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, May 28, 2006

Hoyt Wilhelm

Eligible in 1978.

John (You Can Call Me Grandma) Murphy Posted: May 28, 2006 at 06:52 PM | 196 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 2 pages  < 1 2
   101. jimd Posted: June 03, 2006 at 12:10 AM (#2048696)
Sorry. I just realized it's premium content.
   102. Steve Treder Posted: June 03, 2006 at 12:53 AM (#2048757)
I remember that it was a fairly big deal when Wilhelm broke Cy Young's record for Games Pitched in 1968. He would eventually shatter it by about the same degree as Bonds did the single-season HR mark (by 18%). I could be wrong, but I think that breaking that record played a role in Hoyt's BBWAA election.

That sure is how I remember it. It was pretty big stuff, and every mention of it was sure to point out how amazing it was that it had all been achieved by a guy who had his rookie year at age 29.

Wilhelm spent most of his career well under the media radar, but by the late 60s, the accumulation of his remarkable accomplishments was making him quite well-known. Wilhelm was never a major star equivalent to his performance, but over the last 5 years or so of his career, he was widely recognized as something of a star.
   103. Chris Cobb Posted: June 03, 2006 at 01:51 AM (#2048870)
Sorry. I just realized it's premium content.

Thanks. I thought it might be behind the "statistics" door. Maybe I will have to sign up for premium . . .
   104. Paul Wendt Posted: June 03, 2006 at 02:47 AM (#2048972)
Eck?
I remember when Tekulve broke Wilhelm's record, so I looked it up.
That was the relief games record.
Orosco and Franco passed Tekulve, no one else.
I am stunned that Dan Plesac tied him.
   105. Daryn Posted: June 03, 2006 at 02:54 AM (#2048985)
The reference was to the games record. All time leaders:

Rank Player (age) Games Throws
1. Jesse Orosco* 1252 L
2. John Franco* (44) 1119 L
3. Dennis Eckersley+ 1071 R
4. Hoyt Wilhelm+ 1070 R
5. Dan Plesac* 1064 L
6. Kent Tekulve 1050 R
   106. Chris Cobb Posted: June 03, 2006 at 04:05 AM (#2049067)
Leverage is a measure of the average importance of the pitching situations. It tells nothing about how effective the pitching was in those situations.

So is it the case, then, that the BP leverage stat is calculated so that the leverage of all innings pitched taken together is 1.0?

I know that's how Tangotiger's leverage index is calculated, but I just want to be sure that the same is true for BP.

I'm not sure that's what its number means, particularly since there are lots of starters, apparently, with high-leverage numbers, where Tangotiger found that starters' leverages were around 1.0. Maybe that was different in the 1960s, when complete games, and starters pitching very deep into games, was a lot more common?

If that's the case, will relievers gain value relative to starters as we move toward the present as starters pitch fewer and fewer high-leverage late innings?

Incidentally, it looks to me like Wilhelm's leverage index for 1960-70, taken in total, was 1.52 .
His xip/ip ratio in WARP for those same seasons was 1.24 .
   107. Dr. Chaleeko Posted: June 03, 2006 at 05:08 AM (#2049096)
Yeah, is leverage a little bit like RBI?

I think so.

In the sense that it's an opportunity stat yes. Also in the sense that good hitters generally gather RBI because good hitters are generally put into the lineup slots where RBIs are had, so too, good relievers are likely to be placed in bullpen slots where they can get higher leveraged appearances.

There's some evaluation of performance buried in there, but it's really a description of role. Though it does have the benefit of being used to understand/determine value more directly, where RBI doesn't.

So is it the case, then, that the BP leverage stat is calculated so that the leverage of all innings pitched taken together is 1.0?

Chris, I'd bet dollars to dimes that I recently read in at least one place that Tango had supplied BP with their leverage numbers. I may have even read it in their annual or one of their online pieces. That doesn't preclude them from then adjusting them in some way, I guess.
   108. Joey Numbaz (Scruff) Posted: June 03, 2006 at 05:51 AM (#2049106)
Wow, I think I might have to subscribe - are you guys saying they have LI for every reliever season since 1960? That would be worth the $40 I guess.

Is this stuff anywhere else?
   109. Joey Numbaz (Scruff) Posted: June 03, 2006 at 06:01 AM (#2049107)
"Yeah, is leverage a little bit like RBI?"

I would say leverage is more like (AB w/RISP) / AB.

It's measuring the opportunity sure. But the value is in how well did he do in those opportunities. I mean his whole job is about being leveraged - that's where it's different than RBI.

AB w/RISP are a luck thing more than anything else. I mean sure a guy can hit 3rd,4th,5th and that helps, but he's not truly being leveraged - a manager can't say David Ortiz can save his 4 AB for the 9th inning today, and we'll just let others run for him and let him keep hitting if it's a close game. We can't say Don Mattingly is going to get his 4 AB only with RISP today.

Ace relievers are specifically used by their manager in spots that are more important. Even with the 3-run 9th inning save (which isn't all that common to be honest - that 'issue' with saves is way overblown - kind of like where you can get an RBI on a groundout or a quality start with a 4.50 ERA).

So LI lets us know exactly how much more valuable those situations are - which I think is immensely important when trying to place a value number on these guys.

Assuming I end up getting the data, I plan to just take IP*LI for relievers and use that in my system as a proxy for IP. I'll do it for starters too if I see a need for it (i.e. if all of the numbers aren't between .98 and 1.02 or something).
   110. Arrieta, Gentile Arrieta Posted: June 04, 2006 at 02:52 AM (#2050398)
Sorry. I just realized it's premium content.

Thanks. I thought it might be behind the "statistics" door. Maybe I will have to sign up for premium . . .


Well, I'm not a subscriber and I can <u>view</u> the statistics. It just tells me I don't have access to the sortable reports with extra features. So I can look at 30 players per page sorted by the default stat. Do have access to all years back to 1960. Everything's copy-and-pasteable, but it'd be tedious.

There's no leverage rating for starters: paging through one of the years, all pitchers are listed. Pitchers who only started get leverage of 0.00. Pitchers who both started and relieved get leverage, I guess, based only on relief innings.
   111. Dr. Chaleeko Posted: June 04, 2006 at 03:12 AM (#2050418)
Everything's copy-and-pasteable, but it'd be tedious.

You're correct; it is tedious.... Now, where did I put that cross I was just bearing?
   112. Dr. Chaleeko Posted: June 04, 2006 at 04:29 AM (#2050442)
Here's a nugget on Hoyt's leverage.

I haven't integrated my 1960-1971 and 1972-2005 leverage numbers yet, but here's how Hoyt fits within the first group.
NAME          LEV 
-----------------
Hoerner      1.54
S Miller     1.51
Wilhelm      1.51 
Farrell      1.47
Worthington  1.46
Radatz       1.42
Perranoski   1.41
Face         1.40
Knowles      1.40
Wyatt        1.39
Linzy        1.39
Upshaw       1.38
Aker         1.36
McDaniel     1.34
Giusti       1.30
Lyle         1.30
Bolin        1.29
R Kline      1.29
Fingers      1.28
Granger      1.28
Bob Lee      1.28
Gladding     1.27
Marshall     1.26
Henry        1.25
Sherry       1.25
Abernathy    1.24
Hall         1.24
Locker       1.23
Wood         1.23
Raymond      1.22
McMahon      1.20
Taylor       1.19
McGraw       1.17
Hiller       1.14
Carroll      1.13
Regan        1.12
E Fisher     1.11
J Brewer     1.11
McBean       1.10
Baldschun    1.08
Drabowsky    1.03
B Miller     0.97 


With more data on the other relief aces in Wilhelm's era, it does indeed appear that Wilhelm's leverage is indeed at the elite level for his era. Given the prime importance of leverage to the case of any reliever, looking at the leverage leaders is probably a good idea. Stu Miller probably needs a second look and if Joe Hoerner's already come and gone, maybe just a quick look-see at his career is in order.

I hope to have more information about the inherited/bequeathed runners, the frequency of appearances, and the effictiveness of all these guys soon.
   113. Chris Cobb Posted: June 04, 2006 at 04:21 PM (#2050534)
Thanks, Dr. Chaleeko, for this great work!

I took a quick look at Joe Hoerner, and he's not someone we need to spend much time on. He was an elite reliever in the late 1960s and had some very nice seasons there, though his workload was always pretty light. He tailed off in the 1970s, and although he was in the majors until 1977, he never threw more than 40 innings in a season after 1971, totaling only 562.7 IP at a 3.80 DERA for his career. I think his leverage scores for the 1970s would bring him down out of the elite group by that measure for his career.

Stu Miller needs a more detailed second look. He's definitely in the elite-reliever group, though I don't think even a leverage bonus would be enough to make him a serious candidate.
   114. Dr. Chaleeko Posted: June 04, 2006 at 04:51 PM (#2050566)
A note about those leverage numbers. I just realized they may include starter innings. So I'm going to try to get some starter/reliever splits. I don't know if the LEV on BP includes starter LEVs, however.
   115. Paul Wendt Posted: June 04, 2006 at 05:01 PM (#2050574)
"Yeah, is leverage a little bit like RBI?"

I would say leverage is more like (AB w/RISP) / AB.


That's right.
The better analogy to batter RBI is pitcher decisions (mainly saves and BS here).

Pinch-hitters have batter leverage greater than 1 (presuming that all leverage measures are normalized to one), except those who bat only for pitchers.

Late-inning defensive replacements have fielding leverage greater than 1.
   116. Joey Numbaz (Scruff) Posted: June 04, 2006 at 05:22 PM (#2050601)
Can someone give a link to where you are finding the leverage numbers on say . . . Wilhelm. From there I should be able to figure out how to find it for the other guys . . . thanks!
   117. fables of the deconstruction Posted: June 04, 2006 at 11:26 PM (#2051436)
IIRC, Tek's 179 IBB issued is among the highest totals ever. I could be wrong about that, though, and b-r.com doesn't list out leaders in IBB allowed.

Tekulve does, in fact, hold the record for most career IBB, at least among pitchers for whom we have the records.


