Note that elect 3 is tentative pending our review as discussed back in December.

Discussion threads created for Bernie Williams, Javy Lopez, Tim Salmon, and Brad Radke.

I just tossed in Ruben Sierra and Edgardo Alfonzo too.

I just tossed in Ruben Sierra and Edgardo Alfonzo too.

Two guys who looked, earlier in their careers, like they were going to be more than they turned out to be.

Dan R -

Please lend me some assistance if you have an opportunity.

At Baseball-Fever.com, I voted for Brian Giles, but against Jeff Kent, partially on the basis of your WARP Analysis.

Can you please breakdown why Giles receives and deserves a vote for election and why Jeff Kent might not?
I referenced some excellent material about Kent you posted on this thread, http://www.baseballthinkfactory.org/files/newsstand/discussion/mlbcom_jeff_kent_retires/, so I mainly am looking for a look at Giles.

In addition, Kent rates higher than Giles by Sean Smith's WAR by a huge margin - is this a replacement level and defensive value issue?

Thanks in advance,
Bleed

P.S. - I didn't even plan to mention Giles being blocked in the minor leagues for far to long, but he could be deserving of MLE credit too.

This has the makings a a backlog election. Just as a reminder, here's what the top of the backlog looks like:

4. Palmeiro, 472 points, appeared on 29 ballots
5. Cone, 332, 26
6. Reuschel, 235, 16
7. Rizzuto, 225, 18
8. Tiant, 183, 16
9. Redding, 180, 13
10. Duffy, 172, 14
11. Cravath, 161, 16
12. Newcombe, 157, 12
13. Walters, 137, 10
14. Willis, 132, 11
15. McGriff, 119, 10

The next ten after that: Puckett, (Bobby) Bonds, (Bob) Johnson, Bando, Bridges, Williamson, Grimes, Leach, (Bus) Clarkson, Perez.

This may well be the last backlog election for quite a few years.

I will start beating the warpath for Reuschel as soon as the electorate begins to pay attention.

long before any actual serious review, my hunch is that I'll go Palmeiro and Cone as 1-2. After that, Bernie is in the discussion (among many guys who I don't really support but I have to vote for someone, so....).

I'll be open to lobbying for pet candidates, for sure.

Addie Joss. Centenary this year and a 142 ERA+ The HOM's greatest injustice!

Sure thing, Bleed the Freak. First, here's Giles's line in my WARP--not including 2009 but I think we can agree a 52 OPS+ season doesn't really help Giles's case.

Brian Giles



The case here is pretty straightforward. Even before minor league credit, which I think Giles obviously deserves as he was about as blocked as one could be and then went nuts as soon as he got a starting job, his MLB record is well over the in-out threshold of $150-$155M. The question then is, why is my system so much more favorable to Giles than Sean's?

The first big discrepancy regards Giles's offensive value in San Diego. After converting wins to runs and adjusting for the DH where necessary, I have Giles at 242 batting runs above average through 2002, while Sean has him at 230--close enough. From 2003-08, however, I have him at 171 runs above average, while Sean only gives him 141. The fact that a big gap starts opening up right in 2003 suggests that maybe we are using different park factors? I use the multi-year ones from baseball-reference. Another would be if the Padres consistently scored fewer runs than their offensive stats would suggest over that time, since Sean works backwards from team runs whereas I use the same run values for each event across all teams in the same league-season. Here's a breakdown of his 2005, to use one example, in case anyone can catch a flaw in my work:

103 singles * 0.5 + 38 doubles * .72 + 8 triples * 1.04 + 15 home runs * 1.44 + 112 unintentional walks plus hit by pitches * .33 + 9 intentional walks * .25 + 8 sacrifice flies * .37 + + 317 fielded outs * -.106 + 64 strikeouts * -.115 + -3.3 net double plays * -.37 = 111.2 eXtrapolated Runs. Divide by a .92 park factor to get 120.9 XR. He made 317 + 64 + 8 - 3.3 = 385.7 outs. The average NL team that year made 4,143 batting outs, so on an average team, his teammates would have had 3,757.3 batting outs. The league scored .174 runs per batting out, so his teammates would produce 653.8 runs, plus his 120.9 makes 774.7. The average NL team scored 721 runs that year. 774.7 RS and 721 RA gives a Pythagenpat exponent of 1.88 and a winning percentage of .5337. .0337 * 162 games gives 5.5 wins above average, as can be seen in the BWAA1 column above.

