB's from NL hitting is a good first step. But should pinch-hitters for pitchers also be removed? (The NL would use more pinch-hitters that might be of lesser quality than a starter; do they really reflect "league quality"?) Should DH at-bats be removed from the AL? (They would tend to hit better than the average player because they have no defensive responsibility; should that be included in the league quality measurement?) Trying to compare the leagues offensively becomes much more difficult.

Jeter is not actually going to hit any better* in the NL than the AL

That depends on the "real" difference in league quality.

However, the same stat line (compiled in a park with the same factor) will yield a higher EQA in the NL because the NL averages contain a significant chunk of pitcher at-bats. So Jeter will have a higher EQA in the NL with the same raw stats.

but I think that the effect is mitigated by James' decision to remove negative batting value from pitchers' pitching win shares,

Sorry, but it isn't. AL hitters still have to split the OWS 9 ways, while NL hitters split it up 8.x ways where x varies depending on how many pitching AB's get used by pinch-hitters. AL hitters get less Win Shares for similar stats. If negative Win Shares are used for the pitchers, the problem gets worse because pitching Win Shares get converted into OWS in the non-DH league.

So NL teams have a smaller pool of offense to divide, and a smaller number of regulars to divide it among, which should offset a certain amount of any hosing that might be going on.

It doesn't. They have the same number of wins as AL teams do. Deadball teams earned the same number of OWS as 1930's teams, it just took more runs to get one win share in the 30's. Same thing here; AL teams get less win shares for the same offensive production because it produces less wins.

Some will then argue that the increased offensive production is less valuable. And they MIGHT have a point.

James said in the BJNHBA that Win Shares were split 50-50. Because offense and defense are balanced. But then the Win Shares book said it's really 48-52, because the numbers come out better that way. But as some have pointed out, the DH changes the game structurally. Maybe after the DH, it's still 48-52. OTOH, maybe it's now 52-48. How would we know? How can we tell? Particularly due to the fact we have no explanation for the underlying rationale other than esthetics for 50-50 and "It causes problems if you don't" for 48-52.

Because there is no formula for determining this from real baseball stats, we don't know if the balance subtly changed and whether a different formula needs to be used for DH leagues.
   174. sunnyday2 Posted: September 15, 2005 at 01:41 AM (#1618688)
I think that is what I said. The numbers we've been using to determine league strength are no longer useful to that purpose after 1972 because of the DH noise.

OTOH the numbers still work for another 17 years of HoM voting!
   175. Mark Shirk (jsch) Posted: September 15, 2005 at 02:35 AM (#1618943)
Jim,

just ot make sure, your point about Jeter is thus...

He wouldn't necessarily hit beter in the NL than the AL b/c the pitching and defense and parks aren't necessarily going to be better or worse.

BUT

His usual .315/.390/.450 (or whatever it is) would be more VALUABLE in the NL because the lack of DH changes the run environment. Hence, his Eqa would be higher. Jeter does nothing different.

Correct?
   176. Mark Shirk (jsch) Posted: September 15, 2005 at 02:39 AM (#1618958)
sunny,

I think you are correct, we will have to change what we look at for league strength between leagues and probably between eras for the DH era.

The problem is that there is a league strength adjustment (NL to Al) in WARP, but it isn't singled out in the DT cards so we have no real clue how to pick it out.
   177. KJOK Posted: September 15, 2005 at 04:34 AM (#1619105)
Offensively, the replacement level will be a little higher in the AL, but in theory this would be EXACTLY offset by the defensive replacement level being the corresponding amount of wins/runs/something lower in the AL.

KJOK, could you explain why this would in theory be true? It's not immediately obvious to me.