Two Words about Kent Tekulve's IBB... CHUCK TANNER!
IIRC, Bill James in the "Managers" cited Tanner as one of the most IBB happy managers' ever. 113 of Tekulve's 179 IBB were during Tanner's (1977-1985) tenure in Pittsburgh.

--------
trevise
   118. OCF Posted: June 05, 2006 at 01:36 AM (#2051480)
A Joe Hoerner sighting! You're talking about my team there - the '67-'68 Cardinals. Note that Hoerner was left-handed. In those years, the Cardinals also had the right-handed Ron Willis as a trusted relief pitcher. I would roughly characterize the Cardinals pitching staff at the time as consisting of front-line starters (some number less than a full rotation), trusted pure relievers (in '67, just Hoerner and Willis), mopup men, and swingmen. Part of the story of the '67 pennant was the considerable success of swingmen who were needed to step forward and make many starts - notably Nelson Briles and Dick Hughes.

In 1967, the Cardinals recorded 44 complete games - a middle-of-the-pack number that needs no special explanation. Hoerner had 57 games, 66 innings, 32 games finished, 15 saves, and a 4-4 record. Willis had 65 games, 81 innings, 24 games finished, 10 saves, and a 6-5 record. I haven't checked that out on retrosheet, but it looks like the primary way they were deployed was to come in in the 8th or 9th, either for the starter, or for the long releiver if the starter departed early. Which one of Willis or Hoerner to use may have depended in part on the predominant handedness of the expected batters. It seems that Hoerner slightly outranked Willis, and once in a while they'd both come into the same game, but Hoerner last.

In 1968, Hoerner pitched only 49 innings in 47 games. Part of what happened there is that Gibson was both healthy (he'd been out with a broken leg in the middle of the 1967 season) and spectacularly good. With Gibson completing 28 of his starts, the team CG shot up to 63, although that still didn't lead the league (Steve T. should know who did). Hoerner got the lion's share of the save opportunities, which weren't many. Although I remember a specific game in which Willis came in to start the 5th to protect a 10-8 lead and pitched one-hit shutout ball the rest of the way, it looks like Willis lost part of his job to Wayne Granger.

What was Hoerner's value in 1968, even with leverage, and with good rate stats (which he had), how much value can you accomulate in 49 innings? Of course, would you have dared to try to take Gibson out just so you could bring in Hoerner?

I guarantee you that if Hoerner had come along 20 years later, he'd be a LOOGY - and as such, never given the opportunity to be a closer.
   119. Steve Treder Posted: June 05, 2006 at 02:23 AM (#2051501)
With Gibson completing 28 of his starts, the team CG shot up to 63, although that still didn't lead the league (Steve T. should know who did).

Yeah duh. I believe the names Juan Marichal and Gaylord Perry might be implicated.
   120. Steve Treder Posted: June 05, 2006 at 02:29 AM (#2051503)
WRT Hoerner: yes, a sidearming lefty like him would almost certainly have been a LOOGY in an later era; indeed Hoerner himself was deployed as one of the early LOOGYs in the 1970s. Yet in fact he was only marginally more effective against LHBs (255/299/341) than RHBs (246/314/386).
   121. Chris Cobb Posted: June 05, 2006 at 03:19 AM (#2051525)
Can someone give a link to where you are finding the leverage numbers on say . . . Wilhelm. From there I should be able to figure out how to find it for the other guys . . . thanks!

Joe, there's no place where the leverage numbers for only one player can be found at BP.

The leverage statistic can be found at:

http://www.baseballprospectus.com/statistics/sortable/index.php?cid=18

This page lists all the relief pitchers for a given year (you can go to any year back to 1960). in order of WXRL. Leverage is the rightmost column.

You might do better to ask Dr. Chaleeko to send you his compilation of that data, which he used to produce the table above, if you haven't already done so.

I think I may do that myself in just a moment :-).
   122. jimd Posted: June 05, 2006 at 06:54 PM (#2052011)
Incidentally, it looks to me like Wilhelm's leverage index for 1960-70, taken in total, was 1.52 . His xip/ip ratio in WARP for those same seasons was 1.24 .

A 20% boost in his WARP values for his non-starting years will probably move him from well off-ballot to an elect-me spot competitive with Clemente. Three of his 60's seasons would become star-worthy, as would his first three seasons with the Giants in the early 1950's. 5 star seasons in that long and valuable career (total also gets boosted) would get him on my ballot (competitive with Maranville); 8 star seasons is elect-me on this thin ballot.

I'll have to investigate the season by season leverage factors, and then just guess the Giants.
   123. TomH Posted: June 05, 2006 at 07:28 PM (#2052066)
Question: I cannot figure out if BP's DERA stat is based off of the pitcher's ERA, or his total (including unearned) runs allowed. Which makes quite a difference for guys like Wilhelm. Their glossary just doesn't make it clear to me.

I am surprised that Wilhelm's DERA is only 12% lower than Pierce's for example since his ERA+ career value is more like 19% apart, which makes me think the unearend runs have been factored in. Anybody got a good handle on this?
   124. Chris Cobb Posted: June 05, 2006 at 08:31 PM (#2052183)
Tom,

DERA is derived by starting from the pitcher's total runs allowed, normalized to a 4.5 rpg environment (listed as RA) and adjusting it to reflect average fielding support, based on an assessment of the quality of the team's defense. This gives you PRAA (pitching runs above average), from which DERA is calculated. If a pitcher had above fielding support his NRA (normalized runs allowed average) will be lower than his DERA (defense-adjusted runs allowed average). If a pitcher had below average fielding support, his NRA will be higher than his DERA.

The "earned run" / "unearned run" distinction is not employed in it, although the fielders' propensity to errors is included in the team fielding assessment.

Does that clarify?

The unearned runs might be one reason that Wilhelm's DERA is closer to Pierce's than his ERA+ is. Since Wilhelm's NRA and DERA are equal, and Pierce's NRA is lower than his DERA, the difference between DERA and ERA+ is probably the unearned runs. In general, however, a pitcher whose DERA is worse than his ERA+ may simply have been the beneficiary of excellent fielding support, which boosted his ERA+, since a pitcher's ERA is generally helped by good defensive, because good team defensive efficiency lowers batting average on balls in play.
   125. TomH Posted: June 06, 2006 at 03:57 PM (#2053366)
Thanks Chris!


---

Assignment: in 150 words or less, explain why Hoyt rates higher than any other MLB pitcher on the ballot.

My (and our consensus) highest other MLB SP is Pierce.

Using BP’s DERA, which adjusts for defense and unearned runs and league strength, and their ‘translated IP’ that adjusts for playing condition differences, we have career stats:

Pitcher …transIP DERA
Wilhelm . 2306 …. 3.51
Pierce .… 3428 …. 3.97
NewHoyt. 3124 …. 3.75

I will add .20 to Wilhelm’s DERA to account for the easier RP advantage of his day (accounting that Hoyt did start some, and Pierce relieved a little).
I add 806 (35% of his total) innings to account for Wilhelm’s extra leverage. But for these 806 IP, I use a higher DERA of 3.91 since the non-Wilhelm relief ace he is replacing (and getting extra leverage credit for) would likely be slightly better than average. So Wilhelm’s ‘translated stats’ for me would be [2306 IP of 3.71 + 806 IP of 3.91] the ‘NewHoyt’ line on the bottom.
Sure there is other stuff (peak, pitcher hitting), but Wilhelm is Very comfortably ahead.
   126. Dr. Chaleeko Posted: June 06, 2006 at 04:11 PM (#2053386)
TomH,

Does Hoyt's transIP indicate leverage in any way? I just don't know the answer, which is why I'm asking.
   127. TomH Posted: June 06, 2006 at 04:52 PM (#2053428)
No, they do not.
   128. TomH Posted: June 06, 2006 at 04:57 PM (#2053433)
To make sure I was clear: the 'Wilhelm' line, straight from BP, has no leverage. The 'NewHoyt' line is my personal take on his leveraged equivalent innings.
   129. Dr. Chaleeko Posted: June 06, 2006 at 05:08 PM (#2053462)
Since TomH brought up the question of Billy Pierce's leveraged relief work, let's see what retrosheet says!

We have data from 1957-1964, so mostly the descent from his peak to his last years in the pen in the NL. Here's the total starting/relieving splits for him:
G  GS GF  SV   IP      H   BFP  HR    R   ER   BB  IB   SO  WP HBP  2B  3B GDP  W   L    ERA
-------------------------------------------------------------------------------------------------
SP 195 195  0   0 1309.2 1233  5403 142  541  501  343  20  734  14   8 182  33 112 91  65   3.44
RP  84   0 40  19  103    100   426  13   34   32   22   3   65   2   2  14   1   7  6   6   2.80

     K
/9  BB/9  K/BB HR/9  OAV  OOB  OSG BABIP
----------------------------------------------
SP  5.04  2.36  2.14 0.98 .248 .293 .383  .266
RP  5.68  1.92  2.95 1.14 .257 .291 .398  .280 


There is an interesting effect going on here, however, one that's been detailed elsewhere. Most of Pierce's final two seasons were spent in relief. During which time he gave up 8 of the 13 homers he gave up in relief (that is within the retrosheet sample).

Here are his year-by-year relief splits for INN and ERA

as starter     as reliever
YEAR  INN     ERA    INN      ERA
---------------------------------
1957  253.67  3.16     3.33 10.80
1958  242.67  2.67     2.67  3.38
1959  223     3.63     1     0.00
1960  194.67  3.66     2     0.00
1961  158     3.99    22     2.45 
1962  157.33  3.55     5     1.80 
1963   73.33  4.66    25.67  3.16
1964    7.67  1.17    41.33  2.40 


This only covers the second half of his career. In the first half of his career, here's here G, GS, GF, SV

YEAR    G  GS RelG GF SV
-------------------------
1948   22   5   17  9  0
1949   32  26    6  4  0
1950   33  29    4  3  1 
1951   37  28    9  6  2
1952   33  32    1  1  1
1953   40  33    7  6  3
1954   36  26   10  4  3
1955   33  26    7  5  1
1956   35  33    2  2  1
=========================
TOTAL 301 238   63 40 12 


It looks to me like Pierce was used in a slightly ramped down version of the Grove/Dean model ace/closer hybrid for the first part of his career, during Paul Richards' and Marty Marion's tenures as manager. In those years, through 1955, the White Sox bullpen lacked much definition of roles. Seven or eight guys would each get some saves. By the by, if you pull his 1948 rookie season out of the equation, he's got 46 relief appearances, 31 GF, and 12 saves. (I didn't include his 1945 greenhorn year.)