The second big discrepancy has to do with fielding. I have him as an average fielder for his career; Sean has him with a poor -48 through 2008. Once again, our numbers line up through the Pittsburgh years, and diverge sharply in San Diego. I show Giles as a +8 fielder from 2004-08; Sean has him at -44. Here the play-by-play stats clearly back me up: Dewan's Plus/Minus puts him at +15 and UZR shows -7. Perhaps unsurprisingly, Fangraphs WAR has Giles at 57.8 for his career, a very close match for my score.

"I will start beating the warpath for Reuschel as soon as the electorate begins to pay attention."

He finished 6th last year - how much more attention do they need to pay? :-)

Or did you just mean you'd wait until November . . .

Right, the latter.

Here's a prelim - the backlog election makes it easy

1) Tommy Bridges
2) Rick Reuschel
3) David Cone
4) Bus Clarkson
5) Urban Shocker
6) Phil Rizzuto
7) Gavy Cravath
8) Rafael Palmeiro
9) Luis Tiant
10) Bob Johnson
11) Ben Taylor
12) Bert Campaneris
13) Bucky Walters
14) Dave Bancroft
15) Hilton Smith

For the first time in years I have several non-PHOM guys on the ballot. PHOM additions this year are Rizzuto, Palmeiro and Early Wynn.

DL, where does Bernie land on that (or anyone else new, if above Bernie)?

Bernie's around 150th. Tim Salmon is the best new guy at about 115.

I neglected to address the minor league credit issue for Giles. I would argue that 1995 was his "make-good" year in the minors (a Szymborski MLE of .369 OBP and .452 SLG), and therefore that he starts deserving minor league credit in 1996. His 1996 Symborski MLE of .378/.533, with corresponding playing time at corner outfield using his 1997-98 fielding rate regressed, say, 1/3 to the mean, comes out as:



I also need to credit him for going 5-for-9 with a homer in 1995, which comes out as follows:



However, this takes no account of the time Giles spent rotting on the Indians' bench. He clearly was good enough to start full-time from 1996-98, but was only used as a platoon player over that time. So here's how I'd fill in the "missing" PA:

1. Take his 1999-2008 SFrac per year as a measure of his "true" durability. That gives 0.95 a season.

2. Credit him with 0.95 SFrac for each 1.00 SFrac he did not play while on major league rosters from 1996-98. This gives a "missing" SFrac of .24 in 1996, .34 in 1997, and .17 in 1998, which translates to 170 "missing" PA in 1996, 236 in 1997 and 118 in 1998.

3. Compare his % of PA vs. RHP from 1996-98 to the rest of his career. In 1996 he had 118 PA vs. RHP and 25 vs. LHP; in 1997 he had 371 vs. RHP and 80 vs. LHP; and in 1998 he had 370 vs. RHP and 60 vs. LHP. For the rest of his career he had 4756 vs. RHP and 2055 vs. LHP, which is 69.8% RHP. So to get him to the same 69.8% ratio for 1996-98, we have to add 100 PA vs. RHP and 70 vs. LHP in 1996; 107 PA vs. RHP and 126 vs. LHP in 1997; and 12 vs. RHP and 106 vs. LHP in 1998.

4. Determine his true platoon split. Using Tangotiger's method, I get Giles' regressed career OBP vs. LHP at 89.33% of his OBP vs. RHP, and his regressed career eXtrapolated Runs/out vs. LHP at 71.65% of his XR/out vs. RHP. (The unregressed ratios are 87.7% and 65.77%).

