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

Monday, December 19, 2016

2018 Hall of Merit Ballot Discussion

Vladimir Guerrero (267), Sammy Sosa (255), Luis Tiant (189), Jeff Kent (179), Vic Willis (161), Kenny Lofton (151), Bobby Bonds (133), Ben Taylor (130), Buddy Bell (130) and Tommy Bridges (106) will be the top 10 returnees for 2018.

Jorge Posada (95), Bob Johnson (86), Urban Shocker (80), Dick Redding (79), Phil Rizzuto (79), Wally Schang (79) and Sal Bando (68) rounded out the top 20.

2018 - (December 4 - December 18, 2017) - elect 4

Name               HOFm HOFs Yrs WAR  WAR7 JAWS
Chipper Jones       180  70   19 85.0 46.6 65.8
Jim Thome           156  57   22 72.9 41.5 57.2
Scott Rolen          99  40   17 70.0 43.5 56.8
Andruw Jones        109  34   17 62.8 46.4 54.6
Johan Santana        82  35   12 51.4 44.8 48.1
Johnny Damon         90  45   18 56.0 32.8 44.4
Jamie Moyer          56  39   25 50.4 33.2 41.8
Carlos Zambrano      30  23   12 44.6 39.0 41.8
Omar Vizquel        120  42   24 45.3 26.6 36.0
Chris Carpenter      70  26   15 34.5 29.6 32.0
Livan Hernandez      41  16   17 31.1 27.8 29.4
Orlando Hudson       20  18   11 30.9 27.2 29.1
Kevin Millwood       34  20   16 29.4 24.8 27.1
Kerry Wood           24  14   14 27.7 25.0 26.4
Carlos Lee           78  35   14 28.2 23.4 25.8
Ben Sheets           19  11   10 23.4 22.3 22.8
Jack Wilson          12  16   12 23.5 20.9 22.2
Hideki Matsui        36  21   10 21.3 21.2 21.3
Aubrey Huff          30  20   13 20.2 22.5 21.3
Adam Kennedy         12  16   14 21.0 20.4 20.7
Jeff Suppan          11   9   17 17.4 18.3 17.8
Carl Pavano          16   6   14 16.9 18.5 17.7
Francisco Cordero    77   9   14 17.2 14.6 15.9
Miguel Batista       10   3   18 12.7 15.9 14.3
Jason Isringhausen   71   7   16 13.2 12.2 12.7
Brian Fuentes        48   9   12 10.7 11.3 11.0
Brad Lidge           48  10   11  8.2 12.4 10.3
Scott Podsednik      15  15   11  6.9  7.8  7.4
Guillermo Mota       13   7   14  6.3  7.6  7.0
JoeD has the Imperial March Stuck in His Head Posted: December 19, 2016 at 09:12 PM | 444 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 4 of 5 pages ‹ First  < 2 3 4 5 > 
   301. Carl Goetz Posted: December 06, 2017 at 01:40 PM (#5587418)
"Yeah, so I wouldn't yet trust the Andy Cooper MLE."
Doc, is there a reason you guys didn't do Redding in your 3 pitcher segments? I'm currently using Alex King's numbers and with scaling back Smith and Cooper, I now have Redding #1 among pitchers. Are these the numbers others are using? Is there any reason to believe they are overly generous? Just want to give him another look before I place him high on my ballot.
   302. Dr. Chaleeko Posted: December 06, 2017 at 02:00 PM (#5587461)
I haven't worked up Redding yet. I'm running through all the inducted players from the HOM and HOF first. He might be a while, but I'll try to tuck him in earlier and report on the findings.
   303. Carl Goetz Posted: December 06, 2017 at 02:08 PM (#5587466)
Any opinion on the reasonableness (word?) of the Alex King numbers?
   304. Dr. Chaleeko Posted: December 06, 2017 at 02:15 PM (#5587472)
I don't have an opinion on them. I'm sorry about that.
   305. Carl Goetz Posted: December 06, 2017 at 02:22 PM (#5587478)
No problem. I appreciate all of your work on MLEs. I think I'll stick with Alex King's numbers as they jive pretty well with the anecdotal evidence on Redding. Actually, may be conservative based on some of that evidence. Though I'll admit even as a Redding fan that I find it hard to believe he was as good as Walter Johnson.
   306. Carl Goetz Posted: December 06, 2017 at 10:48 PM (#5587804)
Ok, I worked with Pennants Added but I'm viewing that as more of a modified career value, but I am using it to some degree. I'm incorporating WAA and WAG from Baseball Gauge (Subbing zero for any negative years) as a good proxy for peak/prime value rather than an arbritary Best3/Best5 type system. I have a rough minimum career value of around 45 WAR (adjusted to 162 game season), Then I rank by WAG and make adjustments moving guys up the list if they are significantly better in WAR or WAA or PA than the guy above them. Its got some objective points to it as well as some subjective points. I basically want to focus on Greatness, but not ignore guys who were really good for a long time. Its not perfect, but I'm much more comfortable with this list than with previous iterations. I'll leave this out here for comment and make adjustments to the list based on whether I agree with comments for about the next week and then I'll post whatever iteration of this list is current to the ballot thread.

1 Jones, Chipper
2 Rolen, Scott
3 Thome, Jim
4 Bell, Buddy Still a big Buddy Bell fan and I feel he is the best returnee. We are light on 3B, so I see no reason to adjust at this time.
5 Redding, Dick
6 Appier, Kevin Best pitcher I have all MLBstats for. Love the 19.2 career WAG as a pitcher. Bigger peak than I expected going in.
7 Jones, Andruw
8 Bonds, Bobby
9 Dandridge, Ray
10 Munson, Thurman
11 Leach, Tommy Love the defense even though I adjusted it down a tad.
12 Taylor, Ben
13 Santana, Johan
14 Tiant, Luis
15 Chance, Frank
16 White, Roy
17 Sosa, Sammy
18 Schang, Wally
19 Parrish, Lance
20 Campaneris, Bert
21 Cicotte, Eddie
22 Kent, Jeff I gave him a best player at an under-elected position bump, but can't get higher than this at this time. 56.2 WAR/ 26.1 WAA / 16.6 WAG isn't good enough for me right now.
23 Bancroft, Dave
24 Cruz, Jose Sr
25 Rice, Sam
26 Lofton, Kenny I like him, but I like 7 OFs more, even after adjusting for DRA issues.
27 John, Tommy
28 Bridges, Tommy I'm warming to Tommy, but not to the point where he's on my ballot. Give me time, Tommy fans; this is way up from before.
29 Cey, Ron
30 Evers, Johnny
31 Rizzuto, Phil
32 Olerud, John

I think Vlad Guerrero is the only top 10 guy not listed above. I need more than 52.0 career WAR, 24.6 WAA, and 14.0 WAG from a corner OF. He's currently around #15 on my OF list.
   307. Michael J. Binkley's anxiety closet Posted: December 09, 2017 at 08:10 AM (#5589161)
Brock -

Not to speak for Doc, but as someone else who will have Fletcher on my ballot and Maranville off, let me offer my reasons. I think the main reason is DRA. I use it for 1/3 of my defensive value for players (and IIRC, Doc uses it for 2/3 - at least for non-catchers). And although DRA has Maranville as a very good defensive SS, Fletcher blows him out of the water on that metric (34.4 dWAA vs 13.9, not adjusting for season-length, Maranville missing most of 1918 due to WW1, etc).

This should account for most of the discrepancy in our respective evaluations. I also tend to be peak heavy, so since Fletcher amassed as much or more value than Maranville in fewer years, that also accounts for his higher placement for me.

- Michael Mengel

P.S. I know that you’re not the biggest fan of WAR and prefer WS, so I understand why we arrive at different values.
   308. Dr. Chaleeko Posted: December 09, 2017 at 09:15 AM (#5589182)
Brock,

Responding to your challenge in the ballot thread re Fletcher and Rabbit. I think it can be summed up simply and has relatively little to do with the relative merits of various fielding systems.
Rabbit: -228 batting runs
Fletcher: -11 batting runs

This is, to me, something I failed to acknowledge on my ballot and it is Fletcher’s secret weapon. Not only do you get an amazing glove, you get a league-average bat (if we think about it in a year by year basis). Now think about Ozzie. He’s got an OPS+ in the high 80s, but his 80 baserunning runs and 23 DP avoidance runs are a wealth of additional value. Fletcher and Maranville don’t have PBP generated running and DP figures. I’ve run some estimates on them and Rabbit picks up some value but nothing close to Ozzie. We will have to wait for BBREF and Retrosheet to confirm. Still as a RH hitter Rabbit won’t be more than a few runs to the good on DPs and could be a lot worse. Same for Fletcher, so it’s not as though there is a huge source of hidden value for them as there is for, say, Sam Rice or Harry Hooper (who also would benefit from the value generated by throwing that non PBP doesn’t seem to pick up as much of).

Anyway, so even apart from DRA, the case for Fletcher over Maranville from this perspective is not wrong. I can’t speak to a WS perspective. That said, there is a razor thin margin between any strong backlog candidates. If Rabbit had just missed my ballot, that too is entirely defensible due to how tightly packed these guys are around the borderline. There’s another dozen guys who could have made my ballot too. If I’m honest with myself,
A) my preferred Uberstat (bbref WAR) lacks the precision to split hairs
B) defensive stats are less precise than we’d like
C) my sifting/sorting tool (CHEWS+) lacks this level of necessary precision
D) lore and history and award votes lack a great deal of precision
E) my own analysis, reasoning, and general mental faculties, like the average pointy eared Vulcan, are pretty good, but I doubt they are precise enough for this minute level of differentiation either.

I just can’t say with assurance that something is wrong or correct. What I can say is that I’ve thought it through, I am systematic in my approach, that I’m willing to change my mind, and that I see a big enough gap between Fletcher and Rabbit that the latter isn’t much of a factor in my ballot construction while the former is #7. But, hey, I’ve been known to wrong!
   309. AJMcCringleberry Posted: December 10, 2017 at 12:10 AM (#5589471)
Hey everyone. I haven't voted in a few years, but was trying to get a ballot together for this year. I'm posting a preliminary ballot, I'm using WAR (b-ref and fangraphs) mostly. I've only looked at about half the players I want to (and none of the negro leaguers yet), so if you have questions, comments, suggestions, I'm open.

1. Chipper Jones
2. Scott Rolen
3. Jim McCormick
4. Andruw Jones
5. Jim Thome
6. Kenny Lofton
7. Sammy Sosa
8. Vic Willis
9. Buddy Bell
10. Sal Bando
11. Luis Tiant
12. Tommy John
13. Bobby Bonds
14. John Olerud
15. Vladimir Guerrero
16. Bob Johnson
17. Jeff Kent
18. Johan Santana
19. Urban Shocker
20. Javier Vazquez
   310. Jaack Posted: December 10, 2017 at 01:31 AM (#5589481)
My biggest suggestion would be to take a longer look at Jim McCormick. Dr. Chaleeko made some good points on the previous page about the issues with inducting more 1880s pitchers, and I tend to agree. That era is already well represented and WAR as a model is not particularly good at adapting to the gaudy inning totals of the 1880s. If you look at all time single season WAR leaders you'll notice that of the top 30 seasons by RA9-WAR, 27 are in the 19th century, and 18 are in the 1880s. By FIP it's a bit better, but still 6 of the top 10 are in the 1880s. There are nearly a dozen uninducted pitchers from the 1880s with pretty similar cases. You have to ask why McCormick over Welch, Buffinton, Whitney, or Mullane.
   311. bachslunch Posted: December 10, 2017 at 05:26 AM (#5589488)
@310: I ranked McCormick high on my ballot as well. Of the pitchers mentioned, he was definitely the best available. By BBReF WAR with IP:

McCormick: 75.8/4275.2
Mullane: 67.8/4531.1
Welch: 63.1/4802.0
Buffinton: 62.1/3404.0
Whitney: 55.6/3496.1

McCormick has the best WAR of any pitcher not in the HOM by a good margin.

For the rest, I had Welch in my 16-25 range while the rest did not appear on my ballot. I discounted Mullane because he spent half his career and all his best years in the AA, which may not be the best quality league. Buffinton’s WAR was just a little behind Welch’s but he has fewer innings and I tend to prefer career over peak. Whitney is well behind the rest.

I see the 19th century as a time with a glut of excellent pitchers, but to me that doesn’t mean I shouldn’t vote for the best one available. There are gluts and deficits at various positions at various times in baseball history. Others may feel differently of course.
   312. Jaack Posted: December 10, 2017 at 03:40 PM (#5589607)
As I said before, there are tons of problems with using WAR to evaluate 1880s pitchers. WAR was built for modern seasons - it simply wasn't designed with the conditions of 1880s pitching in mind. Pitchers were able to throw comparably enormous amounts of innings because the mound was closer to home plate. To add on, you can argue pitching was less important than ever - the difficulties of playing defense made that a much larger portion of preventing runs than it is today, or even in the 20s and 30s. Furthermore, I think using BBref pitching WAR as the WAR of choice has some issues - it's defensive adjustments seem shaky for any era, but especially for early baseball. Fangraphs RA9-WAR, which, like bWAR uses RA as its base, does not make these adjustments, so it's a little more pure. I think for the sake of accuracy it's better to start there and make your own defensive adjustments as you see fit. But for now, I'm going to use it because it's a superior raw number.

First we need to start understanding these numbers in context. The scaling of WAR for full time players is supposed to be something like
8 WAR=MVP
5 WAR=All Star
2 WAR=Average
0 WAR=Replacement
These function pretty well for most eras. Here's how the ratios break down over a few arbitrary seasons from MLB history.

2017:
3 MVP type seasons (7+ WAR) - 5%
13 All Star type seasons (4-6.9 WAR) - 22%
37 Average type seasons (1-3.9 WAR) - 64%
5 Replacement type seasons (<1 WAR) - 9%

1996:
7 MVP - 9%
21 All Star - 26%
42 Average - 51%
12 Replacement - 16%

1965:
7 MVP - 10%
16 All Star - 22%
40 Average - 55%
10 Replacement - 14%

1933:
4 MVP - 6%
18 All Star - 29%
30 Average - 48%
10 Replacement - 16%

1909:
7 MVP - 11%
15 All Star - 23%
31 Average - 47%
12 Replacement - 18%

The rates aren't totally consistent, but it's definitely within fudgible margins. It's probably worth noting that modern pitchers are becoming less polarized (my guess is that it's mostly due to increased reliever usage - good pitchers don't pitch quite enough to jump into the next tier, bad pitchers simply don't qualify anymore). But compare those ratios to 1880s ones.