Not sure I can explain it well, but
in the AL, the defensive replacement level is lower due to the DH. It's the reason why a Mike Piazza, when he can no longer catch, might still be able to play in the AL. Players who would be below replacement level overall in the NL (and out of a job) due to their defense/lack of a position can still be above replacement level (and on a roster) in the AL just so long as their offense is above replacement level. The NL may have a Lenny Harris or Dave Magadan, but that will outweighed in the AL by the David Ortiz' and Edgar Martinez' type players, plus a Greg Luzinski or Orlando Cepeda at the end of their careers. Those types of players in effect "migrate towards" the AL, driving the offensive replacement level up vs. the NL, but (assuming 'perfect' talent distribution between the leagues) drive the defensive replacement level down vs the NL as the opposite type of players (Neifi Perez?) will migrate towards the NL roster spots.
   178. KJOK Posted: September 15, 2005 at 04:36 AM (#1619107)
His usual .315/.390/.450 (or whatever it is) would be more VALUABLE in the NL because the lack of DH changes the run environment. Hence, his Eqa would be higher. Jeter does nothing different.

Correct?


Unless we're all really missing the boat here, this is 100% correct.
   179. KJOK Posted: September 15, 2005 at 04:40 AM (#1619112)
but I think that the effect is mitigated by James' decision to remove negative batting value from pitchers' pitching win shares,

Sorry, but it isn't. AL hitters still have to split the OWS 9 ways, while NL hitters split it up 8.x ways where x varies depending on how many pitching AB's get used by pinch-hitters. AL hitters get less Win Shares for similar stats. If negative Win Shares are used for the pitchers, the problem gets worse because pitching Win Shares get converted into OWS in the non-DH league.


I believe Joe Dimino and I had this discussion many elections ago, and after reviewing it I came to the conclusion that Joe was correct when he argued that the AL hitters are "disadvantaged" in the DH era in the accumulation of Batting Win Shares. Just like the above discussion of Jeter and EQA, a hitter with the "same" batting results is more valuable to his team in the NL than in the AL, and this will be reflected in BOTH Win Shares and WARP.
   180. Paul Wendt Posted: September 15, 2005 at 05:09 AM (#1619144)
Jeter +11 !
Among the numbers TomH inferred from WARP1-2-3 differences, that one has attracted a lot of attention but take a look at this series.

Jennings -11
Wagner -11
Hornsby -11
Vaughan 0
Kiner 0


Nothing gradual about that one, forty years before the DH.

Now I lay me down to sleep, thinking of Ralph Kiner in a distinguished line of shortstops.
Thinking o
   181. Paul Wendt Posted: September 15, 2005 at 05:21 AM (#1619158)
a hitter with the "same" batting results is more valuable to his team in the NL than in the AL, and this will be reflected in BOTH Win Shares and WARP.

Right. But not in TPR or WARA (A for Average).

If but 16 players showed up at the Elysian Fields, the Knickerbockers chose two teams of 8. If 20 showed up, they chose two teams of 10. They divided the same three win shares 16 ways on one weekend and 20 ways on the next.
   182. Dr. Chaleeko Posted: September 15, 2005 at 01:40 PM (#1619413)
His usual .315/.390/.450 (or whatever it is) would be more VALUABLE in the NL because the lack of DH changes the run environment. Hence, his Eqa would be higher. Jeter does nothing different.

Agreed, but I would maintain that Jeter is not getting hosed by this at all---he's getting exactly what's due him. His contribution remains the same, but his percentage of the team's total offense in a DH league changes. Just because it's tougher to accumulate value over average or replacement in the post-1972 AL doesn't change the fact that his contribution to an offense, and therefore to the team's wins, is no more or less than the number of runs he creates relative to the total runs scored by his team.

I don't see a problem with Jeter receiving less credit for his offense in an environment where his offense is less meaningful. It's not really different than, say, Hack Wilson's 56 homers being worth less in the 1930 NL than they would have been in the 1915 NL. Or the 1988 NL. It's a question of context: if Jeter hits .315/.380/.450 in a league that hits .315/.380/.450 with or without a DH, he's still just an average hitter in that league. And he should receive credit accordingly.
   183. Mark Shirk (jsch) Posted: September 15, 2005 at 01:47 PM (#1619422)
I see your point Doc. That was essentially what Bill James saidin the Win Shares book.