5. Take his rate of offensive production for the time he did play in 1995-98, which was an OBP of .390 and .262 XR/out while facing 83.9% RHP, and separate it out into platoon splits. At the above ratios, that comes out to a .398 OBP and .275 XR/out vs. RHP, and a .356 OBP and .197 XR/out vs. LHP.

6. Normalize these splits to the corresponding league-seasons. The AL averages for OBP and XR/out from 1996-98 were .350, .340, .340 and .208, .191, and .194. Weighted for Giles's playing time, the overall averages for that period were .341 and .195. So for 1996 we get .398*.35/.341 = a .409 OBP and .275*.208/.195 = .293 XR/out vs. RHP, and a .356*.35/.341 = .365 OBP and .197*.208/.195 = .210 XR/out vs. LHP. Repeating the process, the figures are a .397 OBP and .269 XR/out vs RHP and a .355 OBP and .193 XR/out vs LHP for 1997, and a .397 OBP and .274 XR/out vs RHP and a .355 OBP and .196 XR/out vs LHP in 1998.

7. Fill in the blanks. In 1996 we're adding 100 PA of .409 OBP and .293 XR/out and 70 PA of .365 OBP and .210 XR/out. In 1997 we're adding 107 PA of .397 OBP and .269 XR/Out, and 126 PA of .355 OBP and .193 XR/out. And in 1998 we're adding 12 PA of .397 OBP and .274 XR/out and 106 PA of .355 OBP and .196 XR/out. In total, that's 104 extra outs and 26.7 extra XR in 1996, 146 and 33.0 in 1997, and 76 and 15.4 in 1998.

8. Put him in corner outfield using his 1997-98 fielding rate regressed 1/3 to the mean. Here's what the credit looks like:



...so add it all up, and what do you get??

Brian Giles



Well, that's certainly a Hall of Famer. It's somewhere between Zack Wheat and Al Simmons.

not including 2009 but I think we can agree a 52 OPS+ season doesn't really help Giles's case.

Depending on how you deal with negative WARP values, it could actually hurt Giles's case significantly. BBREF has Giles's 2009 at an atrocious -3.9 WAR in just 253 PA, driven by an unfathomably bad -26 (!) TZ. UZR and +/- are a little more favorable, at -8.3 and -15, respectively, and TZL (Total Zone using Gameday hit location) has Giles at only -2.5, but Giles still accumulated -1.6 Fangraphs WAR.

Interestingly, for the years with TZL data (2005-2009), TZL has Giles at -22 even though TZ has Giles at -60. As far as I understand it, in Total Zone Sean Smith only looks at which OF fielded the ball, but in TZL Smith looks at the Gameday-estimated hit location, so I guess this discrepancy could result from Giles ranging over to field more hits that weren't easily caught.

I certainly don't believe in docking players for sub-replacement seasons, and I don't know if there's a single HoM voter who does. It's not a player's fault that their managers don't know they should be benched. If you're paid to play and asked to play and want to play, you play, even if it's not in the team's best interest. A different case can be made for mid-career collapses like Jimmy Wynn's 1971 or Andruw Jones's 2007-08, since in those situations you can't blame a manager for sticking with his star.

In my 2011 ballot I docked players for negative WAR seasons, though this decision wasn't the result of particularly careful scrutiny: my basic viewpoint is that when a player puts up a sub-replacement season, he cost his team wins on the field and should be punished in a career ranking. But I'm intrigued by your exception for mid-career collapses––in these cases it seems that while a player performed at a below-replacement level, his true talent was still above replacement level (which may not be true for a player at the end of his career). Thus you would exclude any season (or part of a season) in which a player was projected as a below-replacement level, reasoning that he shouldn't have had a job to begin with.

In Giles's case, I don't think it was completely clear at the beginning of 2009 that he was done--after all, he'd just put up 4.6 WARP and 3.9 bWAR in 2008. So maybe he should have been benched earlier than he was, but it still would have taken his managers some number of plate appearances to figure out he was finished--maybe 100 instead of 250 (that's just a total guess though).