1882:
8 MVP - 24%
6 All Star - 18%
6 Average - 18%
14 Replacement - 42% (!)

1886:
12 MVP - 27%
7 All Star - 16%
14 Average - 31%
12 Replacement - 27%

This isn't attributable to there just being a glut of good pitchers. Take those 12 MVP type season in 1886. 4 of them were from HoMers - Caruthers, Keefe, Clarkson, and Galvin. Mullane and McCormick make two more. But the other six - Dave Foutz, Lady Baldwin, Ed Morris, Toad Ramsey, Charlie Ferguson, and Dan Casey - have no one arguing their case. I just don’t think you can look at these numbers an attribute it to a glut of good pitchers. If it were that, we'd see the same names, over and over, but in the 1880s, that just isn't true. Furthermore there were a glut of good pitchers in the 1960s, but the league wide numbers still look reasonably similar to the rest of the 20th century. There's clearly something else at work here.

By RA9 WAR, the top 14 pitching seasons in baseball history happened prior to the 1890s. 90% of the top 50 happened in the 19th century. You can even see a similar effect when you look at negative WAR seasons – nine of the worst ten pitching seasons by RA9-WAR happened in the 19th century, and the only one that didn’t was in 1904. There has been one season of -3.0 WAR or worse in the past century. There were two seasons of -5.0 WAR or worse in 1883 alone. The WAR numbers simply don’t scale at all and taking them at face value is asking for trouble.

To wrap back to McCormick, I’ll admit, he’s probably one of the best 10 pitchers of pre 1893 baseball. But he’s clearly worse than Keefe, Clarkson, and Radbourn. Caruthers was about as good as McCormick, maybe a smidgen worse, but he also had a lot of bat to push him over the edge (and even then, I’m not the biggest fan of Caruthers as a HoMer). The only inductee that is even comparable to McCormick is Pud Galvin. But Galvin pitched 1000 more innings than anyone else in the era, and 1800 more than McCormick. McCormick never led the league in RA9 WAR, and only had three appearances in the top five.

McCormick looks an awful lot like the 1880s adjusted version of Mark Buehrle. Consistently pretty good for a long while, but never really an elite pitcher and didn’t pitch an overwhelming amount of innings like a Tom Glavine, Don Sutton, or Pud Galvin. His case is that his WAR is better than anyone else from his era that’s not yet inducted. But it’s still worse than everyone who is inducted, and WAR is really quite bad ad evaluating pitchers from that era.
   313. bachslunch Posted: December 10, 2017 at 04:40 PM (#5589641)
The Caruthers vs. McCormick comparison caveat for me is that the former pitched half his career and all his best seasons in the AA. And his numbers fell off notably once he got into the NL. As said above, I’m unsure if that league was as good as the NL and am currently treating it as if it wasn’t. McCormick pitched all but one season in the NL.

Having McCormick be one of the best 10 pitchers pre-1893 is a good enough reason to vote for him, am thinking. YMMV.
   314. 6 - 4 - 3 Posted: December 10, 2017 at 08:12 PM (#5589725)
I rely on WAA, not WAR for HOF. This has the effect of penalizing guys who compiled stats by being mediocre (i.e., positive WAR but negative WAA). In my view, above average is a demonstration of greatness; being above replacement level is a demonstration of minimal competence.

So preliminary ballot:

1) C. Jones: Inner circle
2) Rolen: Not quite inner circle, but more than deserving
3) Lofton: Best leadoff man of his generation
4) Thome: One of the best pure power hitters in history
5) A. Jones: Best defensive CFer that I've ever seen in his prime
6) Willis: Going entirely by the stats
7) Tiant: One of the better pitchers not in the HOF
8) Bell: Maybe one of these days, one of the better 3B not in the HOF
9) Santana: HOF peak, but falls short on career value
10) Bo. Bonds: Ibid
11) Appier: Better pitcher than most give him credit for
12) Guerrero: Very steep decline for what appeared to be a surefire HOFer at the start of his career
13) Sosa: Obviously worthy peak, hurt by years of mediocrity on both ends of his career
14) Olerud: Underrated 1B
15) Kent: Late start and the 1994-5 strike hurt him
   315. Mike Webber Posted: December 10, 2017 at 09:25 PM (#5589790)
Having McCormick be one of the best 10 pitchers pre-1893 is a good enough reason to vote for him, am thinking. YMMV.


I decided to look into this, because I wasn't sure if this was true or not. Might be true, heck the top 15 might be true. But after a little bit of messing with the BB ref play index,

Count me as one of the people with varying mileage.


Pre-1893 there were about 272 major league team seasons, roughly. 17 seasons of the NL, plus 17 seasons of the NA, AA, UA, PL combined, with 8 teams per year/per league. That isn't exactly right of course, because the UA didn't really have any teams - Sorry Fred Dunlap - but it's close enough for government work.

Which works out equivalent to 17 seasons of MLB in any given period between 1901 and 1960. After my research my conclusion is "Nobody on this list thinks that all of the ten best pitchers in any of those periods is HOM worthy."

Let's try 1943 - 1960. By BBref War - Spahn, Robin Roberts, Hal Newhouser, Early Wynn, Billy Pierce - ok that's a strong start, but then, Murry Dickson, Harry Breechen, Ned Garver, Virgil Trucks, Bob Lemon. Anyone passionate about Ned Garver?

But Mike, World War 2!

Ok, that's a good point. How about we check the deadball era? Heck we'll include the Federal League to give it a little boost. 1901-1917 -That way we also miss most of WW1 for baseball purposes. And this list is much stronger.

Walter, Christy, Plank, Cy, Big Ed Walsh - those are a strong front five. Pete Alexander, who didn't even start until 1911, Rube Waddell, Mordecai Brown, Vic Willis, Nap Rucker. 11-13 are Iron Man McGinnity, Addie Joss, and Chief Bender. Even here it's pretty clear Willis and Rucker aren't guys that have had much real support.

What about 1921 to 1937 between the Wars? Lefty Grove, Dazzy Vance, Carl Hubbell, Waite Hoyt. Wes Farell, Red Faber, Ted Lyons, Eddie Rommel, George Uhle, Eppa Rixey. Again it would hard to find 10.

Of course we get to 270 major league team seasons in 9 years now - that's a pretty tight window, let's see what it spits out. Kershaw, Verlander, Greinke, Scherzer, Hamels. King Felix, Sale, Price, Lester, Cueto. That's maybe a little better than I thought it would be.


I think it would be fair to say the top 5 in any 17 year period - or 270 season period - are pretty much slam dunks for the HOM. And if you said there are 7 for sure in any period of that length, I wouldn't argue. But finding 3 more? Well, that seems to be stretching the truth just a little too far.

YMMV
   316. Bleed the Freak Posted: December 10, 2017 at 10:45 PM (#5589849)
@310: I ranked McCormick high on my ballot as well. Of the pitchers mentioned, he was definitely the best available. By BBReF WAR with IP:

McCormick: 75.8/4275.2
Mullane: 67.8/4531.1
Welch: 63.1/4802.0
Buffinton: 62.1/3404.0
Whitney: 55.6/3496.1

McCormick has the best WAR of any pitcher not in the HOM by a good margin.

For the rest, I had Welch in my 16-25 range while the rest did not appear on my ballot. I discounted Mullane because he spent half his career and all his best years in the AA, which may not be the best quality league. Buffinton’s WAR was just a little behind Welch’s but he has fewer innings and I tend to prefer career over peak. Whitney is well behind the rest.

I see the 19th century as a time with a glut of excellent pitchers, but to me that doesn’t mean I shouldn’t vote for the best one available. There are gluts and deficits at various positions at various times in baseball history. Others may feel differently of course.


He has the best Baseball-Reference WAR, including his best season destroying weak UA competition.
He drops from 75.8 to 55.9 Baseball Gauge WAR and 41.3 Fangraphs WAR.
He is an interesting case, but a blend of the metrics shows him as a part of the remaining glut of 1880s candidates (not a clear winner), a group of which we have already elected a healthy amount of.
   317. Kiko Sakata Posted: December 11, 2017 at 01:08 AM (#5589879)
My Player won-lost records should be updated now: partial seasons for 1925 - 1940 (typically Retrosheet has play-by-play for 75-80% of games these seasons) along with 1922 and 1921 (NL only for the latter); complete data 1941 - 2017. I will probably post a preliminary ballot in this thread tomorrow (Monday) night and give it a couple of days before moving it to the Ballot Thread. Please let me know if anything doesn't seem to be working quite right on the website.
   318. bachslunch Posted: December 11, 2017 at 08:01 AM (#5589900)
@315: Willis is actually on my ballot and on those of several other voters. If memory serves, I've also seen ballots here with Uhle, Joss, and Rixey. One might argue that not all time periods are equally populated with deserving pitchers.

@316: McCormick did play one season in the UA, which I'm fine with considering an inferior league. But that's the only season he spent outside the NL. Folks like Mullane and Whitney spent half their career outside of that league and didn't pitch as well once they left. Of the rest from the century, Welch is in my 16-25 range and has appeared on other ballots; Buffinton and Mullane have as well.

Look, there are plenty of approaches one can take to this, otherwise everyone's ballot would synchronize in lockstep. For better or worse, except for Negro League candidates, I'm basing things on BBReF WAR and tweaking from there. Maybe one can argue that this is heresy, and maybe I shouldn't be voting at all. If so, so be it -- after lurking for a while, I decided to start voting and I don't plan to turn back any time soon.
   319. Dr. Chaleeko Posted: December 11, 2017 at 08:13 AM (#5589903)
Bachslunch and 6-4-3,

Absolutely vote!!! Please don’t view the commentary as personal or disqualifying. I been voting nearly 90 years, and someone challenged my reasoning this year. What is actually happening here is that sometimes a ballot prompts a return to a discussion topic of ongoing interest. While voters are entirely independent, this entire project is an exercise in Long-term consensus building. Evidence-based persuasion is part of our culture. An important part says me because often it influences me in to see something more deeply than before. As a leading proponent of avoiding McCormick who believes we have overdone it in the 19th century I am totally on board with those challenging your perspective. And I’m on board with you voting you. As long as Jack Morris isn’t in your ballot. (KIDDING!!!!) But there’s plenty good to learn too.
   320. DL from MN Posted: December 11, 2017 at 08:44 AM (#5589908)
Yes, please vote. Everyone has a candidate that nobody else agrees with.
   321. Mike Webber Posted: December 11, 2017 at 12:20 PM (#5590129)
@315: Willis is actually on my ballot and on those of several other voters. If memory serves, I've also seen ballots here with Uhle, Joss, and Rixey. One might argue that not all time periods are equally populated with deserving pitchers.


I agree with that. And I think there's a good argument is not only is McCormick within the top 10, but within the top 5, and it seems that top 5 pitchers almost always HOM worthy.
   322. 6 - 4 - 3 Posted: December 11, 2017 at 01:32 PM (#5590201)
@318: Thanks, will vote. Just wanted to wait a few days for some feedback.

After reading the comments about 19th century pitchers, I'm re-evaluating my inclusion of Willis and will probably replace him with Sal Bando (actually he was an oversight on my initial ballot [32.5 WAA]).

Question for Kiko: how (or does) your methodology adjust for military service? That is, do you make an adjustment for a player who puts up one of his best seasons during a year when many were serving in WWII (e.g., Tommy Bridges' 1943 season)?
   323. Rob_Wood Posted: December 11, 2017 at 02:20 PM (#5590255)
I too want to echo the sentiments expressed above and encourage people to vote.

Debating/discussing players and issues is part of the fun of this exercise (it goes without saying).

If sometimes the debate gets a little too heated or appears to become personal, please don't take offense.
   324. Kiko Sakata Posted: December 11, 2017 at 03:41 PM (#5590357)
Question for Kiko: how (or does) your methodology adjust for military service? That is, do you make an adjustment for a player who puts up one of his best seasons during a year when many were serving in WWII (e.g., Tommy Bridges' 1943 season)?


The raw numbers on my website make no adjustment for league strength and player totals are totals for the games which the player played in (white) major-league baseball (for which Retrosheet has released play-by-play data). A win in 1943 is treated the same as a win in 1933 or 2013.

i.e., You have to make whatever adjustments you want to make for World War II manually yourself. I have ideas of trying to modify my "uber-stat" page to let people weight seasons/leagues individually, but (a) that's really complicated, and (b) that wouldn't solve the "how much credit should we give Tommy Bridges for 1944 when he was in the U.S. Army instead of MLB?" question anyway. So, for now, you have to go in and manually add whatever adjustments you think are appropriate.

One thing I would point out re: Bridges - or anyone from his generation. For seasons prior to 1941, Retrosheet is missing some games (prior to 1925, they're missing most entire seasons, which just won't show up at all). The default numbers on, say, Tommy Bridges' player page only include games for which Retrosheet has released play-by-play data. If you click the "Player records with missing games extrapolated" link, for seasons for which we have some of Bridges' games, but not all of them, I blow up the data that I do have based on his total games played. This assumes he performed the same in the games I'm missing as in the games I have, which will never be literally true, but hopefully it gives some sense of a player's performance. If you do this for Bridges, you get his 1934 - 36 seasons (where we're missing 51 games for Bridges - which is quite a lot, actually) better than his 1943 season (where we have full play-by-play for the entire season).
   325. Bleed the Freak Posted: December 11, 2017 at 10:31 PM (#5590528)
Responses to ronw ballot post #19, but some general points of discussion I think are important, DRA arm values, Andruw Jones defensive evaluation, Buddy Bell's defensive and replacement level assessments, and Sammy Sosa's situational hitting:

11. RF: Bobby Bonds (61.4, 34.5, 23.1) - #16 all-time RF. Just below Reggie Smith. Frankly, he pencils as a better player than partial contemporary Dwight Evans, who sailed in easily.


Bonds is defintely underrated, but careful to vault him ahead of Evans with a mix of bbref/gauge.
For all outfielders, Doc has mentioned previously that the arm ratings are of no value, we can use the baseball-reference R-OF as a proxy for value, for Dwight, that adds a massive ~5 wins to his ledger, Bobby ~1.5.
Kiko's data also shows Evans trouncing Bonds, FWIW.

9. CF: Andruw Jones (63.6, 38.3, 24.9) - #10 all-time CF. We have passed on all other glove-first CF but Jones is off the charts defensively, easily the best all-time outfielder. Less of a bat than I thought when numbers are placed into context, but enough of one to elect him. The fielding rep puts him above Lofton, but not by much.