Then in 1956 in Marion's final season, Pierce's number of per annum relief games dropped off to one, meanwhile in the bullpen, the GFs and saves were all concentrated among four guys (including Ellis Kinder!). In 1957 Al Lopez took the helm, and kept Pierce in the same onesy-twosey pattern. The saves and GFs weren't well concentrated at first, but the roles of bullpen versus top starters were defined more clearly than previously as Lopez sought a bullpen structure. Finally in 1958, Lopez hit on the bullpen duo of Lown and Staley who became the primary go-to guys and ushered in a more modern bullpen structure. They lasted until 1961, Pierce's last year, and their declines probably meant that Pierce got a couple extra relief apps in 1961, breaking the pattern of ones and twos.

Finally in the Giants years, his turn to the pen is probably most explicable by a decline in effectiveness as a starter.

The point here, if there is one, is that Pierce does indeed appear to have the probability of significantly leveraged innings early in his career. And that his later relief work (posted above) shouldn't be generalized over his entire career. Also, that in comparing Wilhelm and Pierce, it is probably appropriate to make some allowance, as TomH has, for Pierce's leveraged appearances.
   130. KJOK Posted: June 06, 2006 at 05:31 PM (#2053546)
IIRC, Bill James in the "Managers" cited Tanner as one of the most IBB happy managers' ever.

And IIRC, this wasn't just a case of walking Barry Bonds type hitters alot, but of almost always walking the 8th place hitter whenever he had the chance, which likely HURT his own pitcher's stats a bit by denying them opportunities to face the weakest hitting of position players.
   131. Joey Numbaz (Scruff) Posted: June 07, 2006 at 08:06 AM (#2055007)
I'm trying to incorporate the numbers for inherited runners into my system, but I'm having trouble getting things to match up.

Prospectus has 4 categories:

InRP (Inherited Runs Prevented)
BRP (Bequeather Runs Prevented)
RA (Runs Allowed per 9 IP)
Fair RA (Runs Allowed per 9 IP, accounting for InRP and BRP)

I can't get the numbers to match.

Take Wilhelm to 1960

InRP 4.0
BRP .7
RA 4.22
Fair RA 3.90

The 4.22 matches up, that's just 69 R in 147 IP.

But using 4.0 and .7 (positive numbers are 'bad' for BRP - in terms of the pitcher, positive means his relief pitchers saved him runs); I get that Fair RA should be 4.02, not 3.90.

So I tried Elroy Face 1960.

InRP 5.1
BRP 2.6
RA 3.10
Fair RA 2.85

But this one's even worse. RA should be 3.18 (he allowed 39 runs that year, not 38. But even if I use 38 runs as the base, I get Fair RA at 2.90. If I use the correct starting point of 3.18 RA/9, I get Fair RA at 2.98.

What other adjustments are they making?
   132. Joey Numbaz (Scruff) Posted: June 07, 2006 at 08:23 AM (#2055016)
OK - in 1960 Wilhelm started 11 games, I missed that. So I could see why we might have some issues there.

But Face relieved every game in 1960, so everything should match, right?
   133. Joey Numbaz (Scruff) Posted: June 07, 2006 at 10:11 AM (#2055031)
OK, I've adjust Wilhelm for leverage. I've adjusted him for his inherited runs prevented (10.0 1960-72), and how well his relievers did with the runners he left behind (they saved him 18.7 from 1960-72).

I normalized his starter IP as discussed on earlier thread. I left his reliever innings unnormalized and adjusted for leverage based on the numbers at Prospectus.

My normalization involves adjusting every season to a 4.50 R/G environment. I adjust like NRA and DERA, by taking the pitchers win expectancy in his actual run environment and crediting him for the runs allowed that would give the same win expectancy at 4.50 R/G.

From 1952-59, I eyeballed a leverage index for each season, based on his saves to IP ratio. It works out to an average of 1.27, wheras from 1960-72, his LI was 1.48. I was conservative. By year:

1952-53 1.4
1954 1.2
1955 1.1
1956 1.2
1957 1.4
1958 1.1
1959 1.0 (treated him as a starter, since he only pitched 15.3 innings in relief).

I set replacement a 5.48 R/G (the equivalent of 6 WS per 220 IP being replacement level, around .400, .403 IIRC).

For relief innings I'm setting replacement to 5.23, allowing for a 'reliever advantage' of .25 R/9).

I also account for pitcher hitting, using RCAP from the Sabermetric Encyclopedia.

Note, my system has significantly higher replacement level than WARP (6.00 R/G). So I'll call mine aWAR (for adjusted Wins Above Replacement).

Wilhelm ends up with 64.2 aWAR.

Here's where he fits on an incomplete list of notables:

Drysdale 69.4
Bunning 68.7
Pierce 65.6
Bridges 65.0 (with war credit)
Wilhelm 64.3
Grimes 60.3
Trucks 57.5 (with war credit)
Newcombe 56.8 (with Neg.L and war credit)
Leonard 56.6 (with 1942-45 demerit)
Willis 56.0
Trout 55.7 (with 1942-45 demerit)
Cooper 55.6
Walters 55.3 (only pitchers years, 1942-45 demerit)

Notice how it starts to crowd up from Trucks down. There are another 9 pitchers between Cooper and 50.0, only among the sample of 27 pitchers I calculated. Guess where I draw the line for being a serious candidate or 'endorsed selection'? Right between Wilhelm and Trucks. Grimes basically is the line.

Note that doesn't credit the starters with leverage for their relief appearances.

Now some will say that this doesn't account for peak which is where relievers lack. That's where Pennants Added comes in . . . here's how Wilhelm stacks up with the same group there:

Drysdale .848
Bunning .843
Pierce .788
Bridges .776
Wilhelm .757
Grimes .724
Trucks .680
Trout .680 (notice how he jumps Willis, Leonard and Newcombe)
Newcombe .678
Walters .677 (notice how he jumps too)
Willis .676
Cooper .668
Leonard .666

Guess where I draw my line here for serious candidates? Still right between Wilhelm and Trucks, with Grimes basically being the line.

I know the lists look similar and like PA doesn't take peak into account much.

But every pitcher input so far (27) except for Bunning and Walter Johnson (the only slam dunk guy) was between 3.9 and 5.6 in t0heir 5th best year. And the guys lower in their 5th best year are the guys with the higher overall peak, so a lot of it just evens out. Trout and Walters jump up the list, as they should, they had the two best seasons of anyone on the list except for Johnson. But while Lefty Gomez had one of the best #1 and #2 seasons, his #4 and #5 are among the worst. So it has a way of evening out. Walter Johnson's aWAR:PA ratio is 76.7:1, Sam Leever's is 85.4:1, which is a pretty significant difference.

In closing on Wilhelm . . . I was pretty conservative with my 1952-59 LI. I think I could be convinced that I shouldn't be lowering the replacement level for relievers at all but went with .25 since that seems to be the consensus. I think this puts Wilhelm in as conservative a light as possible compared to the top starters on the ballot.

I'd appreciate any comments on the methodology and results.

For years pre-1960 I think I should probably use SV:RelIP as a proxy for LI - would anyone have a good guess as to what a fair number for that would be? Once I have that, I can adjust the starters for their relief innings, which should bump guys like Dean and Pierce up a bit.

I can also adjust starters post 1960 for their bullpen support, based on the numbers at Prospectus, which are great. I haven't done that yet, but the way my spreadsheet is set up now, that will be easy to add.
   134. Joey Numbaz (Scruff) Posted: June 07, 2006 at 10:18 AM (#2055033)
Just wanted to note that PA doesn't only take top 5 seasons into account for peak or anything like that. It looks at each season, and the higher the aWAR, the more PA you get, but PA goes up exponentially, it's not a linear relationship.

As a quick example . . .

Wilhelm gets 3.9 aWAR for 1952, which is .046 PA. He gets 3.8 aWAR for 1954, which is .044 PA. That's a total of .090 PA for the two seasons, which total 7.7 aWAR.

In 1959, he had 7.7 aWAR all alone. That gets him .098 PA, so the one big year has more 'pennant value' than the two good years.

That .008 bonus might not look like much, but it's a 9% bonus, and note that only .014 PA separate Trucks (7th) and Leonard (13th) on the list above.
   135. Joey Numbaz (Scruff) Posted: June 07, 2006 at 10:25 AM (#2055034)
"For years pre-1960 I think I should probably use SV:RelIP as a proxy for LI"

Actually, some combination of SV + RW + RL to relIP would be better to use as the proxy. I know my super old MacMillan from 1987 has relief W, relief L and SV in it - I'm pretty sure it has rel IP too. I've got to find that thing now, I probably haven't opened it in 10-15 years. That will be a huge help as retrosheet gets incomplete going back further.
   136. Joey Numbaz (Scruff) Posted: June 07, 2006 at 10:35 AM (#2055036)
Crap - didn't convert relief innings from 154 to 162 game season for Wilhelm from 1952-60. Need to at least give credit there, since most relievers we are going to compare him to will have pitched in a 162 game season. Starters are already normalized, so that's not an issue.

Doing that bumps Wilhelm to 65.2 aWAR and .767 PA.

By the way, his 'adjusted innings' when you account for leverage, schedule length and normalize his starting IP to era norms . . . works out to 3107.4, as opposed to the 2254.3 he actually pitched. His adjusted DERA (mine uses PythaganPat exponents, as opposed to the flat 2 Prospectus uses) is 3.42, in a 4.50 environment.
   137. Joey Numbaz (Scruff) Posted: June 07, 2006 at 10:43 AM (#2055037)
Hmmn . . . another thing I just thought of.

Since starters are adjusted based on the leaders in IP for their season - they really aren't being adjusted to a 162 game season. The number I used as the 'par' is the 25-75% of #1 starters (1 per team, not literal #1 on each team) for the league season. So really they are being normalized to the average season length throughout history, not 162 games.