However, I think there is a difference between Jeter in in a DH league and Wilson in 1930. The conditions of 1930 helped Wilson to reach 56 home runs and 190 RBI whereas the conditions of say, 2005 in a DH legaue don't help Derek Jeter 'achieve' more.
   184. Chris Cobb Posted: September 15, 2005 at 01:52 PM (#1619428)
But, Doc, what do you do when you rank Jeter against Larkin? Larkin's probably the greater player anyway, but would you be comfortable concluding that Larkin is the more meritorious player if it appears that the difference in their value was due to the fact that Larkin played in the non-dh league and Jeter in the dh, such that, if you had dropped them into opposite leagues, their ranking would switch?

That seems highly arbitrary to me.

If we're trying simply to determine how wins are created, win shares and warp1 are accurate. But if we're trying to assess merit, I think we have to take context into account.
   185. Dr. Chaleeko Posted: September 15, 2005 at 02:04 PM (#1619451)
Chris,

Great question. Part of the reason we use these uberstats is that value to team is a big part of how we determine merit. So in a partial sense, yeah, I'm going to line up Jeter and Larkin and say, "Look x was worth more wins to his team than y."

But I think this is where other tools, like the Keltner list, will be especially helpful...especially since we'll know the players so well. One of the Keltner questions is "was this player the best in his league or in the majors at his position?" another is "how many all-star-type seasons did this player have?"

WS and WARP help us answer these questions, and OPS+ and other measures along with the bullshirt dump (I love this term, Sunnyday2) help us get at these questions too.

Of course in Jeter's case, the most important aspect in deliberating whether he'll be on any particular voter's ballot or not is whether the voter is a Sox fan or a Yankee fan.... That's the difference between #1 placement and #100 placement.

; )
   186. sunnyday2 Posted: September 15, 2005 at 02:15 PM (#1619467)
Great discussion. What I've learned:

1) Measures of relative *league* strength will be a lot harder after 1973. But I really don't pay any attention to that anyway. A pennant is a pennant.

2) The bigger issue, which I didn't initially recognize was on the table, is that comparing individual players after the DH (e.g. Jeter and Larkin) is harder, too, and not just because we don't know the relative strength of their leagues, but because they are contributing in different environments. Not just differently strong or weak environments, but environments that are different in kind. This is much harder than AA vs. NL in the 1880s.

But it seems to me that, in another sense, this is no big deal at all. After all, every player in ML history has had a differential opportunity for whatever reason, though I will go back to my old dichotomy here.

a) Those things that affect the opportunity of individual players like Cravath and Averill and Sam Leever and whomever are largely not my concern. #### happens.

b) Those that affect entire cohorts are my concern. Black players kept out of the MLs by racial discrimination. A whole cohort going off to war. The day those players were born, they were doomed to lose opportunities in a systematic way. That is my concern.

And it's true that the DH is an example of #b and therefore it is my concern. The DH distributes offensive opportunity differently (among 9 players instead of 8). How to account for this?

But this is I think less of an issue than pitcher usage patterns over the years. When every pitcher who ever lived was born, they were "doomed" to a particular type of usage (at a macro level) and to a usage pattern that is/was differential to pitchers of other eras. Modern pitching, beginning before the DH, started to see the systematic use of relief pitchers. How to account for this, both for relievers and for starters? And before the DH most teams were migrating from 4 to 5 man rotations. How to account for this, especially when comparing, say, Don Sutton, to a pitcher from the 2 man or 3 man days, much less the 4?

So, anyway, good discussion. It is our place to understand the distribution of opportunity. Maybe we should rate players not by any raw totals but by the percentage of total value they accumulate relative to opportunity. Maybe that is how to do Larkin vs. Jeter. I don't know. But this is not a unique case at a macro level. It is just another day in the salt mine of HoM.
   187. Dr. Chaleeko Posted: September 15, 2005 at 02:35 PM (#1619520)
Sunnyday2,

Nice recap, and I think we're all likely to come back to it very soon...like in 1962 when the first of the interesting relievers hit the ballot: Ellis Kinder.
   188. Chris Cobb Posted: September 15, 2005 at 03:16 PM (#1619608)
More on Hitters

I’ve expanded my slice-of-career hitter study from Medwick, Averill, Johnson, and Sisler to include the leading 1910s major-league outfielder candidates: Cravath, Roush, Veach, and Burns. These candidates are doing quite a bit less well than the leading 1930s candidates, so I was curious to see if that is justified by the WS and WARP1 metrics (we know what W3 will show us . . . ). For Cravath, I have set his MLE career as running from 1906 to 1920. I’ve used Brent’s MLEs and Dr. Chaleeko’s estimated win shares for those minor-league seasons for which Cravath should receive major-league credit.