General question for the electorate, do we believe that Andruw's glove is head and shoulders above anyone in history?
Is it possible he's merely in the discussion for best ever, and if that's the case, he's more of a bubble candidate for the HOM.

10. 3B: Buddy Bell (66.4, 33.6, 22.4) - #15 all-time 3B. Another glove-first player. Much of the backlog are players whose WAR/WAA/WAG numbers are enhanced by fielding. Bell would be the worst hitter we have elected as a modern 3B, but he wasn't a liability at the bat. In fact, he is probably a later version of Jimmy Collins.

Bell falls shy with Dan R's WAR, as his replacement level for 3B analysis shows B-R/B-G allotting credit that might be better classified with the shortstops of that era. Bell's defense is at ~20 wins in Baseball Gauge, ~19 with Baseball Reference, and ~14 with Kiko's. Also a negative in context of his teams, per Kiko's eWins analysis.

required comment RF: Vladimir Guerrero (57.0, 29.7, 17.9) - #20 all-time RF. Vlad wouldn't be a horrible choice, but I think he is clearly behind Bonds and Sosa on all metrics.

In a context neutral sense, Sosa is a clear winner ahead of Vlad, but Sosa was a terrible situational hitter.
His clutch score at Baseball-Reference is a record breaking negative 17 wins, while Vlad is just a mediocre -2.
Sosa was a real detriment to his team in clutch scenarios, how much should we dock him because of this?


Chris Fluit's ballot in post 18:
15. Luis Aparicio, SS (15): +123 base-running (including reaching base w/o a hit) and +149 fielding. Does everything that doesn’t show up in OPS and WAR notices- his 49.9 beats Bert (45.3) and crushes Concepcion (33.6).

Figures are only Baseball-Reference WAR, Bert and Davey kill Aparicio in Kiko's W-L and Dan R's WAR.

Ardo's post 13:

5. Adolfo Luque (6-7-5) - It's uncertain whether his major league career was held back by racism (as I once believed unequivocally, even to the point of ranking him #1) or he was simply a late bloomer. His record in Cuban play is excellent, though a notch below that of Jose Mendez.


How much credit should Luque get for Cuban league/integration credit?
His case is already pretty strong, I easily understand where Ardo is coming from throwing him a bone.


Patrick W post 11:
9. Luis Gonzalez (9), Ariz. – Hou. (N), LF (’91-’07) – A career candidate with a tent pole 2001 season that elevates all his peak scores. I don’t recall thinking of the ’91 Astros as a great collection of talent when I saw them in person at Wrigley, but quite a few of ‘em made careers for themselves, no?

12. Frank Tanana (12), Cal. – Detr. (A) SP (’73-’93) (2000) – No longer seen as having a Koufax peak, but it is still one of the top 175 peaks of all time. Plus the ever- present 10 additional years of average / below avg.


You are a Baseball Prospectus proponent, no?

What makes Gonzo and Tanana stand out here compared to middling rankings elsewhere?
   326. Howie Menckel Posted: December 11, 2017 at 10:34 PM (#5590532)
some food for thought for Vlad fans who won't even consider McGriff on the end ballot (spoiler alert, I'll mention this again in my ballot)

OPS+ in full seasons, best to worst

Vladimir G: 162 160 157 154 150 147 146 139 138 130 119
CrimeDog: 165 165 157 157 153 147 144 144 125 120 119 111 110

doesn't make someone want to put McGriff ahead, given other factors in play - but should he be 50 spots behind? feedback welcomed of course
   327. Bleed the Freak Posted: December 11, 2017 at 11:05 PM (#5590544)
Continuing the discussion:

Tanana is smoking awesome with DRA pitching value at BP, but the guys at Tom Tango's blog have some legitimate arguments against the methods:
http://tangotiger.com/index.php/site/comments/new-dra

And just curious on Luis Gonzalez, I was a big fan growing up, but I can't quite squint to get him ballot worthy.

Bachslunch #10:

4. Jim McCormick. Best WAR for starters not in by a mile. Short career, but played in NL except for one UA season.


If we want to elect one more old timey pitcher, I suggest taking a look at Tommy Bond.
Bond averages a 65 WAR on a 50/50 split for B-R and B-G, all of it concentrated into a 5 year span.
He has a Koufaxian/Santanaian argument to his resume.

Jaack #9:
Probably the biggest difference bewteen me an most voters if I make heavy use of FIP-WAR as my primary pitching metric, and at the current time I do not use bWAR at all for pitchers.
6. Mickey Lolich – I’m definitely his best friend, but I really do think he’s a great choice. He pitched a ton of high quality innings and was excellent in the postseason. I think of him as the player Murray Chass pretends that Jack Morris was.

14. Hack Wilson – I’ve overrated him in the past. He’s not a slam dunk candidate, but I do think his 1926-30 stretch is a HoM worthy peak.


First off, welcome to the show, you are quite thoughtful in your discussions here and greatly appreciated on my end.
What is your basis for selecting FIP based pitching ahead of RA9, or the methods by Kiko's W-L records?
FIP is the preferred short term method of valuing pitcher, but at a certain point, I was under the impression that RA9 held much greater predictive value.
At least that's what the glossary at Fangraphs claims: https://www.fangraphs.com/library/pitching/fdp/ -
"FIP and other DIPS metrics are generally better at predicting things in smaller samples, while in the long-run RA9 tends to be the best."
"When evaluating knuckleball pitchers, it’s better to use RA9-Wins, as those pitchers have been shown to consistently generate weaker contact than typical pitchers."

Hack Wilson has a possibly solid resume with Baseball-Reference, but he's woefully short with Baseball Gauge's measures...are u basing the ranking primarily on B-R with a placement on his strong peak/prime years?

DL from MN #3:

4) Tommy Bridges - have been a supporter since 1970. He's a required disclosure now


Apologies if I've asked previously, but Kiko, can you take a look at Bridges, he's pretty intriguing by a mix of B-R and B-G, but my interpretation of his win-loss records is a guy well short.
   328. Rob_Wood Posted: December 12, 2017 at 12:03 AM (#5590554)
Very good discussion. Thanks to all.

For whatever reason, this is one of my most difficult ballots to craft. I hope I don't ruffle anybody's feathers when I say it is probably because the quality and depth of this ballot is fairly weak (excepting the top three first-timers).

Perhaps another way of saying the same thing is that nearly every other player has "conflicting" information in his resume. Often the various uber-stat methodologies lead to different evaluations. Some key players missed time during a war, but also two of those players (who have been on my ballot for many years) undoubtedly benefited from weaker wartime competition and had great and valuable seasons surely requiring a bit of a discount.

I have put together three full ballots to date and have been deeply unhappy with each of them. I will try another attempt tomorrow and may post my ballot in the ballot thread, but if I am still torn I may post it as a preliminary ballot herein, acknowledging that there wouldn't be a lot of time for feedback before the ballot deadline.
   329. Kiko Sakata Posted: December 12, 2017 at 01:16 AM (#5590572)
Kiko, can you take a look at Bridges, he's pretty intriguing by a mix of B-R and B-G, but my interpretation of his win-loss records is a guy well short.


Retrosheet seems to be missing a disproportionately large amount of his career within seasons for which Retrosheet has released partial data. In general, Retrosheet tends to have 75-80% of games for the incomplete seasons that it's released. But for Bridges, I only have play-by-play data for 264 of his 424 games and that's having complete coverage of his final five seasons (86 games). So, I'm missing 47% of his games prior to 1941. Given that, my concern is that the data I do have are painting an inaccurate picture of Bridges. I don't KNOW that's true, but it's a concern (e.g., it could be that Retrosheet is disproportionately missing Tigers' home games, which could make Tigers players look worse). Because taking my results at face value, yeah, he's nothing special. For example, contemporaries Mel Harder and Dutch Leonard look better than Bridges by my data. Then again, Fangraphs actually shows a higher career fWAR (based on FIP) for both Harder and Leonard, so maybe there's something there.

I hate this as an answer, but I'm really not sure what to make of Bridges. I trust my numbers enough that he'll be off-ballot for me, but I don't have enough confidence as to whether he should be 20th in my consideration set or 100th in my consideration set.

I won't have time to do a deeper dig before balloting ends. But I hope to do so at some point this spring.
   330. Jaack Posted: December 12, 2017 at 01:25 AM (#5590575)
@327

As far as my choice of FIP goes, I'll preface that FIP only makes up about 50% of my pitcher evaluating metric (RA9, BPro's WARP, and Kiko's W-L records are all contributing factors). But I use FIP, at least in part, because I think it's a noteworthy metric that hasn't received too much attention from other voters. As such, I think it can illuminate worthy candidates that are treated poorly by RA9 or similar metrics (bWAR).

I'll use Lolich as an example. RA9 WAR sees him as a good pitcher, but to inconsistent to be a candidate. Two great years in 1971 and 72, but only a few scattered all-star seasons around that. FIP, on the other hand, sees him as a consistent all star. 1970 by RA9 as a down year, but when you dog deeper, you can likely blame the spike in runs on Cesar Gutierrez and a fast fading Don Wert manning the left side of the infield. Once they're gone, Lolich's BABIP goes back into normal range and he has a couple of his best seasons. FIP also recognizes Lolich's decline as actually being pretty solid, and his average appearing record was, in part, due to an aging Detroit defense. Combined with his huge inning totals, even for the time period, and excellent post season record, makes Lolich a fine candidate, who's flaws are exaggerated by RA9.

Of course I could contrast that with Javier Vazquez, Paul Derringer, or Jerry Koosman, who all repeatedly underperformed their FIP, and thus rank more poorly after my adjustments (although I am still intrigued by Derringer, who even after my adjustments looks similar to Tommy Bridges to me).

In the long runs, FIP tends to agree with RA9 more often for not - as it should. It's where they differ I want to take a close look. And starting with FIP helps me do that better.

As far as Hack Wilson goes - I love his bat, and I definitely lean more on the preak/prime side for hitters especially. His bat makes him a serious candidate for me, which means his case rests upon his defense. By total zone (bbref) he was an adequate center fielder, which would put him pretty even with Goose Goslin for me. By DRA he was an awful center fielder, which pushes him more towards Wally Berger range. With the mix I use between the two that lands him close to the borderline mixed in with other flawed outfielders. I'm not in love with him as a candidate but he ranks in my top 15 for now and I think he's worth diving into further.
   331. Rob_Wood Posted: December 12, 2017 at 01:32 AM (#5590577)
Somewhat akin to Kiko's system, my much more modest Win Values system to evaluate starting pitchers relies upon detailed Retrosheet data. So I have just in the past week or so worked up Tommy Bridges from 1941-1943 (recently released seasons).

I find that Bridges's Win Value figures are as follows (Win Values are measured relative to league average so are best thought of as parallel to WAA):

1941: 2.1 Win Value (BB-Ref WAA is 2.5)
1942: 1.9 Win Value (BB-Ref WAA is 2.0)
1943: 2.9 Win Value (BB-Ref WAA is 3.7)

Now, it is not uncommon for a pitcher's Win Value figures to be slightly less than the traditional WAA figures due to the "non-linear" relationship between wins and run prevention. But Bridges' Win Value for these three years being 1.3 wins less than WAA (6.9 vs 8.2) is concerning to me, a long-time voter and advocate for Bridges.

My current view is to include Bridges in a down-ballot position on my ballot. With more data being made available by Retrosheet in the future (next year?), I hope to revisit the case for Bridges with a fuller appreciation of his career at that time.
   332. Dr. Chaleeko Posted: December 12, 2017 at 07:56 AM (#5590600)
Rob Wood, you completely nailed it. There is conflicting info all over the place. Which I think contributes to this long and active thread. I’ve been making this point a lot, and your post crystallizes it. Given where we are in the backlog, we are all in the place where we are inside the margin of error for our own evaluation systems as well as for the numerous stats and uberstats we all depend on. At this point and into the foreseeable future, we will likely discuss less about any candidate’s mutually recognizable qualifications and how he compares to other players and more about the nuances of various systems and stats. Because that is where we are at. Is that a feature or a bug?

Also, and as if to magically rebut what I’m saying just now.... Howie, McGriff’s defense and baserunning appear to comprise the difference between him and Vlad.
   333. Bleed the Freak Posted: December 12, 2017 at 09:00 AM (#5590619)
Formatting is wonky, quoting Doc's post from 332:

...Rob Wood, you completely nailed it. There is conflicting info all over the place. Which I think contributes to this long and active thread. I’ve been making this point a lot, and your post crystallizes it. Given where we are in the backlog, we are all in the place where we are inside the margin of error for our own evaluation systems as well as for the numerous stats and uberstats we all depend on. At this point and into the foreseeable future, we will likely discuss less about any candidate’s mutually recognizable qualifications and how he compares to other players and more about the nuances of various systems and stats. Because that is where we are at. Is that a feature or a bug?...

Provided it generates quality discussion, I would say it's a feature.
We have 3 clear cut newbie candidates and a backlog of interesting, but flawed candidates.
Hopefully my questions weren't vindictive, just attempting to better understand people's methodologies to help me craft a better ballot.

Doc,
The work you have put in on the Negro League MLEs is tremendous, but I don't see anyone eligible in your top 15.
Do you have a current ordinal rank you can share, maybe a tiered approximation of where you stand?
When asking, I realize that the data will continue to improve, and might be considerably more complete by the 2019 election.
I've been a backer of Hilton Smith for quite sometime, but should I vault Ben Taylor ahead or anyone else?
I could easily have 2 Negro leaguers on ballot this year...thanks for the assistance :)
   334. Carl Goetz Posted: December 12, 2017 at 10:28 AM (#5590692)
I'd say that's a feature. None of these so called uberstats are perfect. Its important to know the strengths and weaknesses of the system you use (and that others are using) so you can catch the type of players that a system might overrate and those it might underrate. In short, arguing the systems is another way of pointing out a given player's hidden strengths or weaknesses. People questioning my methods (that led to my underrating Chipper) didn't lead me to say "F U all, my system says Chipper is 5th and that's that". It made me do a lot of research into the Offensive, defensive, and even pitching uberstat that I'm using. Chipper has already moved to first on my ballot as a result of my defensive changes alone. My main issue now is that I don't have time to implement my offensive and pitching changes before this year's election. I'm debating whether I want to be an observer this year or just post my ballot as is and recognize that it will be somewhat different for the holdovers next year.
Long story short; too late; your uberstat of choice is there to begin a discussion, not end it. And when a cloud of players are close on your ballot, its important to understand how your system may over or under rate certain players in that cloud. I am switching to BBRefWAR (RA/9) for pitching next year, but knowing about a guy like Lolich (for example) is going to make me take a long look at him and he'll possibly get a boost of some sort. I want people pointing out players that I should dig deeper into. I may not agree with you when I'm done, but I want to do the deeper dive and find out. A) I want my ballot to be as thorough as possible and B) Its fun to do.
Quick side question: Any reason Willie Davis' offensive WAR would be overrated by BaseRuns or underrated by wRaa? Was just looking at him (missed him before since he hasn't gotten a lot of ballot support) and he looks pretty good by BaseRuns, but in a cloud of OFs by wRaa. I feel like 6+ WAR over a career is significantly more variation between offensive metrics than I would have expected.
   335. DL from MN Posted: December 12, 2017 at 11:15 AM (#5590766)
Post your ballot as-is. I make changes all the time. Don't let "perfect" get in the way of "good enough". This isn't a math problem with one correct answer. Let the wisdom of crowds sort it out.
   336. Rob_Wood Posted: December 12, 2017 at 11:44 AM (#5590800)
Post your ballot as-is. I make changes all the time. Don't let "perfect" get in the way of "good enough". This isn't a math problem with one correct answer. Let the wisdom of crowds sort it out.