That seems like a fair approach to me.

Using a Marcel-projection of Giles's bbref WAR, I calculated that his "true talent" WAR fell below replacement level after 145 PA--so under this method he would be credited with only -2.2 WAR, not -3.9 WAR in 2009 (although I seriously mistrust the TZ numbers for Giles, so don't take this as an endorsement of WAR's rating of Giles).

It's not a player's fault that their managers don't know they should be benched.

What about the all time hits leader :-)

Regarding posts 10 and 16, thank you Dan.

In an election of 32 voters so far, I am the lone voter for Brian Giles, while Jeff Kent has received 23 nominations.

Very Prelim

PHoM: Larry Walker, Bernie Williams, Buddy Bell

1. Rafael Palmeiro
2. Fred McGriff
3. Tony Perez
4. Mickey Welch
5. Tommy Leach
6. Hugh Duffy
7. Bernie Williams
8. Bobby Bonds
9. George Van Haltren
10. Vic Willis
11. Buddy Bell
12. Norm Cash
13. Mark Grace
14. Bill Monroe
15. Dick Redding

16-20. B.Johnson, Puckett, Olerud, Bando, Streeter
21-25. Grimes, J.Greene, Strong, Gleason, Mullane
26-30. N.Robinson, Elliott, Brock, W.Davis, John

Once again Rusty and I are the compare/contrast on how to treat modern pitchers

Hi, everbody! (Hi, Dr. Nick.) I found this project awhile back and have looked through a lot of the previous discussion/ballot/result threads. Wish I had known about it back when it started, but better late than never.

Anyway, I plan on submitting a ballot this year. Obviously there's a lot of time yet to go, but I have begun work on a system to rank players.

Couple questions if anyone is inclinded to answer:

1) Some of the older threads mention a yahoo group for the HOM. Is this still in existance? Are there still useful items on it?

2) What's the best data source for the remaining NgL players? Just their threads? Was the data from the HoF research ever released (it's only been 5 years)? Did the i9s site some of the older threads mention up and vanish (I can't find any trace of it)? I had heard the data gathered for the Strat-o-matic NgL set might be coming to bb-ref at some point, but I don't know the status of that.

3) Are there any metrics I just have to use? There's some I already have plans to use. I started trying to build a spreadsheet to rank guys, using the yearly WAR numbers on bb-ref, but I found it tedious to do what I wanted to do. So instead I've imported the WAR and WSAB numbers from Baseball Gauge (since they allow a csv download of their entire dataset for all years) into a DB I've set up, and am working on some queries to rank guys using those. I've also imported the Lahman data and and am calculating some numbers like wOBA and extrapolated runs, etc. to throw into the pot. It's a decent amount of work already, but it's a lot less boring than copy and pasting to Excel. So if someone wants to plug some metric that I can calculate from the basic numbers (I don't intend to use PBP data at this time) or can easily download a full 1871-2010 dataset for, go wild. I figure more analysis is better than less.

Anyway, when I was planning on doing this by hand with a spreadsheet I ranked about 30ish players (all of the top 10 returning, plus Bernie Williams and selected others), and my #1 player was the immortal Ed Williamson :) It'll be interesting to see if he's still in the lead when I get my new system set up.

Holy smokes, Nate, you went to Hollywood Upstairs Sabermetrical College too?

Yep. Club Med Sabermetrics School too. :)

I will start beating the warpath for Reuschel as soon as the electorate begins to pay attention.

I'm never going to be paying that much attention.

Hey Nate, welcome to the Hall of Merit. For question 2, the best source at this point is probably our own threads for individual Negro League players. You'll find MLEs (major league equivalencies) for most of the current candidates, though that's only one tool that you'll want to consider. The baseball Hall of Fame site had pdfs of their numbers but they took that down a while back. And yes, i9s has unfortunately gone poof. If you could get your hand on the stratomatic research, that would be really helpful. However, stratomatic doesn't necessarily adjust their numbers for the level of league play so somebody who put up monster numbers in a lower league (like Art "Superman" Pennington) looks comparatively better than he should. It's always good to check any one source (including the MLEs here) against other sources and the historical record for context.