All of these players were active during WW1, so as I did with Sisler in the previous posts, I have pro-rated their 1918 and 1919 seasons to 154 games. All PA and g totals affected by this expansion are marked with an asterisk. All WARP and WS counting stats have been prorated to fit these expanded seasons.



Five-year consecutive peak

Cravath         Roush          Veach        Burns
1913-17         1917-21        1915-19      1916-20
2941 PA         2901 PA*       3296 PA*     3402 PA*
723 g           679 g*         762 g*       757 g*
161 OPS+        149 OPS+       143 OPS+     132 OPS+
.321 EQA        .312 EQA       .303 EQA     .300 EQA
-3 FRAA         -4 FRAA        49 FRAA      59 FRAA
81 FRAR         160 FRAR       173 FRAR     182 FRAR
303 BRAR        276 BRAR       273 BRAR     267 BRAR
130.2 bws       121.8 bws      122.0 bws    125.4 bws
12.7 fws        21.7 fws       21.6 fws     19.8 fws


Rank by WS total
Burns 145.2
Veach 143.6
Roush 143.5
Cravath 142.9

Rank by WARP1 total
Burns 449
Veach 446
Roush 436
Cravath 384

There’s remarkably little difference among these four players in five-year consecutive peak! Cravath suffers for his defense in WARP1, but otherwise there’s little basis for choosing between these guys.

I would rank them Roush, Veach/Burns, Cravath.

Roush gets the nod because his rates are so much better even though his raw totals trail Veach/Burns slightly. V & B are truly indistinguishable. Cravath trails slightly for weak defense that slightly overbalances his edge with the stick.

Integrating with Medwick, Averill, Johnson, Sisler, by five-year peak.

By total win shares
Medwick 155.3
Burns 145.2
Veach 143.6
Roush 143.6
Cravath 142.9
Averill 142.6
Sisler 141.5
Johnson 108.4

By total WARP1
Sisler 492
Averill/Burns 449
Veach 446
Roush 436
Medwick 430
Cravath 384
Johnson 368

WS pops Medwick out of the pack, WARP pops Sisler out of the pack. Johnson trails in both; Cravath trails in WARP1.

Ten-year consecutive prime

Cravath         Roush          Veach        Burns
1908-17         1917-26        1914-23      1913-22
5532 PA         5422 PA*       6303 PA*     6856 PA*
1388 g          1265 g*        1484 g*      1521 g*
152 OPS+        141 OPS+       132 OPS+     122 OPS+
-- EQA          .306 EQA       .296 EQA     .291 EQA
-- FRAA         -11 FRAA       111 FRAA     72 FRAA
-- FRAR         287 FRAR       346 FRAR     331 FRAR
-- BRAR         520 BRAR       462 BRAR     455 BRAR
230.6 bws       201.9 bws      197.8 bws    223.1 bws
24.9 fws        42.3 fws       42.1 fws     38.9 fws


Rank by WS total
Burns 262.0
Cravath 255.5
Roush 244.2
Veach 239.9

Rank by WARP1 total
Veach 808
Roush 807
Burns 786
Cravath?

The WARP/WS views are hard to reconcile here. Win share’s high evaluation of Burns’ offense is not backed up by OPS+ or by EQA. Cravath would gain on the pack in WARP1, I think, but not enough to move out of the #4 spot. Roush has the best rates, but trails significantly in playing time. I guess I rank them Roush, Cravath, Veach, Burns, primarily because (1) Roush’s rates are so much stronger in both metrics, (2) I just don’t see a basis for the WS view of Burns’ offensive value, and (3) I’m skeptical that Cravath’s defense really should count as heavily against him as WARP makes it.