For sure post your ballot.

Maybe (note to self) be a little "conservative" this time if you think your evals are in the midst of changing.

As you know, players never lose eligibility so it's not like we won't have another chance to vote for someone. I guess the converse is that we don't want anybody to be voted in on "faulty" reasoning (not that I am accusing you or anyone else of that).
   337. Carl Goetz Posted: December 12, 2017 at 12:13 PM (#5590824)
That was my leaning, especially since I'm pretty comfortable with my top 5.
   338. Dr. Chaleeko Posted: December 12, 2017 at 01:35 PM (#5590883)
Why didn't I vote the MLEs? A few reasons.
1) I wasn't sure I wanted to vote for a first baseman (Ben Taylor)
2) I haven't run up all that many pitchers yet, so someone like Hilton Smith, who looks good not great isn't going to leap out at me until I have more guys under my belt.
3) I also believe that Andy Cooper is a little puffy due to the method I've used so far, and I'm trying to get as many guys out as I can. Some of the fine tuning may wait until I feel some urgency to take a closer look at him or others in a similar spot.
4) A most troublesome aspect of things for me is the defense of players who spent relatively little time in either the Negro Leagues or the majors. Guys with a lot of minor league time (Dandridge, Clarkson, Williams, Wilson) are really difficult to pin down defensively. I've taken a layman's shot at it so far, but I'm not at all confident in the approach I've taken. I've got a lot more research and figuring to do on that front.
5) I sure wasn't going to vote for Judy Johnson! :)

I am hoping that by the end of September 2018 I'll have MLEs for more than 100 total players, including guys that we didn't really get to before whose cases might look better now that we have higher quality data. At that time we'll be able to look at the question of the remaining Negro Leaguers a little more squarely, though my early work on the best of rest guys indicate that while there may be a surprise or two on the way, for the most part, we've done a fine job of locating the top candidates and giving them a good once over.
   339. Rob_Wood Posted: December 12, 2017 at 02:15 PM (#5590922)
Do you have any thoughts on Bus Clarkson who I am contemplating putting on my ballot this year? (I don't need a deep-dive, just a high-level summary.)

Thanks much.
   340. Dr. Chaleeko Posted: December 12, 2017 at 04:01 PM (#5591058)
Rob Wood,

Yeah, here's what I can tell you about Clarkson based on the work I've done so far.
His career is a little short
I'm estimating about 7550 PAs. World War II probably knocked 1500 to 1880 PAs off his career (1943-1945). It also ate up his peak years (ages 28-30).

But I've currently got him with 400, 360, and 310 PAs from ages 39-41.
This is based on those comps who made it that long in their careers. I originally comped him to good hitting shortstops with long careers and to some third basemen with long careers, all of whom are represented in 162 notation: Ripken, Boggs, Da Evans, Cey, Deacon White, Jeter, Appling, G Davis, Larkin, Cronin, Molitor, and Banks. The point wasn't to model their offense only their playing time. But Clarkson remained a positive offensive contributor at age 41. Now there's all kinds of problems with any set of comps, but five of these guys played through age 41, and I simply took their average playing time during those years. Could easily knock down age 40 if you wanted to (though all but three of the comps played through age 40) and 41 by half or more if you wanted to. So there's some squishiness there.

He was a good hitter, but not an offensive force
So far, I'm seeing a top season of a little over 30 Rbat but usually n the high teens or low twenties per annum. I've arrived at an average of about 21 Rbat per 600 PA. We are missing data from 1946 Negro Leagues, 1948 Provincial League, 1949 Negro Leagues, 1950 Negro Leagues, but we have something like 4000 PA of performance available to us. That's enough to have a pretty reasonable degree of confidence. Obviously that's a whale of a hitter for a shortstop, however.

He was probably a decent baserunner.
Bus stole about 60% more bases than his leagues did, adjusted for his teams' tendencies to run. This probably suggests an Rbaser score in the range of 25 to 30 runs. He trailed off badly in 1955 and 1956 at 40 and 41, but who isn't slower when they reach middle age?

His fielding is hard to gauge
There's not much data in the Negro Leagues Database on Clarkson's fielding. Just one season with about 130 innings total, mostly at third base. We have nothing in Mexico, and our first good look at his glove comes in 1950. He played all of 58 games at 3B, and didn't look very good. In 1951, he played 73 games at shortstop and was just a little better than he'd been in 1950. In 1952, he played 19 games at third and looked good, and he played 48 games at shortstop where he looked bad. In 128 games at third in 1953, he looks about average. In 1954 in 77 games at short he look awful again, and in 72 games at 3B, he looked a little below average. Finally, in 1955 at third base he looked awful. But before you trust me about his fielding, let me tell you how I arrived at this. First I used extrapolated runs since it's easy to locate online. I know it's not exactly supposed to be used for pre-1955, but anyone who was in the minors will be close to that time period.
1) I figure that every error below/above the league average at his position in his number of total chances is worth about +/-0.59 runs, half a run for a single and 9/10s for the loss of an out.
2) Every assist above/below is worth the same
3) Every DP turned is worth about 0.37 runs (same as XR values it).

Yeah, it's pretty much just a crappy range-factorish kind of thing. Put it all together, and you have value above/below average. I haven't adjusted for the handedness of his pitchers or anything like that at this juncture. Given all of the chronological information above, I suspect he was an average fielding shortstop who would have outgrown the position by age 30. He got back from the war at 31, so I moved him to 3B, where I figure he probably was above average for a few years, then average, then below. En toto, it works out to about +5 to +10 overall with wide error bars. You might be just as safe calling him a 0 fielder.

Fielding is the place where I have the least confidence in my work so far. I'm not sure if the little routine I outlined above is close enough or totally screwball. Glad for some help on that one!

Anyway, sum it all up, and I've got him around 60 BBREF WAR at this juncture. Doesn't look like a strong peak candidate, more like a guy with several All-Star seasons, several more four win seasons, and a couple shoulder seasons. More like a Lou Whitaker value pattern than a Ryne Sandberg you might say. But there's four seasons of hitting data missing (at least two of which I have little hope of seeing anytime in the near future), defense is a question, and there may be a little more PT at the end than some might agree with.

Does that help? Or just make it worse? ;)
   341. DL from MN Posted: December 12, 2017 at 04:33 PM (#5591113)
More like a Lou Whitaker value pattern than a Ryne Sandberg you might say


I see Barry Larkin with 15 fewer RField and 40 fewer RPos (average 3B instead of slightly above average SS).
   342. Rob_Wood Posted: December 12, 2017 at 04:40 PM (#5591120)
Okay, this supports my view that Clarkson is right around my ballot borderline. Maybe at bottom of my ballot. Maybe just off my ballot.

Thanks.
   343. Howie Menckel Posted: December 12, 2017 at 05:57 PM (#5591183)
my ballot will be up shortly; some comments reference this list of newcomers + Vlad and Crime Dog


OPS+ in full seasons, best to worst (there's a double Chipper, so you can pick if to choose those seasons)

JimThome: 197 170 167 157 156 155 154 153 150 144 142 132 131 127 124
ChipJones: 176 169 165 160 154* 153 151* 148 142 137 137 124* 121 119 117 116 108
ChipJones: 176 169 165 160 153 148 142 137 137 121 119 117 116 108
Vladimir G: 162 160 157 154 150 147 146 139 138 130 119
CrimeDog: 165 165 157 157 153 147 144 144 125 120 119 111 110

ScotRolen: 158 139 138 129 129 127 126 126 121 120* 116 109* 089*
AndrJones: 136 127 126 117 116 114 112 094 093* 087

(CJones had 477, 448, and 432 PA in the asterisked seasons)
(SRolen had 497, 467, and 441 PA in the asterisked seasons)
(AJones had 467 PA in the asterisked season)
   344. DL from MN Posted: December 12, 2017 at 07:08 PM (#5591222)
I asked in ballot thread but probably more appropriate here. Howie - how do you rate Johan Santana?
   345. Howie Menckel Posted: December 12, 2017 at 08:28 PM (#5591247)
re Santana, starting with best year:

2004 - 182 ERA+ (1st) with 265 K (1st) in 228 IP (2nd). unanimous Cy Young
2006 - 162 ERA+ (1st) with 245 K (1st) in 234 IP (1st). unanimous Cy Young
2008 - 166 ERA+ (2nd) with 206 K (2nd) in 234 IP (1st). 3rd in Cy Young voting
2005 - 155 ERA+ (1st) with 238 K (1st) in 232 IP (2nd). 3rd in Cy Young voting
2007 - 129 ERA+ (10th) with 235 K (2nd) in 219 IP (6th). got a 3rd place vote

2010 - 131 ERA+ with 144 K in 199 IP
2009 - 130 ERA+ with 146 K in 167 IP

2003 - 148 ERA+ with 169 K in 158 IP
2002 - 150 ERA+ with 137 K in 108 IP

deserved at least a mention on my ballot. four Koufax-lite years (fewer IP and without the late-season/World Series clutch levels, either). some other solid years, too.

I wasn't as high on Koufax as others, plus the ballot was weaker then - so of course Koufax got in immediately. plus I am less a peak voter than some. I would say that those bullish on Koufax, without a pet P or two on the board, and concerned with HOM P representation could do a lot worse than giving Santana a high rank.

I'll definitely give Santana full consideration next year. Unlike the HOF, he doesn't have to worry about falling off this ballot. kind of a Hughie Jennings case (the big 4 years).
   346. Kiko Sakata Posted: December 12, 2017 at 10:02 PM (#5591299)
Okay, here's my preliminary ballot - a day later than I promised earlier. I posted a (very) preliminary ballot with discussion back on the first page of this thread, which I won't bother to repeat. One caveat: in addition to new data, I also tweaked my Player won-lost records a little. I had a problem with my treatment of double plays that led to some reshuffling vis-a-vis what's on page 1.

For players for whom I have calculated Player won-lost records (MLB since 1925), here's my objective starting point. I try to balance career vs. peak vs. prime with some position-specific adjustments. These are all generally discussed here. I then mix in pre-1925 players in a somewhat more subjective (but still hopefully systematic) way.

1. Chipper Jones - best player eligible and it's not particularly close
2. Jim Thome - second-best eligible player and it's not particularly close. My system values fielding less than WAR (correctly so, I believe)
3. Tommy John - if you've read anything that I've written at the Hall of Merit in the last three years or however long I've been doing this, you'll know that my system LOVES Tommy John and that I believe that he is the most egregious omission of the Hall of Merit electorate to date.

My system has clear #1, #2, and #3 candidates this year. Below this is where the ballot becomes much harder to distinguish the candidates.

4. Vern Stephens - with the release of 1941 data, Retrosheet has data for his entire career. My system loves power hitters at fielding-first positions who can hold their own defensively (Stephens was good enough at SS that the Red Sox moved Pesky to 3B when they acquired him). The big question for Stephens is how much to discount his 1944-45 peak. I have discounted these seasons somewhat in the first link here (see page 1 of this thread for how much). In defense of Stephens, I would point out that, per Player won-lost records, his best season was actually 1949.

5. Jorge Posada - I'm EXTREMELY skeptical of pitch framing data - at a minimum, the numbers I've seen are way, way, way too big to pass the smell test. A catcher with an OPS+ of 121, wRC+ of 123. Yes, please.

6. Wally Schang - he almost entirely predates my system. Statistically, he seems very similar to Posada (OPS+ 117, wRC+ 120, bWAR/fWAR of 45.0/41.0 vs. 42.7/44.7 for Posada). Schang slots below Posada because I'm more uncertain of his numbers, plus he played in a segregated league. On the other hand, Schang has a much stronger claim for being the best catcher in (white) baseball during his career than Posada does (Piazza was, of course, much better than Posada; Ivan Rodriguez was for much of his career, too).

Players 6 - 9 in the first link here basically share the same case: excellent peak as perhaps the best pitcher in baseball, but not quite enough career bulk to be a slam-dunk HOF/HOMer. Objectively, they rank Dizzy Dean, Orel Hershiser, Dwight Gooden, and Johan Santana, in that order. I've rearranged them a bit here.

7. Orel Hershiser
8. Dwight Gooden - career length separates these two from the two below

9. Johan Santana - longest/best peak of the four, but a much shorter career than Hershiser and Gooden (800 fewer innings than Gooden; 1,000 fewer than Hershiser)

10. Dizzy Dean - I've judgmentally bumped him down a tick because some of his peak is extrapolated here (Retrosheet is missing play-by-play data for about one-third of his career). He also has the weakest case of these four for ever being the best pitcher in his league because of the presence of Carl Hubbell.

11. Scott Rolen - I've bumped him up from my objective list in a bit of a nod to consensus. Don't be fooled by the low ranking. He's definitely a deserving Hall-of-Meriter.

12. Jeff Kent - see my comment on Stephens.

13. Ben Taylor - I don't have a good feel for him at all, but this feels rightish.

14. Gil Hodges - I think he benefits from positional averages being calculated empirically every year. He really was the best first baseman in baseball in the 1950's (when Musial wasn't playing there).

15. Luis Tiant - my system has warmed up to him. Close battle for the last ballot slot; Tiant gets it in part in a nod to consensus.

First seven off-ballot (I had trouble narrowing it to five): Dave Concepcion, Toby Harrah, Dale Murphy, Vic Willis, Urban Shocker, Tommy Henrich, Cannonball Dick Redding

Concepcion benefits from playing in an era when most shortstops couldn't hit worth a damn.
Harrah is in the mold of Stephens and Kent.
Murphy had a nice peak and benefits in my system from my zeroing out negatives.
Willis entirely pre-dates my system but seems like the sort of pitcher that my system would like.
Urban Shocker mostly pre-dates my system, but what I have for him (1922, 1925-28) looks very good.