From 1901-2005, that works out to 155.6 games per season. So I guess that's what I should normalize reliever innings to, to put them on par with the starters. Just relIP/tG*155.6.

So instead of a 5% bonus from 1952-60, he gets a 1% bonus from 1952-60, and a 4% dock from 1961-72.

Just to be clear, this isn't a penalty. It just puts the relievers on par with the starters in terms of IP.

If I'm missing something here, or this doesn't seem right, please let me know.

So that puts Wilhelm at 63.0 aWAR and .740 PA.
   138. Joey Numbaz (Scruff) Posted: June 07, 2006 at 10:50 AM (#2055039)
And lowers his adjusted innings to 3002.2. I think I'm done. Really.
   139. Joey Numbaz (Scruff) Posted: June 07, 2006 at 12:57 PM (#2055066)
No dice on the old MacMillan having relief IP, just RW, RL, SV.

Anyone know if it's possible to get splits IP as a starter and reliever 1871-1959? Or any portion of that? It would be incredibly helpful.
   140. Dr. Chaleeko Posted: June 07, 2006 at 01:16 PM (#2055075)
Joe,

About why Face wouldn't add up. Is it that BP are park or lineup adjusting the fair ERA?
   141. Jeff M Posted: June 07, 2006 at 02:24 PM (#2055113)
Don't the WX, WXL, WXR and WXRL stats all incorporate leverage automatically? So why the focus on leverage for Wilhelm, instead of win expectancy?

Win expectancy takes the game situation, score, baserunners, park factor, etc. from when Wilhelm enters the game, and calculates how much he improved (or screwed up) the team's win expectancy during his appearance. If he entered the game in a low leverage situation (e.g., a blowout), he won't get much WX no matter how well he pitches, because his pitching doesn't alter the game much. On the other hand, if he enters in a high leverage situation (leading by one run with runners on), he'll get tons of WX if he pitches well.

I'musing WXR (win expectancy above replacement) for Wilhelm, which basically gives you his true life pitching WARP. I'm avoiding WXRL because there is a "lineup adjustment" in there somewhere, and I don't understand how that is calculated.

I looked at Wilhelm, Face, Perranoski and Stu Miller in terms of their WXR from 1960-1969. Wilhelm was by far the oldest of this group, but Miller and Face were on the downward slope of the age curve too. Perranoski was but a lad. Miller didn't pitch in 1969 and Perranoski didn't pitch in 1960.

I include WS for the decade of the 60s just for a comparison.

<u>Pitcher    60    61   62   63   64   65   66   67   68   69   Tot  WXR/IP   WS</u>
Wilhelm    2.0  4.0  3.8  2.5  4.7  4.7  1.3  4.2  1.9  1.6  30.7   0.03   144 
Miller     0.3  5.1  1.8  5.2  3.5  8.4  4.6  0.6 -0.3  n/a  29.3   0.04   101
Face       3.8  0.9  4.3  1.4 -0.4  0.0  1.3  1.2 -0.1  0.8  13.1   0.02    84
Perranoski n/a  2.2  4.6  5.5  0.8  4.1  0.1  3.0 -0.9  3.8  23.2   0.02   108

Miller was more valuable on a per inning basis, but not quite as valuable as Wilhelm over the decade (though he pitched one fewer year).
   142. Dr. Chaleeko Posted: June 07, 2006 at 03:45 PM (#2055177)
worth noting that miller was three of the four best WXR seasons in the group, and the best season by far. That's kind of interesting.
   143. Steve Treder Posted: June 07, 2006 at 04:25 PM (#2055201)
worth noting that miller was three of the four best WXR seasons in the group, and the best season by far. That's kind of interesting.

It's fascinating.

There is a large opportunity function going on here, of course. Miller was the undisputed ace of the Orioles' pen in 1963-66, and had been as well with the Giants. Wilhelm had competition from Eddie Fisher, Bob Locker, and Wilbur Wood for that role with the White Sox, as they featured amazingly good bullpens in that period (and interestingly, with three of their aces as knuckleballers).

Face was really past his prime for most of this period; his heyday as the Pirates' ace reliever ended in 1962; after that Al McBean was generally the go-to guy. And the Dodgers rotated Perranoski with several other excellent relievers in these years, including Larry Sherry, Bob Miller, and Phil Regan.
   144. Joey Numbaz (Scruff) Posted: June 07, 2006 at 06:21 PM (#2055283)
"Win expectancy takes the game situation, score, baserunners, park factor, etc. from when Wilhelm enters the game, and calculates how much he improved (or screwed up) the team's win expectancy during his appearance. If he entered the game in a low leverage situation (e.g., a blowout), he won't get much WX no matter how well he pitches, because his pitching doesn't alter the game much. On the other hand, if he enters in a high leverage situation (leading by one run with runners on), he'll get tons of WX if he pitches well."

Jeff, I'm just not a big a fan of using Win Expectancy as anything other than a fun junk stat. I'm open to being convinced otherwise.
   145. Dr. Chaleeko Posted: June 07, 2006 at 06:27 PM (#2055287)
Joe,

I remember you saying you weren't a fan of WX stats. Can you describe why? I don't have much footing on them other knowing what they mean, so I'm looking for reasons why or why not to trust them.
   146. jimd Posted: June 07, 2006 at 06:27 PM (#2055288)
Win expectancy takes the game situation, score, baserunners, park factor, etc.

How does Win Expectancy deal with the situation when the ball is going out for a walkoff HR but the CF reaches over the fence and pulls it back for the game-ending out? Is that just another out at the office for the RP? Did the RP get blame for blowing it and the CF credit for the full W?

Alternatively, an easy inning-ending DP ball goes through the 2b-man allowing the tying and winning runs to score. Did the RP blow it in a high leverage situation? Did he do his job while the L gets hung on the fielder?

These are situations that have high impact on the numbers that come out of a leveraged Win Expectancy system. The inherent difficulty (if not near impossibility) of getting the value right makes me highly suspicious of leveraging results, which is not the same as leveraging opportunities.
   147. Dr. Chaleeko Posted: June 07, 2006 at 06:30 PM (#2055294)
makes me highly suspicious of leveraging results, which is not the same as leveraging opportunities.

So Jim, are you suggesting that leverage is, in general, not useful to our analysis of relief pitching? We don't care about leverage except as it relates to results (and magnifies them for RP), so is it a red herring to even look at LEV for HOM purposes?
   148. Joey Numbaz (Scruff) Posted: June 07, 2006 at 06:41 PM (#2055303)
If I can't find relief innings pitched for years before 1960 . . . how is this for an estimation technique?

sD = starter decisions

(CG*9) + ((CG-sD)*9) = sIP (Starter IP)

rIP = IP - sIP

And then check it to make sure it makes sense for that particular pitcher. Meaning a guy with 5 relief appearances doesn't have 45 rIP, stuff like that.
   149. Joey Numbaz (Scruff) Posted: June 07, 2006 at 06:54 PM (#2055313)
I remember you saying you weren't a fan of WX stats. Can you describe why? I don't have much footing on them other knowing what they mean, so I'm looking for reasons why or why not to trust them.


What jim said. Thanks jim!

Dr. C (#147) . . . there's a big difference between using LI and WE. WE is prone to all of the distortions jim mentions above - which in the small samples of a pitcher season (or career) won't even out.

Leverage however, just says how important the innings are. It doesn't magnify each result, for which the pitcher may or may not be responsible.

For me, taking the importance of the innings (LI) and combining that with the overall results (RA/IP, adjusted for league, park and team defense) is a much sounder approach.

I don't know that I explained that great, sometimes it's hard to get across in words and idea that you know works in your head. If I didn't do a great job, someone who understands what I'm saying please pick me up!
   150. jimd Posted: June 07, 2006 at 07:02 PM (#2055321)
are you suggesting that leverage is, in general, not useful to our analysis of relief pitching?

No. Leveraging results could be very useful. But when the values assigned to single plays have the potential to modify league-leading ratings by 20-30%, it is essential that we know how to get the credit/blame right, because the sample size of those highly leveraged plays is way too small for things to have a prayer of evening out over a season.

OTOH, leveraging opportunities chucks the situation-dependent multipliers on the performance, and uses the overall season stats as an indicator of performance in all situations. It measures how important the situations were, assumes that the player's performance was similar in all situations, and comes up with a value. This is how WS and WARP do it, though the leverage estimators are crude.
   151. jimd Posted: June 07, 2006 at 07:18 PM (#2055338)
Leveraging results and leveraging opportunities will give the same final answer if the performance is consistent, independent of the leverage. When the final answer is different, it is because the player performed worse in the highest leverage situations, which may be an inherent problem in his "makeup" or just a case of bad luck combined with insufficient sample size.
   152. Joey Numbaz (Scruff) Posted: June 07, 2006 at 09:09 PM (#2055473)
OK no need to estimate relief IP, just ordered a 1st Edition MacMillan Encyclopedia, which I'm told has relief IP throughout history. Should be here in 4-7 days.

I've also got an actuary friend that I'm asking for help on estimating LI for relief IP based on rW, rL, SV, rIP, LI of pitchers we do know about.
   153. Steve Treder Posted: June 07, 2006 at 10:19 PM (#2055562)
OK no need to estimate relief IP, just ordered a 1st Edition MacMillan Encyclopedia, which I'm told has relief IP throughout history. Should be here in 4-7 days.

That book is a treasure trove, I tell ya. For every pitcher in every season (through 1968, of course), it provided relief games, innings, W-L, and ERA. Also for every batter it provided pinch-hitting at-bats and hits (maybe HRs too, I forget).

It was by far the best edition. The subsequent ones got wimpier and wimpier.
   154. Jeff M Posted: June 07, 2006 at 10:44 PM (#2055577)
How does Win Expectancy deal with the situation when the ball is going out for a walkoff HR but the CF reaches over the fence and pulls it back for the game-ending out? Is that just another out at the office for the RP?...

I don't know what BP does in its process, but you certainly could give all the credit to the defender in that situation, and none to the pitcher. That would produce a more accurate result. I've done it for individual ballgames, just for fun.