Integrating this group with the other:

By WS totals
Averill 266.9
Burns 262.0
Cravath 255.5
Medwick 248.5
Roush 244.2
Veach 239.9
Sisler 236.4
Johnson 217.7

By WARP1 totals
Averill 857
Veach 808
Roush 897
Burns 786
Cravath c. 750??
Sisler 745
Johnson 711
Medwick 694

The high quality of Averill’s 10-year prime is reaffirmed here in both systems. Otherwise it’s a mixed bag. Considering the election this year, it’s notable, however, that both WARP1 and WS prefer Cravath and Burns to Medwick over a ten-year period (if Cravath is given MLE credit).

Career comparisons will follow in the next post.
   189. Chris Cobb Posted: September 15, 2005 at 03:19 PM (#1619614)
Here are the career comparisons.

Cravath         Roush          Veach        Burns
7370 PA         8156 PA        7557 PA      8251 PA
147 OPS+        126 OPS+       127 OPS+     114 OPS+
1909 G          1967 g         1821 g       1853 g
-- EQA          .294 EQA       .292 EQA     .285 EQA
-- FRAA         7 FRAA         102 FRAA     84 FRAA
-- FRAR         430 FRAR       378 FRAR     384 FRAR
-- BRAR         580 BRAR       511 BRAR     491 BRAR
290.4 bws       255.7 bws     219.1 bws     241.3 bws
32.0 fws        58.3 fws      47.0 fws      46.9 fws
1851 adj. g     1931 adj. g   1810 adj. g   1887 adj. g
326.2 adj. ws   321.7 adj. ws 273.0 adj. ws 296.1 adj. ws
(848 est RAR)  (add 24 RAR)  (add 20 RAR)  (add 23 RAR)


Career value by win shares
Cravath
Roush
Burns
Veach

Career value by WARP1
Roush 1034
Veach 909
Burns 898
Cravath 848

Clearly, Cravath is the candidate here whose position needs careful consideration because of the differences in the two systems’ view of him. Cravath’s 20-point advantage in career OPS+ more than makes up for his fielding shortcomings in win shares’ reckoning, with minor-league credit given for 1906-7 and 1909-11. That would probably not be true in WARP, though one can’t say for sure. WARP likes Cravath’s offense less than WS does, and it gives more weight to defense, so Cravath’s below-average defense hurts him more in WARP. Also, since he is a right fielder, he is playing a position WARP1 sees as distinctly less valuable than LF in the deadball era. I have taken a shot at estimating Cravath’s career value in WARP simply by prorating his WARP1 totals from 1220 ML games to 1909 ML eq. Games. His OPS+ drops only slightly from his ML career to his ML eq. career (150 to 147) and I suspect his added footspeed in his earlier seasons (when he had good sb totals in the AA), might balance out that slight OPS+ drop in EQA. WARP also never likes his fielding, so I felt comfortable extrapolating his late rate to his whole career. Even if this sells his fielding a bit short, his fielding value would probably not be high enough to move him up to Burns in WARP1 career value,

The other three are more straightforward. Both systems take Roush over Veach and Burns, both by total and by career rate, though if Roush were docked for his FL play, rate would be close between him and Veach in WARP or him and Burns in ws.

Integrating this group with Medwick/Averill/Johnson/Sisler

By career win shares
Cravath 326.2
Roush 321.7
Medwick 311.8
Averill 308.4
Sisler 300.0
Burns 296.1
Johnson 287.4
Veach 273.0

By career WARP1
Roush 1034
Averill 977
Johnson 933
Veach 909
Burns 898
Medwick 878
Sisler 869
Cravath 848

When career rate is factored in, along with WWII deductions and FL deductions, I would put Averill ahead of Roush, and overall rank them by career as follows:

Averill, Roush, Johnson, Cravath, Medwick, Veach, Burns, Sisler

But that’s just my take on one part of the ranking. Adding up their placements (a method with many problems, of course), here’s how they rank overall:

Averill 16
Burns 20
Roush 20
Veach 27
Medwick 28
Cravath 29
Sisler 33
Johnson 41

This ranking includes minor-league credit for Averill and Cravath and it expands WWI shortened seasons to 154 games. It does not discount federal league seasons or WWII seasons.
   190. sunnyday2 Posted: September 15, 2005 at 03:42 PM (#1619679)
Being more of a peak voter, I think Chris' list undervalues Medwick and Sisler. And my metrics tend to be WS and OPS+, so I think he overrates Burns. And I'm not much for MiL XC, so knock Averill and Cravath down a bit, but that's a fairly minor consideration.