Tommy Henrich requires three years of WWII credit to get here. He could actually rate higher depending on how generous you are with your WWII credit. He was much better in 1941 and 1947 than in 1942 and 1946. If you use the former two in building your WWII credit, I think you could make a case to slip him onto the end of the ballot. Thanks to Bleed for pointing out to me how much my system likes him (and how big an impact the WWII credit makes in his case - three full seasons would increase his career length by 25-30%.

Redding looks like the best available Negro League pitcher left and I tend to think the HOM has inducted too few pitchers. But I don't feel confident enough in his specific case to pull the trigger here. I hope to have some time in the coming year to do some more in-depth analysis of the top remaining Negro League candidates.

Another guy who my system likes that hasn't gotten a lot of love here is Kiki Cuyler. He's probably somewhere around #30 or so in my consideration set.

Required disclosures will be in the next message.
   347. Kiko Sakata Posted: December 12, 2017 at 10:17 PM (#5591311)
Required disclosures:

Sammy Sosa, Vlad Guerrero, and Indian Bob Johnson rate similarly, falling generally in my top 50. All three would be reasonable HOF/HOMers. They mostly rate behind guys who my system thinks the HOM has missed on - as well as a very strong freshman class (4 on my ballot).

Bobby Bonds is a tick below those guys because my system doesn't think as much of his fielding (~ -0.5 win for his career) as BB-Ref (+48 fielding runs).

Kenny Lofton is probably around 80th or so in my system. As I said above, WAR over-rates fielding.

Same story for Buddy Bell, who rates even lower in my system. In the first link in my previous message Bell shows up at #161, very close to Omar Vizquel. I see Bell's case as similar: very good defense, but not nearly as good as the all-time best (Ozzie and Brooks). Bell was a better hitter than Vizquel but at a position where it was expected and his lack of power hurts him. My system thinks that the Indians clearly won the Bell-for-Harrah deal after the 1978 season. Take a look at the two teams' records before and after that trade.

Tommy Bridges looks worse than several contemporaries in my system. I've discussed this a bit earlier in the thread.

Phil Rizzuto. Being as generous as I can plausibly be with World War II credit, I can maybe get him up as high as #40 or so in my system. Same general issue as Lofton and Bell.

Sal Bando ranks around top 70 or so in my system. Kind of a short career without enough career bulk to make up for it. Also same issue as Bobby Bonds - my system isn't in love with his defense (same as Bonds, -0.5 win for his career vs. +36 fielding runs on BB-Ref).

I think that's everybody.
   348. Bleed the Freak Posted: December 12, 2017 at 10:44 PM (#5591344)
Have been a lurker since the early days and a voter since the elections went annual, hope the turnout improves in 2018 versus 2016 and 2017. My system incorporates for hitters: Kiko's Win-Loss Records, DRA/Baseball Gauge WAR, Baseball-Reference WAR, Dan R WAR, War Credit, MLE credit (mostly for pre-integration players), Negro League/integration credit, CSAA/catcher values from BP where available, and RE24/clutch contextual value. For pitchers, I use the same systems, as well as Baseball Prospectus Deserved Run Average (although just a little), FG FIP WAR, and a dose of WPA.


PRELIM: Top 3 should be elected this year, 4th spot up for grabs! Please post to ballot thread if I'm unable to login/make updates prior to deadline.
1 Chipper Jones - easy #1 upper level HOMer.

2 Jim Thome - Kiko's narrower defensive spectrum tips the scales to Thome ahead of Rolen...great to meet in person, forever grateful for his work in his hometown community, was blessed to play on a field he donated, great to listen to on the MLB network.

3 Scott Rolen - won't earn much more than a sniff by BBWAA like Jim Edmonds last year, easily clears the backlog/mid-level HOMer. Part of a great 2004 Cardinals team that ran into the destiny Red Sox, made up for by a mediocre 2006 squad topping the superior Tigers.

4 Urban Shocker - strong marks besides FIP WAR, deserving of WWI credit, too bad he passed away SOOOOOO young.

5 Bobby Veach - please read the good doctors comments on him. Was a stud for 3 seasons in the AA ages 38-40, in today's game, does he get a chance to pad some stats at the end of his career? A Dale Murphy level hitter with plus defense.

6 Bert Campaneris - a stud in Dan R WAR, worthy by Kiko's and Baseball Gauge WAR, massive value in RE24 offense.

7 Art Fletcher - quality hitter for a SS, DRA D otherworldly, Baseball Reference merely excellent.

8 Tommy Bond - super peak 1870s, could be at top of backlog or outside of someone's consideration set, putting him here.

9 Johan Santana - or Johan Santa as linked at Fangraphs frequently :); compelling arguments that he's REALLY close to Sandy Koufax by regular season value, looked to be bouncing back from injury with a no-hitter in his final season, but 134 pitches/stress a contributor to his early demise.

10 Bob Johnson - quite stronger performer in all metrics I review, dock some for WWII, but bump some for PCL time.

11 Joe Tinker - see Art Fletcher.

12 Andruw Jones - Knocking his defense to an upper elite but not best ever gets him this ballot spot, will be interesting if he's one and done here (elected) and in the BBWAA (ballot dropped).

13 Wally Schang - warming to the non-Dan R WAR and Baseball Reference proponents, want a catcher on ballot, passing on Munson this year.

14 Don Newcombe – Howie gives him a shout as of interest, lack of 50s players in a low-standard deviation era, needs all bonus value we can find (Integration, Korea).

15 Orel Hershiser - overlooked mid to late 80s peak, historic post-season run, feels like we are light on players from his peak time frame.

16 Ben Taylor/Hilton Smith - contemplating whether to push one of the Negro League guys to ballot.


Guys who are close you could argue for at least a bottom half of the ballot spot:
Bobby Abreu, Babe Adams, Dave Bancroft, Bobby Bonds, Perucho Cepeda, Kiki Cuyler, Brian Giles, Dwight Gooden, Vladimir Guerrero, Gil Hodges, Harry Hooper, Jeff Kent, Tommy Leach, Kenny Lofton, Thurman Munson, Johnny Pesky, Sam Rice, Phil Rizzuto, Sammy Sosa, Vern Stephens, Vic Willis.

Required Disclosures (top 10 returnees) outside of my preferred consideration set:
Vladimir Guerrero – HOM worthy, in a pile with Bobby Abreu, Brian Giles, and Kenny Lofton modern era outfield candidates.

Sammy Sosa - dependent on defensive estimations on how much emphasis you place on clutch/late game situations, he was godawful, bumping him this year.

Jeff Kent - borderline or worse besides Kiko's pretty strong placement, leaving off this year.

Luis Tiant - HOM level and real close, killer Baseball-Reference and Baseball Gauge, enough WPA, just shy by Kiko, well short by Fangraphs and Baseball Prospectus.

Vic Willis – Top 60 hurler with Baseball-Reference, 70 with Baseball Gauge, 90 with Fangraphs, endorsed by Kiko...very close.

Kenny Lofton - If you don't buy TZ defensive valuations, he's a bubble or shy candidate, I'm ok with him as a personal HOM, but shy of ballot.

Bobby Bonds - Baseball Prospectus, Baseball Gauge, and Baseball Reference show him in, Dan R lukewarm, Kiko after the adjustment to right fielders shows Bobby short.

Buddy Bell - buy in for Kiko's system, slam dunk by Baseball-Reference, Baseball Gauge, and Baseball Prospectus, borderline with Dan R WAR.

Tommy Bridges - after PCL and WWII credit, squeezes in by Baseball Gauge and Fangraphs, short by Baseball-Reference, well short with Kiko.


Others:
Jorge Posada - looks like a good choice before incorporating Baseball Prospectus defensive metrics, which show him as truly dreadful. Splitting the difference would still keep him off ballot. Terrible post-season WPA and RE24 clutch metrics.

Excited for the Negro Leaguers in next years election, better data = more confidence in our placements.


None of the other newbies high in consideration set:
Omar Vizquel comparable to Mark Belanger, BBWAA lunacy.

Jamie Moyer to Jerry Koosman, great to see Moyer succeed after hitting the minors well into his career and throwing <70 mph breaking balls.

Carlos Zambrano a victim of the Dusty Baker arm shredding machine, on a HOF path...same might be argued to some extent for Kerry Wood and Chris Carpenter.

Kerry Wood - thanks for your graciousness in person, the 20k game as a rookie, and for pitching your ass off with the In Dusty "We Don't" Trusty times.

Johnny Damon - thanks for helping keep females entertained and the 2004 series run.

Hideki Matsui - a Japan monster, a more solid MLB career and we would need to take him seriously as a candidate.


Too difficult to place into the upper backlog:
Sal Maglie and Orlando Hernandez - what credit do these guys get for Mexican/Cuban leagues? The value they provided at the ages playing in MLB suggest lower tier HOM level or very close.
   349. Bleed the Freak Posted: December 12, 2017 at 10:57 PM (#5591360)
Kiko - 347 - where did you come out on Andruw Jones?
   350. Kiko Sakata Posted: December 12, 2017 at 11:06 PM (#5591368)
Kiko - 347 - where did you come out on Andruw Jones?


#27 for players for whom I have data. So, he'd be somewhere in the #30 - #35 range probably if my ballot went that deep. Similar placement to Kiki Cuyler probably; a step above the Vlad, Sammy, Indian Bob trio. He'd probably make my pHOM eventually, although he's #5 among the debut candidates this year, so he'd have to wait until at least next year. One of the best defensive CF of all-time, but (a) merely "one of", and (b) as I said probably a dozen times already tonight, fielding isn't as important as WAR thinks it is. That said, he was also a damn fine hitter in his prime.
   351. Rob_Wood Posted: December 12, 2017 at 11:14 PM (#5591377)
Here's my prelim ballot (sans explanations). Like everybody else I am incorporating a myriad of information. For pitchers I consult my own Win Values stat. Generally speaking compared to my previous ballots I am moving towards the "peak" side of the peak vs career argument based upon the new-fangled CPASR stat some nut recently developed.

1. Chipper Jones
2. Jim Thome
3. Scott Rolen
4. Kenny Lofton
5. Luis Tiant

6. Ben Taylor
7. Tommy Bridges
8. Vlad Guerrero
9. Johan Santana
10. Sammy Sosa

11. Jeff Kent
12. Buddy Bell
13. Urban Shocker
14. Andruw Jones
15. Sal Bando

Others I seriously considered voting for (in no particular order):

Bob Johnson, Bobby Bonds, Bus Clarkson, Bucky Walters, Phil Rizzuto.
   352. Carl Goetz Posted: December 13, 2017 at 10:32 AM (#5591564)
"(b) as I said probably a dozen times already tonight, fielding isn't as important as WAR thinks it is."

I guess I'd be curious as to how important you think defense is.

FIP certainly has flaws, but I do buy that pitchers have a lot less impact on BIP than his fielders do. According to BBref, in 2017, there were 120,524 BIP and 185,295 BF in the MLB or 65% of Batter Faced put the ball into play and were thus affected by the fielders to varying degrees. If we give the pitcher credit for .7 outs per BIP (BABIP average around .300), then that implies a pitcher/fielder split of around 80.5/19.5 and that is setting fielder replacement level at average (Average BABIP assumes average fielders). The Tigers were MLB worst by allowing .320 BABIP. Using that as a proxy for Replacement level defense, we'd give pitchers .68 outs per BIP and change the split to 79/21.

BBRef uses a Position Player to pitcher split of 59/41 which I will assume implies 50% Offense, 41% Pitching, 9% Fielding. This translates to an 82/18 split between Pitching and fielding on the defensive side of the ball. This seems about right to me, if not slightly conservative towards fielders.



   353. Kiko Sakata Posted: December 13, 2017 at 10:54 AM (#5591595)
I guess I'd be curious as to how important you think defense is.

BBRef uses a Position Player to pitcher split of 59/41 which I will assume implies 50% Offense, 41% Pitching, 9% Fielding. This translates to an 82/18 split between Pitching and fielding on the defensive side of the ball. This seems about right to me, if not slightly conservative towards fielders.


I'm not sure what you mean that "BBRef uses a Position Player to pitcher split of 59/41". BB-Ref keys everything off of average (0). The issue is the spread around zero, which is too large for them (and Fangraphs - it's actually more clear in Fangraphs; the issue is somewhat clouded in BB-Ref because they key defensive WAR off of actual runs allowed, which is right by construction at the team level).

This article compares my stat to WAR and gets into the fielding issue. This article was published in the Baseball Research Journal about a year ago and is a chapter in my book. I compare my fielding numbers to UZR here.

Mitchel Lichtman (MGL, creator of UZR) wrote about the topic a year-and-a-half ago here.

Tom Tango has come to a similar conclusion in looking at Statcast data here and here.

The range of fielding numbers at Fangraphs (UZR) and BB-Ref (Total Zone, DRS) is too wide.
   354. Carl Goetz Posted: December 13, 2017 at 10:57 AM (#5591599)
"I'm not sure what you mean that "BBRef uses a Position Player to pitcher split of 59/41"."
They set the league to a total of 1000 WAR, 590 to position players and 410 to Pitchers.

I'll take some time to read the many links you've provided. I may not have a response today.
   355. Kiko Sakata Posted: December 13, 2017 at 11:38 AM (#5591657)
They set the league to a total of 1000 WAR, 590 to position players and 410 to Pitchers.


WAR is built up from WAA. For pitchers, BB-Ref starts from actual runs allowed and, among various adjustments, adjusts pitchers up or down based on the fielding numbers of the defenses behind them. But those defensive adjustments average zero by construction. So, total WAR is unaffected by overall league-wide fielding, regardless of how wide the spread is - you'd get the same total league-wide pitcher WAR if you multiplied all of the defensive adjustments by 3 or divided them all by 7; you'd merely affect the spread of pitcher WARs.

Similarly, one can think of position player WAR as being constructed by calculating oWAR for everybody - i.e., based on batting, baserunning, and positional adjustments. By construction, total oWAR should equal total WAR for a league. Individual players' WAR are then adjusted up or down based on their fielding. But these adjustments, again, should average zero, so total position-player WAR - and, hence, the split between position-player and pitcher WAR - would be the same if you multiplied all of the defensive adjustments by 5 or divided them all by pi. All that would be affected would be how we value Andruw Jones and Scott Rolen vis-a-vis, say, Gary Sheffield and Adam Dunn, to pick some extreme fielders.