It is EXTREMELY labor-intensive to do that though. Plus, there's a huge subjective factor on balls put in play. Does the diving shortstop get more credit for the out than the pitcher, or was the shortstop positioned incorrectly? Should the pitcher only get 1/2 credit on every ground ball? 1/3? You can't possibly get it "right." Maybe you'd want to adjust the win expectancy based on a DERA vs. NERA factor or something, to take into account the general level of defense the pitcher had behind him.

It is imperfect, but it seems to add something to our understanding of leverage. All the other stats we use are simulators. They take some raw stats and convert them into "what should have happened," all things being equal. Leverage is the same way.

I'm not saying there's anything wrong with leverage, as long as it is understood that it is a crude pattern of pitcher use. To me, it isn't that helpful alone, and maybe not even when it is applied to other stats as an adjustment.

Take Face's appearances in 1965. It isn't a representative season...it's just one that isn't so labor-intensive as to take 10 hours to evaluate, because he only had 16 appearances and 20 or so IP. His leverage is basically average (1.04). His raw numbers indicate an ERA+ of 132. He had no unearned runs. He won 5 games in 16 appearances, and didn't earn any saves, although he finished 12 games.

Considering he didn't pitch much, it doesn't look like a horrible season. He got 3 win shares for not much work. He got a 1.6 WARP. Project his 20 IP to an 80 IP season and he picks up 12 win shares and 6.4 WARP. Not bad.

His win expectancy above replacement is 0, as reported by BP. He was a replacement level pitcher. I actually looked at the play-by-play files (3 are not available). He sucked. Had he not been a veteran, he would have been in the minors.

I don't intend this as a commentary on Face...he was obviously not himself. I just use it to illustrate how his raw stats and the generalized uberstats can mislead. Applying his leverage factor to his stats wouldn't correct the misimpression.

If I scrap win expectancy from my analysis (and I might), then I'll also scrap leverage, because I think the two have to go together. I'm just thinking out loud here (not always a safe course). After all, this isn't anything more than clutch pitching, is it?

Incidentally, as sunnyday mentioned (I think), leverage and win expectancy are going to get more random as time goes on. Managers plug the closer into any save situation, without heed to the importance of the situation.
   155. Steve Treder Posted: June 07, 2006 at 11:06 PM (#2055591)
I actually looked at the play-by-play files (3 are not available). He sucked. Had he not been a veteran, he would have been in the minors.

I'm looking at the same play-by-play files you are, and I'm just not seeing what you're seeing. He had a couple of rough outings, but never got bombed, only once surrendering as many as 2 runs. 11 of his 16 outings were scoreless; 7 were hitless. Yes, he vultured a bunch of cheap wins, but his overall stats were rather good: 20 innings, 20 hits, only 1 HR, only 2 unintentional walks, and 19 strikeouts.

What causes you to say he "sucked"?

(Face was hurt that year, BTW; he had a knee injury that kept him on the DL from early May through late August.)
   156. Jeff M Posted: June 07, 2006 at 11:29 PM (#2055604)
Here's adjusted runs prevented, from BP, which takes a lot of stuff into account, including inherited runners prevented from scoring and how many runners the reliever left for subsequent relievers to clean up.

This paints a different picture than win expectancy and leverage by widening the gap between Wilhelm and Miller during this decade.

<u>Pitcher     60     61    62    63    64    65    66    67    68    69    Tot  ARP/IP</u>
Wilhelm   13.3   26.9  23.1   8.9  30.3  33.4  11.1  11.5  15.0   6.7  180.2   .163    
Miller     1.0   40.4   0.4  23.0  10.0  33.7  21.7   5.6  -2.7   n/a  133.1   .159   
Face      19.7   10.1  25.6   3.4 -11.6   2.0  12.9   7.6   3.7   4.7   78.1   .132   
Perranoski n/a   27.2  10.2  29.1   0.8  22.4   5.9  12.2  -8.2  29.7  129.3   .135   

Replacement level appears to be about -1.4.
   157. Jeff M Posted: June 07, 2006 at 11:34 PM (#2055609)
What causes you to say he "sucked"?

I didn't save my spreadsheet, but when I was running him through the win expectancy calculator, he was just ineffective when it counted (which wasn't very often in those games). He didn't get "bombed," but you can be a pretty ineffective reliever by giving up just a few runs at the wrong times or in important situations. I wish I had saved the spreadsheet so I could be more specific.

Maybe sucked is a relative term. :) He sucked from a Win Expectancy perspective, since he was replacement level. ARP shows he did prevent two runs, so I guess it wasn't a total loss.

Anyway, he sucked a lot more than his stats indicate, IMO. That was my point, more than casting any aspersions on Face's season.
   158. Steve Treder Posted: June 07, 2006 at 11:40 PM (#2055610)
Anyway, he sucked a lot more than his stats indicate, IMO. That was my point, more than casting any aspersions on Face's season.

Gotcha.
   159. jimd Posted: June 07, 2006 at 11:52 PM (#2055613)
The point I'm trying to make has little to do with measuring the leverage, and everything to do with applying the leverage.

Measuring leverage is fine, it's what tangotiger does. It's the weight (importance) that we attach to any given situation. From tango's work, it appears to be a solved problem. Average leverage is a simple extension of that concept. WARP and WS attempt to estimate the average leverage without benefit of PBP. The average leverage indicators that go along with the WXRL are most likely better measurements than those.

Where Joe and I have a problem is with the idea of applying the specific leverage number to the result of the specific situation without having a very precise description of that situation. Assuming consistency of performance and combining that with an improved leverage measurement yields an improved WARP/WS number, and that's good. Atomically analyzing (result of each play)x(leverage of situation) without an accurate model for determining credit/blame makes the error bands unacceptably large, and that's bad. If the RP who was saved by his CF gets a lot of value from that situation while the RP that was victimized by his 2B takes a heavy penalty, each due to the large leverage multipliers on the performances, then the stat has a problem (at least IMO).
   160. Dr. Chaleeko Posted: June 08, 2006 at 12:14 AM (#2055627)
so that takes us back to what Joe said earlier: take WARP1, remove the LEV estimation; insert the Tango/BP-WXRL LEV; recalculate value; presto! New, more LEVyed WARP1.

Turning attention to Jeff M's ARP table above, it's interesting to note that again Miller's got the biggest season on the board and the second biggest. He appears to be a peak/prime candidate to Wilhelm's career candidate. Maybe a Chance or Sisler to Wilhelm's Beckley (though Wilhelm's peak is better than Beckley's relative to their positions).

Miller's 1957-1959, ages 29-31, are also excellent (at least by ERA+: 108, 157, 135 in 473 innings---136 for the three years), but his years before that---when he was starting a lot more often are not good at all. I do wonder if when we get full 1950s splits, his relief and starting splits will be extreme. Wilhelm from 57-59 is even better (95, 176, 153 in 415 innings---157 for the three years). Incidentally, Miller started 42 times in those three years, and Wilhelm 37.

Anyway, this all leads me to say that it seems like Wilhelm has a substantial enough career advantage and Miller insubstantial enough value outside of peak to make it obvious that Wilhelm goes first. But in that era, it's also pretty clear that Miller has a great case as the number two guy in MLB relief wise during that span.
   161. Harold can be a fun sponge Posted: June 08, 2006 at 12:17 AM (#2055630)
If I scrap win expectancy from my analysis (and I might), then I'll also scrap leverage, because I think the two have to go together. I'm just thinking out loud here (not always a safe course). After all, this isn't anything more than clutch pitching, is it?

No. Using Win Expectancy is measuring clutch pitching. Using (average) Leverage is appreciating that a relief ace can be used in a way to maximize his value, without actually looking at how he does in each situation.

A relief ace with a high LI faces a lot of clutch situations. Muliplying his IP by LI (or doing some adjustment like that) gives him credit for the ability to pitch in more clutch situations; it doesn't attempt to measure his performance in the clutch. Using Win Expectancy does that, however.
   162. Harold can be a fun sponge Posted: June 08, 2006 at 12:21 AM (#2055633)
I don't know what BP does in its process, but you certainly could give all the credit to the defender in that situation, and none to the pitcher. That would produce a more accurate result. I've done it for individual ballgames, just for fun.

It is EXTREMELY labor-intensive to do that though. Plus, there's a huge subjective factor on balls put in play. Does the diving shortstop get more credit for the out than the pitcher, or was the shortstop positioned incorrectly? Should the pitcher only get 1/2 credit on every ground ball? 1/3? You can't possibly get it "right."


And even your examples don't get it right. It's not about apportioning the credit between pitcher and defense by fractions. In the first example (potential home run pulled back over the fence), the fielder should get negative credit, and the OF should get even more positive credit.

Really, the way to do it is to break things up into smaller fragments -- the pitcher and batter get credit (positive or negative) for the batted ball, and the fielder gets credit for what he does with that batted ball compared to what average fielders would do.
   163. Steve Treder Posted: June 08, 2006 at 02:10 AM (#2055807)
But in that era, it's also pretty clear that Miller has a great case as the number two guy in MLB relief wise during that span.

And the f@cking Giants f@cking traded him for f@cking Billy f@cking Hoeft.

Sorry. Carry on.
   164. Steve Treder Posted: June 08, 2006 at 02:13 AM (#2055810)
But in that era, it's also pretty clear that Miller has a great case as the number two guy in MLB relief wise during that span.

And the f@cking Giants f@cking traded him for f@cking Billy f@cking Hoeft.

Sorry. Carry on.
   165. Steve Treder Posted: June 08, 2006 at 02:16 AM (#2055812)
I guess the point has been made ...
   166. Dr. Chaleeko Posted: June 08, 2006 at 05:05 PM (#2056248)
Chris C. and Joe D.,

At risk of sounding presumptuous or entitled, is there any chance that either of you are

1) working on rejigging WARP or WS for relievers to make use of Tango's LEV versus the estimated LEVs in those systems already?

2) if so, will you be posting results for Wilhelm by Monday? (if so, I'll hold off on voting til then; it feels that important to my placement of Wilhelm.)

3) do you forsee this substitution of real LEV for estimated LEV causing Wilhelm's WARP or WS to shift dramatically?
3a) if it did shift his WS or WARP dramatically, would that cause big team-level problems?