So, in the end, just swap Medwick and Sisler for Burns and the rest of the list is pretty good.

Here's a challenge for you Chris. Integrate these guys, then, with the Moneyball guys. This of course is in anticipation of Ralph Kiner, and as has been said, if you're going to consider Ralph Kiner, and I am, then you gotta consider some pretty comp guys like Hack Wilson, Charley Keller, Chuck Klein (and Averill and Bob Johnson, whom you have already done). Oh, and Pete Browning and Berger and McGraw and Chance and Mike Tiernan. The high OPS, short career guys.

That is my next task. My gut is that Kiner belongs very high on the ballot while Keller is not near. Why?
   191. DavidFoss Posted: September 15, 2005 at 04:03 PM (#1619713)
That is my next task. My gut is that Kiner belongs very high on the ballot while Keller is not near. Why?

Its an issue I frequently encounter as a peak voter. How short a career will I support?

Keller's got 5-6 full seasons (rookie year looks like an early June call-up or a platoon year) and 4604 PA. Kiner has nine full seasons and 6256 PA.

Complicating things are the 1000 PA Keller missed to the war (and adding two the the full season total). Also complicating things are Keller's final 800 PA occur over five seasons as an oft-injured role player. A possible complication is anyone who might give Kiner any war credit at all.

Its a difficult decision. I've been more forgiving of McGraw because of a positional shortage. (I wonder what I'll do with Al Rosen, then.) Short-career high peak corner OF's opens up a large box of guys, though.
   192. Chris Cobb Posted: September 15, 2005 at 04:05 PM (#1619717)
I have comparable data prepared for Kiner, Wilson, Klein, and Berger, but I am waiting until I work out war-credit elements for Keller and for Kiner to become eligible to post it.

Next after that is a way to effectively compare the NeL hitters using this way of viewing peak, prime, and career.

Then I'll turn to the pre-1910 hitting stars.

I should note, also, that, despite all this work on sorting these outfielders, the top half of my 1960 ballot is mostly pitchers. Part of the reason it's so hard to distinguish these outfielders from one another is that we've gotten down right to the in-out line in the outfielder pool, but we're not there yet on pitchers.

We should elect Newhouser, Griffith, Rixey, and Ferrell among currently eligible pitchers before we take another outfielder from the backlog. And Ruffing, Grimes, Redding, Mendez, and Waddell (!) have just about as good a case, imo, as any of the current outfield eligibles. (The exclamation point after Waddell is there because I think that this may be the first time in 40+ years that I have made a statement that sounds favorable towards Waddell's election.)
   193. sunnyday2 Posted: September 15, 2005 at 04:10 PM (#1619725)
>Short-career high peak corner OF's opens up a large box of guys, though.

David, yes it does, I have a list of 20 guys who made 2 or 3 of the top 100 all-time lists on OPS, OPS+ and OBA, and who are eligible now or shortly. All are cornermen who heavy-hitting CF (Wilson, Averill, Berger, Doby, Browning), except for McGraw and Rosen. It is tempting to just push the CFs and 3Bs to the top of the ballot, especially the latter, but that is what some more study hopefully will validate (or not).

Many are short career guys. I wonder if they are on the top 100 list because they had short careers (no/short decline?) or whether they had short careers because they were Moneyball players (do these skills deteriorate rapidly?). That is what I hope to determine or at least gain some insight on.

Still it has come as a surprise to me that in this Moneyball age that there is significant opposition to Kiner's candidacy. But as you say, how short of a career can one support? I agree (now, subject to change) that the line is between Kiner and Keller.
   194. TomH Posted: September 15, 2005 at 04:17 PM (#1619746)
re: WARP3 league qual

Maybe someone has already concluded this amidst our discussion, but anyway here is what I see:

The way I first used raw EqA to adjusted EqA is NOT a good measure of even how the BP cards would determine league quality among different leagues in similar years. IOW, the "Jeter +11 / Bonds 0" is a red herring.