So, the split of position-player WAR vs. pitcher WAR doesn't tell us anything about what the spread of fielding numbers around zero is, or, hence, how reasonable that spread is.
   356. Carl Goetz Posted: December 13, 2017 at 12:13 PM (#5591727)
Quick question. So far I've only read your article comparing your stat to bWAR and fWAR. If I'm understanding your method correctly, you are taking each event in a game and measuring the impact on Win probability and then assigning that impact to the various players involved in the play. Therefore your system starts and ends with wins as primary measurement. Ignoring completely that your method is context-dependent and theirs are context-neutral, bWAR and fWAR start with components that are measured by their impact on runs scored and later converted. Aren't bWAR and fWAR at a slight disadvantage when comparing to actual team wins since runs scored/allowed do not perfectly correlate with actual wins. Wouldn't it be more appropriate to regress the run-based systems to Pythagorean Wins rather than actual wins to remove this portion of the error?

As for the actual topic of discussion here, should I take your conclusion to be that you would advocate an 8% reduction in Fielding WAA? If so, would you bump pitching WAA by a similar amount then?
   357. Carl Goetz Posted: December 13, 2017 at 12:18 PM (#5591734)
So total fielding WAR for a league is zero? Wouldn't that imply fielding is underweighted since its measured against average while offense and pitching are measured against replacement level?
   358. Kiko Sakata Posted: December 13, 2017 at 12:47 PM (#5591792)
So total fielding WAR for a league is zero? Wouldn't that imply fielding is underweighted since its measured against average while offense and pitching are measured against replacement level?


As I said above, WAR is built up from WAA. This is most explicit for position players on BB-Ref. In their WAR tables, all of the numbers for batting, baserunning, fielding, etc. are relative to average and add up to what they call RAA. RAA is then converted to RAR by adding a number this shifts the baseline from average to replacement.

In effect, the way they build oWAR/dWAR is that they assume that the entire difference between an average player and a replacement player is in their offense - i.e., that replacement-level players are average fielders. Which is probably close to true, but the truth of it isn't really necessary for what they're doing. It doesn't really make sense to think of "replacement level" as being a fixed offensive and defensive level. It's a fixed overall level: in some cases, your replacement will be a worse hitter but a better fielder; in other cases, the reverse; and, of course, in some cases, your replacement will be worse at both.
   359. Kiko Sakata Posted: December 13, 2017 at 01:01 PM (#5591817)
Therefore your system starts and ends with wins as primary measurement. Ignoring completely that your method is context-dependent and theirs are context-neutral, bWAR and fWAR start with components that are measured by their impact on runs scored and later converted. Aren't bWAR and fWAR at a slight disadvantage when comparing to actual team wins since runs scored/allowed do not perfectly correlate with actual wins.


First, as to the "sidebar", after constructing context-dependent wins (pWins), I then construct context-neutral wins, which are based on pWins but averaged across all contexts (e.g., all home runs hit at Wrigley Field in 2017 generated the same number of eWins, regardless of when they were hit). It is these eWins, then, which I compare to bWAR and fWAR in that article.

Second, "[a]ren't bWAR and fWAR at a slight disadvantage ... since runs scored/allowed do not perfectly correlate with actual wins." Yes, that's precisely the point of the article. WAR purports to measure wins; it's what the W stands for, after all. One can more accurately measure something when that is the thing being directly measured, rather than trying to impute a measurement of one thing from a direct measurement of a second, related thing. Hence, if you want to measure wins, you're better off starting by measuring wins, not runs.

As for the actual topic of discussion here, should I take your conclusion to be that you would advocate an 8% reduction in Fielding WAA? If so, would you bump pitching WAA by a similar amount then?


Obviously, my first preference would be to use my numbers. Alternately, if one wants to insist on using WAR, I would (a) advocate for ignoring BB-Ref's pitcher WAR entirely - I find their adjustments inscrutable and difficult to override. Better to use Fangraphs, look at both of their numbers (based on RA-9 and FIP) and mentally adjust as needed (simple average probably makes a plausible starting point; the longer the career and/or the more different teams pitched for, the more likely any gaps between RA-9 and FIP are due to actual pitcher skill, so RA-9 probably converges to a true-talent value measure over time). For position players, I would probably say that something like oWAR + (1/2)*(Fielding Runs) would be preferable to raw WAR. The (1/2) is just a ballpark estimate - and perhaps a bit too strong - but conceptually, I think that's the way to go.
   360. Carl Goetz Posted: December 13, 2017 at 01:02 PM (#5591818)
Ok, that makes sense, but if they set total MLB WAR to 1000 and pitchers make up 410 (expressly stated in explanation though its not clear as to whether that's forced or just the result of their methodology). If offense (runs scored) and defense (runs allowed) are each assumed to be 50/50 of total value, does that not imply 500 Offensive WAR, 410 Pitching WAR, and 90 Fielding WAR? In other words, why would position players get 59% of WAR if the extra 9% isn't due to their contributions on defense?
   361. Kiko Sakata Posted: December 13, 2017 at 01:11 PM (#5591838)
In other words, why would position players get 59% of WAR if the extra 9% isn't due to their contributions on defense?


I think that is the thinking: defense is split 82/18 pitching/fielding. I'm not entirely sure where they got that split and I think it is just imposed. I presume the "410" comes into play when they convert from WAA to WAR. The total difference across all pitchers between league-wide WAA (which should be zero, by construction) and WAR is 410 and that 410 is divided across pitchers proportionally or something like that.

I'll be honest: I don't really understand how either BB-Ref or Fangraphs get to pitcher WAR. I get what they're trying to do conceptually, but I don't get the math of all of the intermediate steps that get you from "This pitcher allowed 74 runs in 225 innings" to "This pitcher had a WAA of 2.8" to "This pitcher had a WAR of 4.5" (all numbers made up).
   362. Carl Goetz Posted: December 13, 2017 at 01:54 PM (#5591900)
"I'll be honest: I don't really understand how either BB-Ref or Fangraphs get to pitcher WAR."
Same page. I tend to agree with a roughly 80/20 split between pitching and fielding in terms of overall impact of runs scored in a game. The problem comes if there a wider range of pitcher quality than fielder quality. ie the best pitcher is more ahead of the replacement pitcher than the best fielder is better than the replacement fielder. This would give the best pitcher more value compared to the 80/20 split than the best fielder if that makes sense.
What would be interesting to me to see would be a more basic regression comparing RAA for pitchers and RAA for fielders with the team's runs allowed above average. Unfortunately, my level of statistical sophistication makes me well equipped to understand the results of an analysis, but not so much for running one myself :) That, and not sure I have the free time for that.
   363. Rob_Wood Posted: December 13, 2017 at 06:49 PM (#5592149)
Kiko or anyone else who may be familiar with the Retrosheet data:

How can I find out how many games (raw total or percent) are "missing" from the play-by-play data files of seasons prior to 1941? Is there a season-by-season tabulation available anywhere? If not, is there an easy way to figure this out myself?

I am not really looking for team-by-team tallies (though that would be ideal), but just a total for each league for each season. I am hoping there is an alternative to simply downloading the files and counting/determining how many games are missing.

Thanks much.
   364. Bleed the Freak Posted: December 13, 2017 at 07:14 PM (#5592155)
Rob in 363:

Retrosheet Missing Games Via Kiko's Website:
http://baseball.tomthress.com/Articles/Seasons.php
   365. Rob_Wood Posted: December 13, 2017 at 08:01 PM (#5592181)
Thanks Bleed (and Kiko). That is exactly what I was looking for!

For some reason I forgot, or never realized, that Kiko had that information on his website.

Thanks again.
   366. Bleed the Freak Posted: December 13, 2017 at 08:24 PM (#5592197)
6-4-3 ballot thread post #24:

1) C. Jones (53.2 WAA): Inner circle
2) Rolen (43.9 WAA): Not quite inner circle, but more than deserving
3) Lofton (38.2 WAA): Best leadoff man of his generation
4) Thome (37.5 WAA): One of the best pure power hitters in history
5) A. Jones (36.0 WAA): Best defensive CFer that I've ever seen in his prime
6) Tiant (34.5 WAA): One of the better pitchers not in the HOF/HOM

---PERSONAL DIVIDING HOM LINE---

7) Bando (32.5 WAA): Peak compensates for short career
8) Bell (32.4 WAA): Maybe one of these days, one of the better 3B not in the HOF
9) Santana (32.3 WAA): HOF peak, but falls short on career value
10) Bo. Bonds (31.8 WAA)): Ibid
11) Appier (30.7 WAA): Better pitcher than most give him credit for
12) Guerrero (29.4 WAA): Very steep decline for what appeared to be a surefire HOFer at the start of his career
13) Sosa (28.0 WAA): Obviously worthy peak, hurt by years of mediocrity on both ends of his career
14) Olerud (27.3 WAA): Underrated 1B
15) Kent (26.3 WAA): Relatively late start and the 1994-5 strike hurt him


Is this just Baseball-Reference WAR?
And I don't see a single player from pre 1960s era.

If you are going to run with a straight WAA results, I recommend at least splitting 50/50 with Baseball Reference and Baseball Gauge:
http://www.thebaseballgauge.com/history.php?tab=players_at&first=min&last=max&lgID=All&division=All&pos=All&age_low=16&age_high=59&bats=All&throws=All&HOF=All&active=All&living=All&results=500&page=1&sort=tot_a

01 - Chipper Jones - 43.9
02 - Scott Rolen - 41.9
03 - Joe Tinker - 40.7
04 - Art Fletcher - 36.2
05 - Jim Thome - 35.1
06 - Andruw Jones - 35.0
07 - Johan Santana - 33.5
08 - Tommy Bond - 32.9
09 - Bobby Bonds - 32.7
10 - Tony Mullane - 32.7 - no penalty for AA seasons, no credit for blacklisting
11 - Luis Tiant - 32.4
12 - Bob Johnson - 31.7
13 - Charlie Buffinton - 31.6
14 - Frank Chance - 31.4
15 - Urban Shocker - 30.9

16 - Buddy Bell - 30.6
17 - Kenny Lofton - 30.3
18 - Kevin Appier - 29.3
19 - Silver King - 28.4 - needs heavy discount for weak leagues
20 - Norm Cash - 28.3
21 - Sammy Sosa - 28.3
22 - Dizzy Dean - 27.9
23 - Vic Willis - 27.8
24 - Dizzy Trout - 27.8 - needs WWII discount
25 - Eddie Cicotte - 27.8
26 - Jim McCormick - 27.6
27 - Noodles Hahn - 27.6
28 - Tommy Leach - 27.5
29 - Jose Cruz - 27.4
30 - Roy White - 27.2
31 - Jeff Kent - 27.2
32 - Willie Davis - 27.2
33 - Babe Adams - 27.1
34 - Dwight Gooden - 27
35 - Chet Lemon - 27

If you place a 2x to 1x weight on defensive measures by Gauge vs Reference, you get the following:
01 - Joe Tinker - 43.5
02 - Chipper Jones - 41.9
03 - Scott Rolen - 41.3
04 - Art Fletcher - 39.2
05 - Jim Thome - 35.2
06 - Johan Santana - 33.5
07 - Bobby Bonds - 33
08 - Tommy Bond - 32.9
09 - Tony Mullane - 32.8
10 - Bob Johnson - 32.7
11 - Luis Tiant - 32.4
12 - Frank Chance - 31.8
13 - Charlie Buffinton - 31.8
14 - Urban Shocker - 30.9
15 - Buddy Bell - 30.9

16 - Tommy Leach - 30.6
17 - Roy White - 30.3
18 - Bobby Veach - 29.7
19 - Kevin Appier - 29.3
20 - Norm Cash - 29.2
21 - Kenny Lofton - 28.9
22 - Jose Cruz - 28.6
23 - Silver King - 28.4
24 - Sammy Sosa - 28.1
25 - Dizzy Dean - 27.9
26 - Vic Willis - 27.8
27 - Jeff Kent - 27.8
28 - Dizzy Trout - 27.8
29 - Eddie Cicotte - 27.8
30 - Jim McCormick - 27.7
31 - Noodles Hahn - 27.6
32 - Willie Davis - 27.5
33 - Babe Adams - 27.1
34 - Chet Lemon - 27.1
35 - Dwight Gooden - 27.0

You could also look at doubling WAA and then adding WAR to get a mixture of both, at 50/50 defense:
01 - Chipper Jones - 164.6
02 - Scott Rolen - 152.6
03 - Joe Tinker - 146.1
04 - Jim Thome - 141.4
05 - Andruw Jones - 132.5
06 - Tony Mullane - 132.3
07 - Tommy Bond - 130.7
08 - Art Fletcher - 129.2
09 - Luis Tiant - 129.1
10 - Charlie Buffinton - 127.7
11 - Buddy Bell - 127.6
12 - Bobby Bonds - 126.5
13 - Kenny Lofton - 122.2
14 - Bob Johnson - 121.7
15 - Jim McCormick - 121.1
16 - Johan Santana - 120.3
17 - Urban Shocker - 120.2
18 - Willie Davis - 119
19 - Vic Willis - 118.5
20 - Sammy Sosa - 116.4
21 - Eddie Cicotte - 115.6
22 - Jose Cruz - 114.4
23 - Kevin Appier - 114.2
24 - Tommy Leach - 113.6
25 - Norm Cash - 112.4
26 - Frank Chance - 112.3
27 - Jeff Kent - 111.5
28 - John Olerud - 111
29 - Roy White - 110.7
30 - Vladimir Guerrero - 110.4
31 - Bobby Veach - 109.7
32 - Silver King - 109.4
33 - Chet Lemon - 108.5
34 - Sal Bando - 107.2
35 - Chuck Finley - 107.0


You could also look at doubling WAA and then adding WAR to get a mixture of both, at 67 B-G/33 B-R defense:
01 - Chipper Jones - 158.6
02 - Joe Tinker - 154.5
03 - Scott Rolen - 150.8
04 - Jim Thome - 141.7
05 - Art Fletcher - 138.1
06 - Tony Mullane - 132.6
07 - Tommy Bond - 130.7
08 - Andruw Jones - 130.7
09 - Luis Tiant - 129.2
10 - Buddy Bell - 128.5
11 - Charlie Buffinton - 128.3
12 - Bobby Bonds - 127.4
13 - Bob Johnson - 124.7
14 - Tommy Leach - 122.8
15 - Jim McCormick - 121.4
16 - Johan Santana - 120.3
17 - Urban Shocker - 120.2
18 - Roy White - 120.0
19 - Willie Davis - 119.9
20 - Bobby Veach - 118.9
21 - Vic Willis - 118.6
22 - Kenny Lofton - 118
23 - Jose Cruz - 117.9
24 - Sammy Sosa - 115.7
25 - Eddie Cicotte - 115.6
26 - Norm Cash - 115
27 - Kevin Appier - 114.2
28 - Frank Chance - 113.6
29 - Harry Hooper - 113.3
30 - Jeff Kent - 113.3
31 - John Olerud - 111.3
32 - Tony Phillips - 110.4
33 - Vladimir Guerrero - 110.1
34 - Silver King - 109.4
35 - Chet Lemon - 108.7
   367. Rob_Wood Posted: December 13, 2017 at 08:53 PM (#5592212)
Kiko, Bleed, or anyone:

I have a simple question about Retrosheet deduced games ...