4) Thanks, as always.
   167. TomH Posted: June 08, 2006 at 07:22 PM (#2056369)
If I can't find relief innings pitched for years before 1960 . . . how is this for an estimation technique?
-------------- sD = starter decisions
-------------- (CG*9) + ((CG-sD)*9) = sIP (Starter IP)
-------------- rIP = IP - sIP
And then check it to make sure it makes sense for that particular pitcher. Meaning a guy with 5 relief appearances doesn't have 45 rIP, stuff like that.


Joe, I don't get this. If CG = 0, the formula reduces to -sD*9 = sIP, and of course neither sD nor sIP are negative. Is it a simple minus sign somewhere?
   168. DanG Posted: June 08, 2006 at 07:42 PM (#2056391)
Just for fun, here are the "leading" relievers for Hoyt's prime decade, 1961-70:

1961-70         RG  GF  SV  ERA
H
Wilhelm     565 407 158 2.10
R
Perranoski  655 414 172 2.54
B
Locker      388 190  61 2.67
D
McMahon     558 308  96 2.70
S
Miller      421 312 136 2.72
A
Worthington 383 247  98 2.78
T
Abernathy   500 329 120 2.87
D
Radatz      381 297 122 3.13
E
Fisher      499 273  73 3.15
R
Face        456 326 118 3.21
R
Kline       465 310 103 3.21
L
McDaniel    550 339 115 3.29 
   169. Dr. Chaleeko Posted: June 08, 2006 at 08:04 PM (#2056429)
OK, here's a little something. I don't quite know what to do with it yet.

I went looking for a way to estimate leverage for Wilhelm-type relievers, hoping to come kind of close to Tango/BP's LEV numbers, but using only the stats at hand like SV, G, GF, Rel W. Reason for doing so? Well, mostly because Holds aren't easy to find, and it seemed like something that we could apply it to the pre-Retrosheet relievers, maybe....

Well, after a little futzing with about the forty top guys of 1960-1970ish---theWilhem era---, I've got a decent proxy I think....

Turns out that

((SV+RelW)*3) + ((GF-SV)*8)

will do a pretty good job of getting close to the Tango LEV of a larger group. In fact, the mean error rate for this equation (comparing the result to the Tango LEV for each guy in my study group) was 1.002. 30 of the 40 fell within 10 percentage points of the mean error.

Not sure if this is helpful, or even if I'm examining the rate of error correctly, but I thought I'd put it out there in case it's useful to anyone.
   170. Joey Numbaz (Scruff) Posted: June 09, 2006 at 02:54 AM (#2057231)
"Joe, I don't get this. If CG = 0, the formula reduces to -sD*9 = sIP, and of course neither sD nor sIP are negative. Is it a simple minus sign somewhere?"

Tom - flip flop CG-sD to sD-CG. Whoops.

Doesn't matter though, since I think I've found relief innings. Thanks for pointing it out though!

*****

Dr. C - we discussed this off line, but I should be able to a list of all the pitchers I've done up sometime soon.

My 're-jigging' as you call it is pretty substantial though - numbers won't be directly comparable to WARP - I use a signficantly higher placement level (5.23 for rIP, 5.48 for sIP, WARP uses 6.00 - all of those are in relation to a 4.50 R/G environment). There are other tweaks as well.

Also, I won't be able to do the relief innings adjustments for SP before Monday. So the starters will all just be compared as normal. Eventually (hopefully for 1979), I'll have starters adjusted for the importance of their relief innings.

If there are particular pitchers you want to see on the list, please let me know ASAP.

**************

I have a question regarding my system. As those who have been following understand, I use a system for translating IP that is similar to what Prospectus uses, however, I tweaked it, I don't use the top 5 in the league = 275 IP, I take an average of the league leaders in a range from 25-75% of the number of teams in the league. So in an 8 team league, I use the #3, #4, #5 and #6 in the league in IP, and make that equivalent to 258.3 IP.

But here's the rub - if I'm going to give the starters credit for leverage in their relief IP, is this going to throw off my translated IP numbers.

My plan is to not translate relief innings. Older relief aces will get an advantage from their innings. Modern relief aces will get an advantage in that they prevent more runs per inning (higher ERA+) because of their reduced workload.

Here's the specific issue.

Take Walter Johnson he started 30 games in 1908, relieved in 6, finished them all. He threw 256.3 innings that year, which translates to 213.3 in my system.

Let's say he threw 12 relief innings that year (I'll know the actual answer next week). If those innings are leveraged at 1.0 his translated innings in theory, should stay at 213.3. If they were leveraged higher than 1.0, translated innings would go higher, if lower than 1.0, translated innings go lower. In theory.

The problem is that it doesn't work out that way. If I give him credit for 12 rIP with an LI of 1.0, his tIP goes from 213.3 to 215.4.

The reason is that all of Johnson's innings are reduced by 17% because pitchers in 1908 threw more innings than they have historically.

But while my system says starter innings are adjusted for historical workload, it says and inning is an inning for a reliever (I have my reasons for both of those choices). So those 12 innings are getting 'full credit' while the 244.3 sIP are getting 83% credit.

1) Is this a problem?

2) Assuming it is - how should I rectify it?

Instead of normalizing starting IP to league leaders in total IP, I could normalize starter IP to the league leaders in starting IP. But that's an issue because I have idea where to find as many as the top 12 in each league in starting IP. I don't know if such lists exist. I could subtract relief IP from the leaders in total IP as a proxy, but that would still be a ton of work. It doesn't take very long to just input a list of numbers, but it will take a long time to review that list of numbers and make a subtraction from each one - especially when each subtraction will require flipping to a new page of an encylopedia.

Any help or suggestions would be greatly appreciated.
   171. Rob_Wood Posted: June 09, 2006 at 04:37 AM (#2057376)
Joe, this may or may not help. I would estimate the pct of innings of the top starting pitchers achieved in relief for each year (perhaps by decade). Then estimate the typical leverage of these relief innings. Then use the percentage and the leverage to adjust the top innings pitched to derive a "mega-"innings pitched average that is used in the normalization. In this way you only have to look up the relief innings of the pitchers you are interested in. (Of course, BB-Ref has the list of top innings for each year that you are probably already using.) Hope this makes sense.
   172. Howie Menckel Posted: June 09, 2006 at 04:41 AM (#2057377)
Well, I consider holds and save pcts to be fairly useless in the fireman era.

When you come into a 3-2 game in the 7th with bases loaded one out, and allow a sac fly to tie it and retire the side, that is a plus for your team vs a "blown save" or "no hold."

Holds nowadays are often "pitched the 8th of a 4-2 game, and left with the lead."

That is an entirely different task.

I love the research efforts, because we may be able to pluck out the underappreciated guy from 1960-90.
But the usage patterns may not make it easy for us.

Is it my imagination, or could Win totals be more relevant for firemen than for SPs at any pt since the early 20th century? It'll never be perfect, but there's a stretch heere where these guys usually pitched multiple innings with the game on the line. SPs always have had many games where they basically can't win or can't lose, but this window has pitchers entering games at the most crucial pts - not with cushy leads in the 9th...
   173. Joey Numbaz (Scruff) Posted: June 09, 2006 at 05:10 AM (#2057382)
Howie - I like holds more than most. You can get an RBI with a groundout, you can get a Quality Start giving up a 4.50 ERA. The worst case scenario doesn't render a stat meaningless.

Generally the top set up guys are the guys with the most holds. It is a generally meaningful stat. It's not perfect, but it's better than not using anything.

I certainly agree that Wins for relievers could be very meaningful. Sometimes they are 'vultured' but again, worst case scenarios don't kill a stat.

I will definitely be incorporating rW, rL, SV into any guesstimate at LI for years before 1960. Also considering GF and whether or not to base it on those numbers per IP or per G. Dr. C and I have been discussing this over email.

*********

Rob - as I said above, I'm planning to actually use rW, rL, SV, GF, rIP, rG to estimate a LI for each starter's reliever innings pre-1960. I assume that will work even better than your suggestion of using era norms, right?

You are right, I am using the top IP lists from BB-Ref, and the Sabermetric Encyclopedia. Unfortunately neither breaks it down into starter IP, which would be ideal. I have a lead for a source of relief innings in digital format, if that ends up coming through, perhaps I could use that to generate top starter IP lists (if starter innings are also included). I'll keep you posted there.

Until I can get to that point though, or until I get a suggestion as to something else, I'm thinking of just going with:

sIP normalized based on tIP of the league leaders

+

(rIP * LI)

. . . is that a major issue?
   174. TomH Posted: June 09, 2006 at 11:50 AM (#2057444)
Joe, your last equation seems useful, and pretty close if I understand it correctly to what Rob suggested, which I was also happy with.
   175. Joey Numbaz (Scruff) Posted: June 09, 2006 at 11:57 AM (#2057448)
As requested by Dr. C a list that shows where Wilhelm stands compared to his contemporary relief aces that are retired by 1978. I'm not an expert on this era, if I missed an obvious guy, let me know:

Pitcher   PenAdd aWAR Top3Con Top5  AdjIP
Wilhelm    .733  62.5  13.3   27.7  2974.3
Miller     .503  42.3  15.1   28.5  2098.3
McDaniel   .431  37.0  10.2   22.8  2495.0
Face       .367  32.1  12.2   18.8  1801.3
Radatz     .285  23.1  19.6   23.1   952.7
Fisher     .218  19.0  11.2   16.3  1612.0
Abernathy  .185  16.3   8.2   13.2  1331.0 


Remember that aWAR is not directly comparable to WARP. My replacement level is much higher.

Some notable starting pitchers:

Pitcher   PenAdd aWAR Top3Con Top5  AdjIP
Johnson   1.940 148.9  36.9   57.0  5207.7
Drysdale   .848  69.4  21.2   33.8  3263.3
Bunning    .843  68.7  23.2   36.4  3692.0
Pierce     .788  65.5  18.7   33.5  3363.0
Bridges    .773  64.8  17.5   29.6  3087.3
Wilhelm    .733  62.5  13.3   27.7  2974.3
Grimes     .722  60.2  16.7   31.7  3900.3
Walters    .688  56.1  24.6   36.0  3045.0
Trout      .681  55.7  22.2   33.7  2770.3
Trucks     .679  57.4  15.4   28.5  3122.3
Newcombe   .678  56.8  16.4   29.8  3099.0
Willis     .676  56.0  18.2   32.5  3174.0 


Those numbers include war credit and demerits.