I looked at George Brett and Mike Schmidt in years 1982-85, when both parks played as neutral ofensive parks.

Brett gains +6 or +7 EqA most years, while Schmidt gains +1. Brett gains 6 more BRAA runs per year also.

But if you check out their real stats to the "translated" stats, Schmidt gains just as much as Brett (as meausre by OPS differences).

The adjustment in EqA is somehow 'correcting' for the difference in 'average' or 'replacement' skill level brought on by the differing offensive league levels.

I conclude that the WARP3 numbers are not tainted by a phony DH/no DH adjustment. Of course, they could be tainted by other things, just not this one :)

I do not know if this would apply to non-DH leagues like the 1950s NL-better-than-AL, or the 1930s when the AL offense was much higher. I have not checked.
   195. DavidFoss Posted: September 15, 2005 at 04:23 PM (#1619758)
Moneyball players (do these skills deteriorate rapidly?).

Not sure what exactly is meant by a Moneyball skill. I'll assume the low-BA, powers-and-walks guys.

Batting average is one of the first skills to deteriorate and power/walks often peak late. On the contrary, if one is a Moneyball player very young, then their batting average may deteriorate lower than the minimum required to be of any use. Also, Moneyball players tend to be much less athletic. If you are less athletic at 25, you are more prone to simply getting too fat by age 35 (Boog Powell, Kent Hrbek, Frank Howard, Cecil Fielder, etc) in which case its not really the Moneyball skills that deteriorate, but the correlation between those skills and a players fitness. Moneyball teams would try to avoid long term contracts past age 30 anyways.

(I feel strange using the word 'moneyball' so much and as an adjective).
   196. Mike Webber Posted: September 15, 2005 at 04:27 PM (#1619765)
Integrating this group with Medwick/Averill/Johnson/Sisler

By career win shares
Cravath 326.2
Roush 321.7
Medwick 311.8
Averill 308.4
Sisler 300.0
Burns 296.1
Johnson 287.4
Veach 273.0

By career WARP1
Roush 1034
Averill 977
Johnson 933
Veach 909
Burns 898
Medwick 878
Sisler 869
Cravath 848


First, thank you Chris.

2nd, I hope some info like this does influence how people think, rather than "well I had Mr X at 12 last year, and the top 2 are gone, so now he is 10."

3rd, I would really like to push the fact that other than very small tweaks due to a shortened WW1 schedule Roush credit is all from things done in the major leagues. I think when you balance out the War discount, and peer hard at the minor league credit, that Roush is really the top candidate in this group.
   197. sunnyday2 Posted: September 15, 2005 at 04:28 PM (#1619768)
Chris,

I am more sanguine about our current distribution of HoMers, but I do already have Newhouser, Mendez and Waddell (and Tommy Bond) in my PHoM (and Redding on my ballot--and I would ask you also to consider Joss, who has been on my ballot for several years). I do not have Stovey, Kelley, Sheckard and Hill. So I am in perfect balance ;-)

I do not have Griffith or Rixey on my ballot but would not be displeased by their election. Meanwhile, even with Medwick and Sisler in my PHoM I am not as top heavy with cornermen as the consensus.

But seriously, it is hard for me to support a guy I don't support because a) he is in the consensus top 10 and b) we don't have enough of his position.

I'd be curious to know your answer to David's question. How short of a career (no matter how good) can you support? The question doesn't arise from the players you evaluate above. Is there a threshold, a lå the good Doctor, below which players do not even get consideration? How about Keller and his 4600 PAs (or 5600 adjPAs)? Or as a practical matter, among players in your consideration set, how short of a career is pretty much a disqualifier?

And I'd be curious at some point to hear you opinion on my question--why so many of the Moneyball players do have short careers? No hurry, though, it can be after you've done the next hitter study.
   198. Mark Shirk (jsch) Posted: September 15, 2005 at 04:38 PM (#1619788)
Chris,

You put Roush ahead of Burns, Veach, and Cravath because of Roush's higher per game WARP and per game WS despite his lower five year totals. However, we know the reason he has lower five year totals is that he happened to miss a lot of games, either through holdout or injury. I am not willing ot give him credit for this.