Let me pose a highly stylized example for my question. Suppose there is a game in which the total number of putout and assists are known for each fielder. Suppose we know Team A went down in order in the bottom of the ninth inning. Suppose we know that Joe, John, and Jack were the three batters (in order). Suppose we know that there was one strikeout, one flyout to centerfield, and one groundout to the shortstop, but we don't know which batter did which.

If this were a deduced game in the Retrosheet database, what would appear in the play-by-play portion of the bottom of the ninth inning? Would these fields be blank; would they be filled in using best judgment (or truing up to end-of-season batter stats to the extent that is relevant, pure guesses just to fill something in, etc.)?

I am perfectly fine with whatever Retrosheet decides is the appropriate thing (best judgment, etc.). Here I am simply asking an "administrative" question. Basically I think I am asking in a deduced game does every at bat have a specific outcome in the play-by-play fields even if those outcomes are not known with certainty or are some of the outcomes left "blank" (missing?).

Thanks much.

   368. 6 - 4 - 3 Posted: December 13, 2017 at 09:26 PM (#5592233)
I'm not sure that I understand the logic of doing a weighted average of WAA and WAR, when one is basically a linear transformation of the other. WAA is a sufficient statistic in my view as it encapsulates both career and peak value. In future ballots (as I just did with 2010 MMP), I'll do a simple average of pWOPA and WAA, since that is based on dramatically different methodology and so there's value in averaging... but WAA is essentially WAR, with the exception of it rewarding sustained mediocrity. I'm not familiar with Gauge.

There's no pre-1960s players in large part because the most deserving ones are already in, IMHO. Note: I had included Vic Willis on my preliminary ballot, but the previous discussion in this thread about 19th century pitchers convinced me to remove him. So no Tommy Bond, Charlie Buffinton, Tony Mullane, etc. either.

That said, Joe Tinker, Frank Chance, and Urban Shocker were oversights (I assumed Chance and Tinker were already in and mistakenly thought Shocker was a 19th century pitcher). I mainly researched the players mentioned in the initial post that listed those new to the ballot and returning players who finished well on the previous ballot (on that note, having a link to a list of everyone would be helpful to new voters). Anyway, all three would place outside the Top 10 (i.e., should have been on my ballot, but short of PHOM). I'll consider all in future ballots. Sorry, first time trying this.

I completely discount seasons during the WWII years, when most of the best players were not in MLB. That's why Bob Johnson fell off my ballot (4.5 WAA in 1944). The competition level just wasn't MLB in 1943-45 due to the absence of so many players who otherwise would have been in the majors and without a great season (1944) and two decent seasons (1943 and 1945), Johnson's down to 23.6 WAA, which is well below the threshold of consideration.
   369. Kiko Sakata Posted: December 13, 2017 at 09:56 PM (#5592249)
Kiko, Bleed, or anyone:

I have a simple question about Retrosheet deduced games ...

Let me pose a highly stylized example for my question. Suppose there is a game in which the total number of putout and assists are known for each fielder. Suppose we know Team A went down in order in the bottom of the ninth inning. Suppose we know that Joe, John, and Jack were the three batters (in order). Suppose we know that there was one strikeout, one flyout to centerfield, and one groundout to the shortstop, but we don't know which batter did which.

If this were a deduced game in the Retrosheet database, what would appear in the play-by-play portion of the bottom of the ninth inning? Would these fields be blank; would they be filled in using best judgment (or truing up to end-of-season batter stats to the extent that is relevant, pure guesses just to fill something in, etc.)?

I am perfectly fine with whatever Retrosheet decides is the appropriate thing (best judgment, etc.). Here I am simply asking an "administrative" question. Basically I think I am asking in a deduced game does every at bat have a specific outcome in the play-by-play fields even if those outcomes are not known with certainty or are some of the outcomes left "blank" (missing?).

Thanks much.


Rob,

In the specific example you cite: K, 8, 63, in some order (which is fairly unusual), I think the answer might depend on the deducer. The K should be knowable from box scores (batter K's is an element of box scores and has been an official statistic at least as far back as we've been deducing games so far). I might be inclined to put the 8 and 63 in some order just for completeness sake. Alternately, batting outs where the details are unknown are coded as "99". You can add some details if they're known: e.g., 99/G or 99/SH, or 99(1)/FO. If we have detailed play-by-play for part of a game (which sometimes does happen for whatever reason), I may try to deduce the missing outs if there are few enough (as in your example) based on Retrosheet's fielding records. But for games where we have the vast majority of outs that are simply unknown (e.g., all we have from newspaper sources is something like, "The only White Sox batter to reach base in the first five innings was Luke Appling when he singled through the box in the fourth inning") those will all simply be coded as "99".

If you go to Retrosheet's web site for such a game (e.g., here), the box score will show fielding totals that come from official records but they won't be distributed within the play-by-play - you'll have a bunch of "Warren out on unknown play; Blanton out on unknown play". If it's a deduced file, the page on Retrosheet's web site will say that at the top (as in the link in this paragraph). For event files, there may also be unknown batting outs. Again, the fielding records in the box score would come from official totals, but the play-by-play would include references to "unknown play[s]" - here's an example of such a game.

If we know that a batter struck out at some point in the game, but don't know when, the deducing standard is to put the K as early in the game as possible. More generally, that's a general deducing rule that Dave Smith told me when I started doing this: try to get the unusual stuff out of the way as early as possible. The end result will be a complete play-by-play in the sense of every at bat will have some play coded. If we know a player went 2-for-4 but know nothing else, his four AB would be coded "S", "S", "99", and "99" in some order.
   370. Kiko Sakata Posted: December 13, 2017 at 10:09 PM (#5592257)
I completely discount seasons during the WWII years


This seems unduly harsh. Actually, in the case you cite - Indian Bob Johnson - where he was 37 years old in 1943, I can see the argument: Bob Johnson may well not have had a job as a 38-year-old corner outfielder in a non-war-ravaged 1944 American League. Fair enough.

But, to pick my World War II era pet candidate: in 1943, Vern Stephens was a 22-year-old shortstop who was coming off a rookie year in which he batted .294/14/92 (thinking in terms of how the GM/managers of 1943 would have viewed him). To just say - "well, too bad, 1943 - 1945 don't exist; it's as if Vern Stephens just didn't play those three years" seems like it's not being fair to that generation. It's essentially the same argument why you need to give Johnny Pesky - who also had a strong rookie year in 1942 (.331, 205 H, 105 R) - WWII credit for the same three seasons. Discount Stephens' 1943-45 seasons, absolutely: either via direct deduction or even by substituting in some kind of guess based on 1942 and 1946. But if you just eliminate Stephens' age 22-24 seasons (and Pesky's age 23-25 seasons) I would think you'll end up with too few players from that generation.
   371. Rob_Wood Posted: December 13, 2017 at 10:38 PM (#5592270)
Thanks Kiko for the explanation of Retrosheet deduced games.

I second Kiko's concern about how Aspiring OTP Dancing Monkey (6-4-3) is completely ignoring performances during the WWII years. We have very few rules around here but "be fair to all eras" is one of them. There are many players whose performance during WWII seasons could influence whether a voter includes the player on this or future ballots. To simply wipe out three (?) seasons from players' careers seems unfair to players of that era.

In theory we have a ballot committee which is responsible for making the ultimate decision to include/exclude each and every ballot submitted. My personal opinion would be to ask the voter to reconsider his ballot in light of that edict (perhaps by applying a reasonable discount of wartime performance). From reading the account above, the voter in question also understandably overlooked a few old-timey players that he now thinks could well have been on his ballot.

Taken together with the fact that any modifications of his ballot would only take a few minutes of his time, I think it is worth asking for a resubmittal.
   372. Howie Menckel Posted: December 13, 2017 at 11:31 PM (#5592295)
I completely discount seasons during the WWII years

not only is that unreasonable, I think it may violate our Constitution.

I have already voted for Bob Johnson, yet in my comments on him I express my concerns about his WW II peak season. so I factored in the issue, but I didn't zero it out.

Hal Newhouser is a good example of the foolishness of a "complete discount" of 1942-45

1944: 29-9, 159 ERA+ in 312 IP
1945: 25-9, 195 ERA+ in 313 IP
1946: 26-9, 190 ERA+ in 292 IP

so he dominated 1946 AL hitters, but no credit for the same stats the previous two seasons?

Newhouser was in his age 23-25 seasons, and he didn't care who he was facing
   373. DL from MN Posted: December 14, 2017 at 08:55 AM (#5592340)
I'll do a simple average of pWOPA and WAA


I also incorporate WAA heavily but I like to add in WAPA (Wins Above Positional Average). The Hall of Merit's goal is to be fair to all eras (including WWII) and positions. I look at a season when the top 4 WAA finishers are outfielders and realize that it wouldn't necessarily be helpful for a team to acquire all of them at the expense of their catcher and shortstop.
   374. Carl Goetz Posted: December 14, 2017 at 09:32 AM (#5592353)
"First, as to the "sidebar", after constructing context-dependent wins (pWins), I then construct context-neutral wins, which are based on pWins but averaged across all contexts (e.g., all home runs hit at Wrigley Field in 2017 generated the same number of eWins, regardless of when they were hit). It is these eWins, then, which I compare to bWAR and fWAR in that article."
Sorry, got my pWins and my eWins confused. So if I understand you correctly, eWins are developed after pWins by taking every like event (ie HR in Wrigley) and averaging the pWin value and assigning that average to each player who did it?

Also, I did confirm that BBRefWAR does total zero for a given year. It appears that they are giving the extra WAR to the offensive side to start and then adjusting individual players up and down on the defensive side from there. I do tend to agree that player brought up are usually ready defensively before they are ready offensively. And if I remember the defensive peak chart correctly, it is typically fairly young and followed by a slow decline. This may make some sense as an approach, but I'll need to think on it a bit more.

I read the Lichtman and Tango stuff. I get the Bayes argument, but 35-40% reduction seems like a lot. How would that look in WAR terms? Would a reduction in Fielding WAR be an addition in Pitching WAR for a given season?
   375. Kiko Sakata Posted: December 14, 2017 at 10:35 AM (#5592394)
So if I understand you correctly, eWins are developed after pWins by taking every like event (ie HR in Wrigley) and averaging the pWin value and assigning that average to each player who did it?


Basically, yes.

I get the Bayes argument, but 35-40% reduction seems like a lot. How would that look in WAR terms? Would a reduction in Fielding WAR be an addition in Pitching WAR for a given season?


If you were to regress all of the fielding numbers toward zero by, say, 40%, it wouldn't affect the pitcher / non-pitcher split at all. It would just affect the split across fielders.

Let's imagine a simple world where these are all of the fielders in a particular league, with fielding runs as follows:

+20, +10, +10, +6, +2, 0, 0, -2, -2, -4, -8, -12, -20

That adds up to zero. The average fielder is perfectly average, as you'd expect.

Now, regress all of these numbers toward zero by 50% - i.e., cut all of these numbers in half:

+10, +5, +5, +3, +1, 0, 0, -1, -1, -2, -4, -6, -10

The players' rank is unchanged. And the sum is still zero.

But, at 10 runs = 1 win, the best fielder loses 1 WAR, the worst fielder gains 1 WAR, and the guys in between gain or lose smaller or lesser amounts - except for the guys who were already at zero, of course.
   376. DL from MN Posted: December 14, 2017 at 11:57 AM (#5592525)
So Andruw Jones gives WAR back to Manny and Sheffield.
   377. Carl Goetz Posted: December 14, 2017 at 12:03 PM (#5592535)
I get the distribution end of it for fielders, but that only works if WAR starts from a position of 50% value for offense and 50% value for defense. If Position players get 59% of WAR in BBRef and we reduce the fielding distribution, aren't we effectively "giving" the fielding WAR to hitters/baserunners? Presumably the extra 9% of Overall WAR to offense is actually for the defensive contribution of position players. Ie Starting with higher offensive credit is a way to compensate for using fielding above/below average (rather than replacement level) for the individual players. Even if the direct effect is simply to reduce the range of fielding WAR outcomes, we are still allocating 59% to position players and 41% to Pitchers. If total fielding WAR is zero, that doesn't change the overall allocation. Given that offense and defense as a whole should be close to 50/50 in value, shouldn't any regression of fielding also adjust the Position Player / Pitcher split?
For example, if we regress fielding by 40%, should the extra 9% for position players also be regress by 40%? In this case, 9- .4(9) = 5.4 which would move the split to about 55.4%/44.6% to keep things constant.
If we fail to make any adjustment, we are effectively increasing the value of offensive contributions relatively to pitching. If we assume the extra 9% is for fielding contributions, the effective offense/pitching/fielding split is 50/41/9 even if the actual reported WAR is a straight 59/41/0. If we reduce the effective fielding component to 5.4, we have effectively changed the effective split to 50/41/5.4; which no longer totals 100. Since the overall WAR and pitching WAR have not changed, all benefit of the fielding regression is effectively being accrued by the offense by default. Realistically, this changes the split to 53.6/41/5.4 which overweights offense.
I feel like this is a flaw in the WAR methodology. Assuming we agree with every component of WAR, its fine. But if I want to adjust one component, other things fall apart. While I disagree with a lot of James' WS methodology, one thing he got right was the starting step that assumed Offense and Defense are equally weighted. I haven't decided if I want to regress fielding or not (or to what degree if I do), but it seems clear to me that any reduction in credit/blame to fielders should have an equal and opposite reaction for pitchers.
   378. Kiko Sakata Posted: December 14, 2017 at 12:53 PM (#5592609)
So Andruw Jones gives WAR back to Manny and Sheffield.


Yes. Well, technically, Jones would give WAR to bad centerfielders (e.g., Bernie Williams). But, yes.