Miller's .503 PA puts him below Hippo Vaughn (.565), Mickey Welch (.562), Eddie Rommel (.545) and above Addie Joss (.496) and Sam Leever (.468), just for some perpsective.
   176. Dr. Chaleeko Posted: June 09, 2006 at 01:02 PM (#2057487)
Thanks, Joe! Boy this is tough. Wilhelm's got plenty of innings, and he does well by pennants added, yet he's got the worst WAR peak among this group of pitchers. Oy....

Maybe instead of Wins Above Replacement, we could call it Pitcher Excellence Above Cohort Expectation (PEACE).
   177. TomH Posted: June 09, 2006 at 02:17 PM (#2057553)
Doc, you brought a smile to my face....

Joe, nice work. Gotta Q, tho. Using your table, dividing PenAdd by AdjIP (to get a effectivess rate type of measure), I find Bridges has a higher pennants added per 1000 leveraged-innings than Wilhelm (.250 to .246). Given the huge career ERA+ difference between them (146 to 126), which would come out to about .50 in ERA, I'm surprised Bridges closes that gap, since you didn't adjust relief innings to that extent. Can you find something in your numbers that might explain this? (of course, I'd have Wilhelm slightly higher on league strength anyway, but....)
   178. Steve Treder Posted: June 09, 2006 at 03:32 PM (#2057639)
As requested by Dr. C a list that shows where Wilhelm stands compared to his contemporary relief aces that are retired by 1978. I'm not an expert on this era, if I missed an obvious guy, let me know:

Fascinating stuff, Joe.

Some other names you might consider running it for would be Ron Perranoski, Al Worthington, Ron Kline, Phil Regan, Frank Linzy, Bob Lee, Hal Woodeshick, and Jack Baldschun.
   179. sunnyday2 Posted: June 09, 2006 at 05:32 PM (#2057798)
Before I ran those guys, I would be interested in seeing Ellis Kinder and Firpo.
   180. Joey Numbaz (Scruff) Posted: June 09, 2006 at 07:10 PM (#2057911)
Tom - I have a higher replacement level for relief innings - that closes the gap.

Dr. C - how do you get Wilhelm has having the worst peak? Only Radatz and Miller have a higher Top 3 consecutive and only Miller has a higher top 5.

Oh wait - you mean compared to the starters. Yeah, but as we've said, it's awfully hard for a reliever to get up to par peak wise, especially since the starters have a 25 point ERA head start over replacement and throw more innings.

Wilhelm is a weird case in that his best years don't line up, so 3 consecutive is kind of unfair to a guy like that. I just listed it because some people like it.

Pennants Added pulls that together, it takes longer to get there with less big year value, but if you pitch effectively for 2 decades, you can.

I think what the numbers show is that an ace reliever, used in 1960s fashion can get to the point where they bump up against and sometimes cross over into the bottom rung of elite pitchers. I mean Radatz's top 3 fits right in with the starters, but it cost him his career in terms of wear and tear. Wilhelm's best 5 isn't very far behind and his career value is a lot better than most which gets him up there.

Also note that list gives no credit for Bucky Walters before he started pitching - forgot to mention that. And look at Burleigh Grimes, I think we need to go back and look at him again. His hitting is what breaks him out of that crowd. And poor hitting is what drops Trucks back into it.

By the way Pennants Added also shows what Bill James' study in The Politcs of Glory did. That if you have two pitchers with similar career value, the guys with the higher peaks are slightly better. Note how Walters and Trout are slightly ahead of the others around 55-57 aWAR.

But it also shows what that study did too - that no one mentions, that once you get significantly behind on career value a strong peak usually isn't enough to pull you past. That's because the guys significantly ahead of you on career value generally couldn't have done it without putting up a least a decent peak of their own. I've said it before that study shows Drysdale ahead of Pappas and Carlton ahead of Sutton. It most definitely doesn't show Drysdale ahead of Sutton - which is the leap many have taken.

Also don't forget that PA are calculated only based on the NL from 1876-1977 for the 1978 election. I've got to work the AL in at some point, but that's half a night of data entry that I haven't had time for yet. As pennant winners move closer to .500 and the leagues tighten up from top to bottom, the higher peak seasons will move up a bit.

Steve, I'll try to get to those guys tonight. After I spent 15-20 minutes running Abernathy (guys who switch teams mid-season are a pain in the butt and it's even worse if they play in two different leagues in the same year) and another 15 on Fisher and saw that they weren't close, I decided to tighten up the list a bit. I had Worthington there, but decided to pass in the interests of time. I threw in Radatz because I had to see that insane peak. I'll try to get to as many of those guys as I can tonight.

I'll eventually (next week) get Kinder and Marberry in there too Marc - but the problem is pre-1960 I'm estimating Leverage Index and starting IP vs. relief IP.
   181. Joey Numbaz (Scruff) Posted: June 09, 2006 at 07:12 PM (#2057917)
Missed the PEACE comment the first time - that's hysterical. I like it.
   182. Joey Numbaz (Scruff) Posted: June 09, 2006 at 07:17 PM (#2057925)
"but the problem is pre-1960 I'm estimating Leverage Index and starting IP vs. relief IP."

Note that that is only until I get my 1969 MacMillan with relief IP broken out. I'm still going to have come up with something to estimate LI. Dr. C is taking a good crack at it right now, and I'm hoping to talk to an actuary friend of mine, but she's blown me off so far, hasn't returned my email yet. She's probably thinking that I'm a geek and laughing about it :-)

I'll drop Tango a note on it as well.
   183. Steve Treder Posted: June 09, 2006 at 07:35 PM (#2057958)
Steve, I'll try to get to those guys tonight. After I spent 15-20 minutes running Abernathy (guys who switch teams mid-season are a pain in the butt and it's even worse if they play in two different leagues in the same year) and another 15 on Fisher and saw that they weren't close, I decided to tighten up the list a bit. I had Worthington there, but decided to pass in the interests of time. I threw in Radatz because I had to see that insane peak. I'll try to get to as many of those guys as I can tonight.

Don't sweat it. None of those guys is anything close to a serious HOM candidate; it would just be interesting to see how they compare against each other.

And speaking of insane peaks: Bob Lee's was close to that of Radatz, and his subsequent flameout was even more sudden and total.
   184. Steve Treder Posted: June 09, 2006 at 07:38 PM (#2057962)
Bob Lee's ERA+ figures, by season: 218, 177, 123, 81, 62.

Do you think there might be a trend of some kind there?
   185. Joey Numbaz (Scruff) Posted: June 09, 2006 at 07:57 PM (#2057983)
Wow. I'll definitely run Lee. Doesn't look like he'll take long :-)

A hard core peak guy should love Radatz if he gives any kind of positional bonus.

The relievers take a long time because I have to reload the inherited runners thing on Prospectus for each season, etc.. You'd think they have career lists season by season for the pitchers too, instead of just seasonal lists. If I'm missing that, please let me know! I shelled out for a 4.95 monthly sub so I could sort the lists.

I've got every serious Hall of Merit candidate in there at least. After finishing off the 1960s relief aces, the next step is to fill the non-HoM, HoF guys, like Marquard, Haines, etc.. I should be able to get to all of that tonight.

Then I've got to start next week with entering all of the starter work as relievers.

Eventually I have to go and get all of the HoMers too, so we have some perspective on how all of these guys fair relative to their era. I just thought it was more important to get the current candidates first. That will help me to identify if I've got any inherent bias in the system against a certain time frame. I already know of one - the pre-1893 guys, when normalized aren't going to end up with career value like post-1893 guys. Their careers just weren't as long.

I also want to update the post 1960 starters with information on their bullpen support regarding inherited runners, but again, that involved reloading the pages. And did you see when you sort by name alphabetically, it sorts by FIRST name, WTF???

Actually, I'd be better off sorting by season alphabetically and just printing it all out. Because it's a secure site, I have to click a stinking YES/NO box with each seasonal reload. It'd be faster to sit there with all of the paper than having to click through. Less carpal tunnel too, my right pinky is starting to go numb . . .
   186. Dr. Chaleeko Posted: June 10, 2006 at 02:00 PM (#2059057)
I shelled out for a 4.95 monthly sub so I could sort the lists.
Not to shill for a third-party product, but if that $4.95 per month also buys you Joe Sheehan's "Prospectus Today" column, you'll be getting your money's worth.
   187. Joey Numbaz (Scruff) Posted: June 10, 2006 at 04:46 PM (#2059126)
Hey guys a few updates . . .

I finally got to putting in the AL 1901-77 for the Pennants Added calculation. Brought down the standard deviation a bit, which increased the spread on the peak guys a little bit. Unfortunately, I don't have time to update the lists above now, nothing earth shattering, but a slight bump for the peak guys from my eyeball observation.

I found (actually Tango pointed me in the right direction and Rob Wood gave me Pete Palmer's LI estimation formula.

[9*(rW+rL+(k*SV))]/IP

where k = the smaller of .25 or 1/[10*(leagSV/leagW)]. Basically once 40% of the league wins are saved, it starts decreasing the value of a save.

The final answer can be no higher than 2.00 or lower than .50.

I'll use that pre-1960 unless I or someone else comes up with something better. There's no real urgency in that department at this point, though my actuary friend did get back to me last night, so we'll see if I can convince her to go a little further with this.

This tweaked Wilhelm's numbers, but nothing earth shattering since I was already eyeball estimating anyway.

From eyeballing it with the actual LI's we know, I would say that it probably gets to the extremes too quickly (the .50 and 2.0 limits), meaning when it's high it might overestimate a little, and occasionally it may be too low when it's low, though this is less of an issue. I might take the final answer and multiply it by .9 unless that leaves a number under 1.00. If that leaves an answer under 1, then I'll probably leave it alone. That's open to debate though.

I wasn't able to run any new pitchers last night, the combination of lack of sleep, the numbing of my right pinky and thumb and overall burnout caused me to stop where I did. I'll be back on the horse with it no later than Tuesday evening . . .
   188. Paul Wendt Posted: June 10, 2006 at 06:54 PM (#2059260)
Pete Palmer's LI estimation formula.