Therefore, I believe that his per game rates, while superior, are easily balanced out by his lack of in season 'durability' (games played). He may have been better when he was in there, but since he wasn't playing as often as the others his value goes down overall.

I rank the four:
Cravath, Veach, Roush, Burns
   199. Mike Webber Posted: September 15, 2005 at 04:47 PM (#1619808)
Five-year consecutive peak

Rank by WS total
Burns 145.2
Veach 143.6
Roush 143.5
Cravath 142.9

Rank by WARP1 total
Burns 449
Veach 446
Roush 436
Cravath 384


Posted by jschmeagol on September 15, 2005 at 12:38 PM (#1619788)
Chris,

You put Roush ahead of Burns, Veach, and Cravath because of Roush's higher per game WARP and per game WS despite his lower five year totals. However, we know the reason he has lower five year totals is that he happened to miss a lot of games, either through holdout or injury. I am not willing ot give him credit for this.Posted by jschmeagol on September 15, 2005 at 12:38 PM (#1619788)


Come on jschmeagol, those numbers are too close together to put any of them "ahead". 2.3 WS in a 5 year period? Less than one win.

If anything, the only conclusion I could possibly see based on this is Cravath being last based on his significant gap in WARP.
   200. Dr. Chaleeko Posted: September 15, 2005 at 05:14 PM (#1619870)
Just to clear up something...I do have a sort of minimum threshold for consideration, but it's not based on playing time or anything, just on comparison within groups. I take the top 25 guys historically at each position (that is, in my opinion) and draw a line there. That's my way of trying to ensure positional balance of some sort. Pitchers are different, of course....

As to the Moneyball guys with short careers, I tend to think there's a couple things in rank order:
1) They mostly happen to be short-career guys with short decline phases
2) If you've got old-player skills AND you're not athletic to begin with (in the speed/tools way), injuries to the areas important to power-hitting (wrists, core, back, plant leg, etc) will be especially damaging. Once your power goes, your batting eye becomes much less significant because no one's going to be afraid to challenge you anymore. This probably contributes to number 1, (it certainly does for Keller, Berger, and Kiner, for instance).
3) Bad fielding probably plays some part, and it's also deeply effected by injury to the legs, feet, and back.
3) Then throw in a sprinkling of park effects, industry-wide ignorance of take-n-rake, intolerance of low batting averages, etc....
Page 2 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
Darren
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

2015 Hall of Merit Ballot Discussion
(105 - 11:14pm, Dec 20)
Last: Dr. Chaleeko

Herman Long
(10 - 7:54pm, Dec 20)
Last: The District Attorney

Most Meritorious Player: 1960 Ballot
(10 - 4:34pm, Dec 19)
Last: DL from MN

2015 Hall of Merit Ballot
(93 - 1:16pm, Dec 19)
Last: DL from MN

Most Meritorious Player: 1901 Discussion
(29 - 4:50pm, Dec 18)
Last: John (You Can Call Me Grandma) Murphy

Ben Taylor
(81 - 11:04am, Dec 15)
Last: DL from MN

New Eligibles Year by Year
(962 - 10:09am, Dec 15)
Last: Chris Fluit

Cristóbal Torriente
(82 - 11:34pm, Dec 14)
Last: Brent

Brian Giles
(59 - 4:08pm, Dec 14)
Last: Joey Numbaz (Scruff)

Most Meritorious Player: 2014 Discussion
(38 - 12:16pm, Dec 13)
Last: lieiam

Larry Doby
(95 - 10:00pm, Dec 12)
Last: Howie Menckel

Pedro Martinez
(159 - 6:44am, Dec 11)
Last: bjhanke

Most Meritorious Player: 1960 Discussion
(25 - 2:12pm, Dec 09)
Last: DL from MN

Most Meritorious Player: 2014 Results
(7 - 9:24pm, Dec 05)
Last: Chris Fluit

Most Meritorious Player: 2014 Ballot
(29 - 4:20pm, Dec 05)
Last: DL from MN

Syndicate

Page rendered in 1.0404 seconds
43 querie(s) executed