I get the distribution end of it for fielders, but that only works if WAR starts from a position of 50% value for offense and 50% value for defense. If Position players get 59% of WAR in BBRef and we reduce the fielding distribution, aren't we effectively "giving" the fielding WAR to hitters/baserunners?


Yes. WAR doesn't start from zero and build up. My system does and Win Shares does. But WAR starts from average. The starting point for position players is WAA = 0. They then make a final addition to get from WAA to WAR. And for THAT, they split the Rrep 59/41 between position players and pitchers (apparently). The final adjustment has no relation to the pitcher/fielder or offense/fielding splits. In theory, it should - i.e., in theory it should be based on how you believe pitcher/fielding value should be split.

What drives the relationship between offensive and fielding value is the range of batting runs, baserunning runs, and fielding runs above/below average.

And, in fact, I believe that the problem with WAR is that the offense/fielding split implied by the ranges on their batting runs and fielding runs numbers is wildly different from the split implied by their choice of a 59/41 split of replacement runs.

If value should split 50% offense / 9% fielding / 41% pitching that implies that, for position players, value splits 85% offense (50/59), 15% fielding (9/59), which would be consistent with the range on batting runs (technically, batting + baserunning runs) being 5.7 times greater than the range on fielding runs (85 / 15 = 5.7).

As a super simple oversimplification, BB-Ref says that in 2017, the top 5 players in Total Zone Runs had +32, +28, +25, +22, +20 and the top 5 players in Batting Runs had +66, +61, +57, +56, and +56 (Buxton and Votto led in these, respectively). The batting numbers are 2.3 times greater than the fielding numbers, which would imply a split between offense and fielding - assuming 59% total, which is fixed and unaffected by the batting and fielding numbers (both of which should sum to zero no matter how they're scaled) - of 41% offense, 18% defense (41/18 = 2.3).

So, yes, implicitly, WAR is ACTUALLY distributing value 41% offense, 18% fielding(*), 41% pitching. Which I do not believe to be correct.

(*) - I believe the correct way to figure out the 41/18 split would be to calculate the standard errors associated with batting runs, baserunning runs, and fielding runs, and use those to calculate the actual weights. I would not be surprised if that moved the fielding number by 2-3%; I'd be shocked if it moved it by the 9% it would have to move to push offense up to 50%.
   379. Carl Goetz Posted: December 14, 2017 at 01:31 PM (#5592669)
"So, yes, implicitly, WAR is ACTUALLY distributing value 41% offense, 18% fielding(*), 41% pitching. Which I do not believe to be correct."

So to summarize:
1) You feel the implied split of 50/41/9 is reasonably accurate to reality.
2) BBRef is simply not accomplishing that implied split because of problems with their methodology in merging offensive and defensive statistics to create total player value for position players.
3) If you were to use WAR (which you wouldn't because of your issues with run based systems), you would regress fielding WAR by the roughly the 35-40% recommended in the Lichtman article.

Are these fair statements as to your position on the matter?

You've definitely made a convincing argument to regress dWAR statistics (which thankfully Baseball Gauge makes easy) but I'm not sure how much yet. I'm also rereading Wizardry and learning what I can about TZ to decide which dWAR (DRA/TZ/Hybrid) to use for Pre-2003 years.




   380. Kiko Sakata Posted: December 14, 2017 at 01:53 PM (#5592687)
Carl,

Your summary of my position in #379 is correct in all regards. Thanks for engaging in this discussion with me. I've both enjoyed it and found it very helpful in clarifying my thoughts on the subject.
   381. Carl Goetz Posted: December 14, 2017 at 02:15 PM (#5592703)
Thanks to you as well. I've got a lot to figure out now for my 2019 ballot.
   382. Rob_Wood Posted: December 14, 2017 at 03:06 PM (#5592769)
I think this is relevant.

There are two underlying factors that get inexorably intertwined. Suppose there is a game in which the outcome is determined A% by the "offense" and (100-A)% by the "defense".

Suppose further that, for whatever reasons, the underlying dispersion of "offensive talent" (OS) is much different than the underlying dispersion of "defensive talent" (DS).

Now play thousands of games and keep careful records of the results for each offensive player (hitter) and each defensive player (pitcher). You will find that the resulting spread in the "values" (games won and lost) will be a mixture of A, OS, and DS.

That is, there are too many unknown variables to "fix" values to any pre-conceived notions. Put another way, there is no guaranteed relationship between the underlying split A and the resulting spreads of observables (such as "runs" in this stylized example). So jiggering a "value" formula until the observables match the assumed underlying parameters is not recommended.
   383. Dr. Chaleeko Posted: December 14, 2017 at 03:25 PM (#5592787)
Rob, in post 382 are you agreeing with or challenging Kiko's point? Or neither? Thanks!
   384. Rob_Wood Posted: December 14, 2017 at 04:09 PM (#5592821)
If I am understanding Kiko's position (which is a big "If"), then I am challenging his point.
   385. Carl Goetz Posted: December 14, 2017 at 04:23 PM (#5592827)
"There are two underlying factors that get inexorably intertwined. Suppose there is a game in which the outcome is determined A% by the "offense" and (100-A)% by the "defense"."

I get what you're saying, but ultimately a run saved on defense is just as valuable as a run earned on offense. I have a hard time with A being something different than 50% for a league for a season though. I get saying a 10-9 game was decided by the offenses and a 1-0 game by the defenses, but you really can't extricate the offense of one team from the defense of their opponent. the 10-9 game could have been good pitching facing great hitting or crappy pitching facing ok hitting. You can say a HR was a good pitch and great hitting or that he hit a mistake, but most of the time its hard to determine that. The interplay in a game is too complicated to say the offenses were 60% and the defenses were 40% responsible for the outcome. That may be true and it may not be, but the value of a run saved vs a run scored remains very close to equal as long as the game is close.
   386. Rob_Wood Posted: December 14, 2017 at 05:01 PM (#5592860)
I don't think it is so obviously 50/50. Consider slow-pitch softball. In my opinion the split in slow-pitch softball between offense and defense is something like 80/20 even though it is true that "a run saved on defense is just as valuable as a run earned on offense". Those two things are not the same thing.
   387. Kiko Sakata Posted: December 14, 2017 at 05:20 PM (#5592886)
If I am understanding Kiko's position (which is a big "If"), then I am challenging his point.


I think Carl gets at my issue very well in #385. Baseball has a perfect double-sided accounting structure. A positive event for one team's offense is a negative event of exactly equal value for the other team's defense.

If you build up from zero, this will inevitably lead you to having offense and defense being equal - because they're the two sides of the same coin - and to a pitching / fielding split that is "real" in the sense that it simply falls out of the data (there are points where subjectivity comes into play in determining that split, but given a specific set of subjective assumptions, the answer falls out of the data). My system does this. For every play, there are win pieces and loss pieces of equal size which are distributed to specific players at specific positions. And these are then added up to get your final totals. The relationship between pitching and fielding that comes out of my system falls out of that naturally. Win Shares is also built up from zero in this way (this is not a coincidence; Win Shares was essentially the inspiration for my Player won-lost records) and, because of this, ends up in a very similar place (fielding is less important relative to batting in Win Shares than in WAR).

The problem with WAR is that it's not built up from zero. It's built from average. But, because of this, there's no natural mechanism for ensuring that everything is appropriately scaled (Bill James discussed this issue somewhat with respect to Pete Palmer's fielding numbers in his original Win Shares book).

With fielding, there are two issues. At the aggregate (team) level, what is the correct split between pitching and fielding? And at the player level, what is the correct split between batting, baserunning, and fielding?

In my opinion, the "right" level of fielding has to yield a reasonable answer to both of these. And I don't believe that WAR does so. The 59/41 split of position-player vs. pitcher WAR implies that fielding is 18% of defense. But that implies a specific ratio of fielding to offense for position players (15% / 85%) which is not, I believe, consistent with the relative weighting of fielding vis-a-vis batting (and baserunning) within position-player WAR.

I think to really do this correctly, you have to have two pieces. You have to track wins and losses (or runs scored and runs allowed), because a win on one side (run scored) is a loss on the other side (run allowed), and that's the only way to be sure that you're tracking offense and defense and pitching and fielding on the same scale. In other words, it's not enough to say that Andruw Jones is +20 runs; to verify that things are balanced correctly, you have to say that Jones has, say, 40 runs saved against 20 runs allowed, so that you can then sum up the "runs" and the "runs allowed" on both offense and defense and for both fielders and pitchers and make sure that the numbers all make sense together at a league/season-wide level.
   388. Rob_Wood Posted: December 14, 2017 at 06:21 PM (#5592928)
I agree with all of that.

It sounds like I misinterpreted what you said earlier about using the spread in offensive and defensive runs to adjust value metrics.

I apologize for that.
   389. Howie Menckel Posted: December 14, 2017 at 07:19 PM (#5592958)
Consider slow-pitch softball.

well, NOT the annual BBTF summer get-together in Central Park where defense is...... elusive.
:)
   390. DL from MN Posted: December 15, 2017 at 09:25 AM (#5593110)
Brock - I'm okay with not double-counting Lou Brock's fielding errors. I didn't think WAR did that but would be interested if it did. There is no reason to count a ball that dropped after hitting a glove different than a ball that dropped because the fielder was out of position. However, a throwing error is more serious. How many of Lou Brock's errors were throwing errors or misplaying grounders?
   391. Fridas Boss Posted: December 15, 2017 at 09:47 AM (#5593118)
Will there be a BBTF mock hall of fame vote this year? Spruces up the holiday season!
   392. DL from MN Posted: December 15, 2017 at 10:22 AM (#5593141)
I can probably run a mock vote next week after HoM voting is done
   393. #6bid Posted: December 15, 2017 at 10:58 AM (#5593199)
Not a HoM voter but am curious about Brock Hanke's statement in the ballot thread:

"[Lou Brock] played in a ballpark that savagely suppressed homers"

His home-away splits on bb-ref: 74 HR home, 75 HR away.

What am I missing? Players do hit better at home I believe but is it a significant split? Would we expect a normal HR split to look something like 80H/75A?
   394. Fridas Boss Posted: December 15, 2017 at 02:23 PM (#5593394)
TYVM DL from MN!
   395. 6 - 4 - 3 Posted: December 16, 2017 at 01:45 PM (#5593727)
In theory we have a ballot committee which is responsible for making the ultimate decision to include/exclude each and every ballot submitted. My personal opinion would be to ask the voter to reconsider his ballot in light of that edict (perhaps by applying a reasonable discount of wartime performance). From reading the account above, the voter in question also understandably overlooked a few old-timey players that he now thinks could well have been on his ballot.

Taken together with the fact that any modifications of his ballot would only take a few minutes of his time, I think it is worth asking for a resubmittal.

As requested, resubmitted a ballot that includes Tinker, Chance, and Shocker while excluding Sosa, Olerud, and Kent.

With respect to treatment of WWII players, at this point in "HOM history" my views are really not an issue since all the deserving players from that era are already enshrined. The only player that is possibly up for debate is Johnson (Bridges has only 27.0 WAA, so he wouldn't make my ballot even if given full credit for WWII).

If you take away Johnson' 1943-45 season (age 37-39 seasons), then he's nowhere on the radar screen given the late start to his MLB career. Given him full credit for all his MLB years (30.9 WAA), then he's still borderline in my estimation. But given the circumstances (i.e., his age and relatively weaker performance in years immediately preceding WWII), I'm inclined to heavily discount the career year that came out of nowhere in 1944. So it's really within the question of whether to include him in the bottom third of my ballot or not. I think that it's within my discretion to determine that he doesn't make it. If you insert him as #12 or whatever on my ballot or whatever, then he would simply displace another pre-integration player (Chance). So I don't think that there should be an admissibility question related to my revised ballot.
   396. bachslunch Posted: December 16, 2017 at 02:32 PM (#5593753)
Question: when was the plaque room last updated? It looks like it may have been a few years.
   397. Mike Webber Posted: December 16, 2017 at 03:11 PM (#5593779)
@393 - I agree with you, Brock will have to explain why he thinks Brock might have more 20 homer seasons if St. Louis wasn't stick as a homerun park.

His three best road homer seasons
Year - Road Homers - Total Homers
1966 - 9 - 15
1964 - 8 - 14
1967 - 8 - 21

Three seasons he hit 6 road homers, which in includes 1965 when he hit 10 homers at home. His 16 homer output in 1965 is the second best of his career.

If I used the BB Ref play index correctly, from 1960 to 1980 NL teams hit 27,446 home runs. 13,756 of those were in home games, 13,690 in road games. So as a whole, it seems that you didn't get a huge bump in homers by simply being at home during that era.

   398. Rob_Wood Posted: December 16, 2017 at 03:53 PM (#5593792)
Yes, but many times the home team does not bat in the bottom of the ninth inning (I know you know that).

This can skew home/road totals by a significant amount.

I used to have at my fingertips (meaning on my computer) the total number of home vs road innings batted for major league seasons, but I cannot locate it at the moment.
   399. Bleed the Freak Posted: December 16, 2017 at 04:30 PM (#5593808)
I've been asked to resubmit my ballot:

3) Lofton (38.2 WAA): Best leadoff man of his generation
11) Tinker (30.5 WAA): Relatively short career, but always above average
14) Shocker (29.0 WAA): Too short a career without enough of a peak to compensate
15) Chance (28.3 WAA): Not quite enough of a peak to compensate for lackluster career value


Thanks for the updates 6-4-3.

Do you adjust for schedule length or war credit?
The 4 quoted have truncated seasons, and Shocker was a stud that missed a partial 1918 campaign.

With respect to treatment of WWII players, at this point in "HOM history" my views are really not an issue since all the deserving players from that era are already enshrined.


New information comes to light with the passing of time, so I wouldn't say all deserving players are enshrined, or that any electorate is 100% full proof.
A challenge with the 40s/50s guys is how much to discount the 1943-1945 seasons, how much war credit to estimate for players serving in WWII and Korea, integration/negro league credits.

Guys who are intriguing in this mold include: Bus Clarkson, Dolph Camili, Luke Easter, Tommy Henrich, Don Newcombe, Johnny Pesky, Phil Rizzuto, Hilton Smith, Vern Stephens, Dizzy Trout, and others.
   400. bachslunch Posted: December 16, 2017 at 04:50 PM (#5593816)
Flip
Page 4 of 5 pages ‹ First  < 2 3 4 5 > 

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Dynasty League Baseball

Support BBTF

donate

Thanks to
robneyer
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 1.0598 seconds
41 querie(s) executed