[9*(rW+rL+(k*SV))]/IP
where k = the smaller of .25 or 1/[10*(leagSV/leagW)].


The misnamed "Relief Ranking" RNK is familiar Adjusted Pitching Runs, limited to relief work only, inflated by this decision factor. Introduction of the decision factor in TB4 (1995) bumped John Hiller 1973 up from 3.6 to 6.7 in the total pitcher index TPI. Total Baseball editions -- stats in the registers
   189. Joey Numbaz (Scruff) Posted: June 11, 2006 at 05:21 AM (#2059738)
I got the Encyclopedia today, was just skimming through it . . . Eddie Rommel might get the biggest relief bonus of any non-closer. Marberry is going to do pretty well too I think. Can't wait to get cranking on it early next week!

Pretty weird to get a book in the mail that's older than I am. I even smells like an old book. It's wild looking at an all-time HR list that has Hank Aaron 8th!
   190. Steve Treder Posted: June 11, 2006 at 06:53 AM (#2059762)
I even smells like an old book.

I recommends getting outside in the fresh air once in a while.

;-p
   191. Paul Wendt Posted: June 11, 2006 at 02:02 PM (#2059791)
Pretty weird to get a book in the mail that's older than I am.

I have the first Official Encyclopedia of Baseball (1951), also at a bargain price. That one will cover most of the older generation. Ernie Lanigan's Cyclopedia (1922) is always expensive but what is social security for?
   192. Paul Wendt Posted: June 11, 2006 at 02:38 PM (#2059795)
I have posted more on the evolution of the relief pitcher by win shares including the league- and staff-leading relievers every ten years. "Relievers" are all pitchers who worked more relief games than starts; it's interesting nonetheless. And a work in progress.
Relief Pitcher seasons by win shares - in progress
   193. John (You Can Call Me Grandma) Murphy Posted: June 11, 2006 at 03:13 PM (#2059801)
I have the first Official Encyclopedia of Baseball (1951), also at a bargain price. That one will cover most of the older generation.

That was the first baseball encylopedia that I ever checked out from a library in my life when I was in high school. Though it pales in comparison to later versions, I'm still fond of that old trailblazer.

I have the original Mac from '69, BTW. It's battered (I got it that way by trading for it for some inconsequential baseball books of mine), but I doubt I will ever get rid of it.
   194. jimd Posted: June 12, 2006 at 10:20 PM (#2061507)
That was the first baseball encylopedia that I ever checked out from a library in my life when I was in high school.

That was the first BB encylopedia I ever encountered, too. It was in the reference section of the public library attached to my junior high. Couldn't check it out, so I spent too many hours with it after school. Nerds R Us ;-)

It was a decade later that I got my first encylopedia of my own. It was a Neft, Cohen, et al, 1975 edition, paperbound. (Macmillan was too expensive for a young recently married couple.) I highlighted in yellow marker all the seasons by HOMers, which is when I first noticed the unusual concentration of early 1920's Giants, few of whom looked truly distinguished.
   195. Steve Treder Posted: June 12, 2006 at 10:30 PM (#2061517)
I have the original Mac from '69

Oh, me too. The best reference book I'd had before then was S.C. Thompson's All-Time Rosters of Major League Baseball Clubs, which was great, but suffered from the enormous limitation of including only batting average and won-lost record as its individual performance stats.

I'll never forget flipping open the MacMillan after unwrapping it from its mail-order package, in early '69. The first player's record I came across was George Caster ... omigod ... one year he was 4-19, with a 6.56 ERA!!

It was days before they could get me to sleep, eat, or do anything other than absorb the contents of that book.
   196. Joey Numbaz (Scruff) Posted: June 25, 2006 at 09:24 PM (#2075506)
Pitcher           PA  aDRA  tIP   WAR RSAR  PSup InRP  BRP  LI  LIP     1    2    3    4    5   Top3  Top5
Hoyt Wilhelm     .850 3.38 2905.7 69.6 665  1.7  10.0 18.7 1.4 1871.0  7.8  6.5  6.0  5.1  5.0  15.8  30.4
Stu Miller       .577 3.59 2087.3 46.4 443  0.4  27.0 
-1.6 1.4 1103.3  8.6  7.0  5.8  4.5  4.4  16.8  30.3
Lindy McDaniel   .541 4.08 2590.7 44.5 425  0.7   2.5 16.5 1.3 1672.7  7.5  7.1  5.1  3.7  3.5  13.2  27.0
Firpo Marberry   .501 3.96 2334.7 40.9 391  0.0   0.0  0.0 1.4  730.0  7.9  5.6  5.1  4.9  3.5  16.3  26.9
Roy Face         .463 3.75 1894.0 38.7 370  0.0  18.8 11.6 1.4 1186.3  5.7  5.4  5.1  4.8  3.0  13.7  23.9
Ron Perranoski   .382 3.90 1562.0 31.3 299  0.0   0.3 28.3 1.4 1170.7  6.2  6.1  4.4  3.9  3.7  12.0  24.3
Dick Radatz      .339 3.29  952.7 25.8 246  0.0   3.0  5.0 1.4  693.7  8.3  7.8  5.6  4.0  0.1  21.7  25.8
Clint Brown      .324 4.20 1688.7 27.6 263  0.0   0.0  0.0 1.4  595.0  4.9  4.5  3.5  3.5  2.7  11.1  19.2
Ron Kline        .302 4.61 2400.3 26.1 250 
-3.5 -28.6  3.7 1.2  871.0  5.2  3.8  3.3  2.2  2.2  10.3  16.8
Johnny Murphy    .298 4.09 1582.0 25.5 244  0.0   0.0  0.0 1.7  764.0  4.9  4.4  3.2  3.0  2.6   8.5  18.1
Al Worthington   .273 4.11 1470.7 23.8 227  0.0  
-4.2  4.0 1.3  838.0  4.8  3.4  2.3  2.3  1.9  10.5  14.7
Eddie Fisher     .263 4.46 1633.7 22.1 211  0.0   3.8 11.6 1.1 1186.0  6.4  3.6  3.4  2.7  2.3  12.8  18.5
Jim Konstanty    .248 4.09 1303.7 20.8 199  0.0   0.0  0.0 1.3  749.0  7.1  3.3  2.2  1.9  1.6  12.0  16.1
Ted Abernathy    .225 4.40 1330.7 19.3 184  0.0  
-2.9 20.9 1.2  913.0  5.5  3.7  2.2  1.8  1.7   9.3  14.9
Joe Page         .214 4.37 1137.3 16.9 162  0.0   0.0  0.0 1.7  510.3  7.1  6.5  1.4  0.8  0.7  15.0  16.5
Phil Regan       .205 4.54 1522.7 16.9 162  3.6  
-6.8  8.0 1.2  721.0  7.2  3.1  2.4  1.3  0.9  10.2  14.8
Frank Linzy      .175 4.34 1052.7 15.2 145 
-1.1 -11.6 20.6 1.3  812.3  4.3  2.8  2.6  2.3  1.5   9.4  13.5
Bob Lee          .152 3.74  597.3 12.4 119  0.3  
-4.5  2.3 1.3  457.7  5.3  5.2  1.5  0.4  0.0  12.0  12.4
Hal Woodeshick   .135 4.85  950.0 11.6 111 
-3.2 -13.2 28.6 1.2  567.3  4.1  3.0  2.2  1.7  0.6   9.3  11.6
Jack Baldschun   .058 5.33  737.7  5.3  51  0.0 
-29.2 22.6 1.1  704.0  2.1  1.4  1.3  0.6  0.0   4.7   5.3 

PA - Pennants Added; aDRA - my version of Defense adjusted runs allowed, which uses PythaganPat exponents, and the Baseball Prospectus adjustments from NRA to DERA, 4.50 is league average; tIP - my version of translated IP, which accounts for leverage of relief innings, and adjusts starters based on era norms based on the league leaders IP and the size of the league; WAR - my wins above replacement, using aDERA, tIP and accounting for pitcher hitting; RSAR - my version of runs saved above replacement; PSup - starting pitcher bullpen support, taken directly from Baseball Prospectus (negative means good support); InRP - Inherited Runs Prevented, taken directly from Baseball Prospectus (positive is good); BRP - bequeathed runs prevented, taken directly from Baseball Prospectus (positive means good support); LI - Leverage Index, 1960-2005, taken directly from Baseball Prospectus, 1871-1959 estimated based on Pete Palmer's formula, the only difference being that I cap it at 3.00, not 2.00. 1 - pitcher's best season WAR; 2 - pitcher's second best season WAR; 3,4,5 - figure it out, you are smart; Top3 - pitcher's top 3 consecutive seasons of WAR; Top5 - pitcher's top 5 individual WAR seasons.

The numbers in the chart include all seasons, not just reliever seasons. Any years where WAR or RSAR would be negative are zeroed out, under the premise that as long as someone is willing to play you, you can't have negative value.

The best individual reliever seasons that I've found among relievers retired prior to the 1979 election:

x1. 8.6 Stu Miller (1965)
x2. 8.3 Dick Radatz (1964)
x3. 7.8 Dick Radatz (1963)
x4. 7.5 Lindy McDaniel (1960)
x5. 7.2 Phil Regan (1966)
x6. 7.1 Joe Page (1949)
x7. 7.1 Jim Konstanty (1950)
x8. 7.1 Lindy McDaniel (1970)
x9. 7.0 Stu Miller (1961)
10. 6.5 Hoyt Wilhelm (1965)
11. 6.5 Joe Page (1947)
12. 6.4 Eddie Fisher (1965)
13. 6.2 Ron Perranoski (1969)
14. 6.1 Ron Perranoski (1963)
15. 5.7 Roy Face (1959)

My career value rankings would be exactly as on the chart above.

My peak rankings (highest established/sustained 2-3 season level) would be:

1. Stu Miller
2. Dick Radatz
3. Hoyt Wilhelm
4. Lindy McDaniel
5. Firpo Marberry
6. Joe Page
7. Ron Perranoski
8. Roy Face
Page 2 of 2 pages  < 1 2

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
Sebastian
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 0.9907 seconds
49 querie(s) executed