Alright, I'll look through the threads and see what else I can find. Seems like the candidates that it's most important to get up to date on are Redding and Clarkson, with Hilton Smith and Ben Taylor trailing a bit behind. There's some others I'd like to take a good look at (Stovey, Cepeda, Chino Smith, Silvio Garcia, White, Baro, Andy Cooper, maybe some others), but realistically it doesn't look like anyone outside of those first four have a shot this year, so I'll start with them.

Was the data from the HoF research ever released (it's only been 5 years)? Did the i9s site some of the older threads mention up and vanish (I can't find any trace of it)?

I downloaded the HOF pdfs last year, although I can't remember where I found the link. I can email them to you though. i9s is gone, but some of the player pages can be accessed on the Internet Archive.

Stovey, Cepeda, Chino Smith, Silvio Garcia, White, Baro, Andy Cooper, maybe some others

An interesting list, given that some of these guys have received scant attention from the HOM recently (Stovey, Garcia, White, Baro, Cooper). Cooper's case looks especially intriguing, since the HOF inducted him in 2006 based on their new data. I don't think we ever computed any MLEs for Cooper based on the HOF numbers, so he might be due for another look.

I downloaded the HOF pdfs last year, although I can't remember where I found the link. I can email them to you though. i9s is gone, but some of the player pages can be accessed on the Internet Archive.

If you could email me the files, that'd be great. The email in my profile should work.

An interesting list, given that some of these guys have received scant attention from the HOM recently (Stovey, Garcia, White, Baro, Cooper). Cooper's case looks especially intriguing, since the HOF inducted him in 2006 based on their new data. I don't think we ever computed any MLEs for Cooper based on the HOF numbers, so he might be due for another look.

Having looked through the older threads, it seems there was a feeling among some posters that the next tier of NgL pitchers, after the obvious or near-obvious greats, got scant attention. Pitchers either tended to be voted in, or to at least hang around the upper portion of the results (e.g. Mendez and Redding), or completely disappear. With the exception of Leroy Matlock, who a couple voters liked, there really weren't any pitchers in the Eddie Cicotte or Urban Shocker mold, who have hung around the lower portions of the results forever. Though I suppose Hilton Smith has taken up that role now. But I suspect for a lot of voters, distinguishing between players like Nip Winters, Bill Byrd, Andy Cooper, Hilton Smith, Leroy Matlock, and Chet Brewer (and I suppose you could throw in guys like Verdell Mathis or Tiant Sr, who got even less attention), with the combined problems of evaluating pitchers and NgL players, was just a confusing mess. So none of them really rose to the top (with the exception of Smith more recently). Which isn't to suggest that any of these guys have a strong case for the HOM, just that while there was/is substantive disagreement over white pre-integration pitchers (I mean, Sam Leever got 2 votes last year...), the problems of evaluating these players resulted in a stronger consensus.

Anyway, I looked at the Cooper thread. His last MLEs were from June of 2005 and labeled a rough draft. In light of the HOF results, another look at him might be warranted. Of course, he had great success as a manager with the Monarchs as well, which would be appropriate for the HOF to consider, but outside the scope of the HOM (his HOF plague mentions both his playing and managing, though more attention is given to his playing career), which could mean he was a good choice for the HOF but wouldn't be for this project.

Nate, I can't email the pdfs to you through the BBTF email console, can you reply to my email so I have your email address, and then I'll attach the pdfs. I sent this to you as an email, but I'm not certain that it went through. Thanks.

I like Cooper as a re-look candidate, even moreso in a dreadful newcomer year.

Also I think we need to also look hard at the 2010 near-misses, who are a far different crop from 2011 counterparts in large part due to a smaller/different electorate. Probably not fair compared to some years to focus a bit more on just the latest runnerup class.

Might also be worth some positional discussions.

I think the consensus is that we're fully stocked on HOM catchers from the options, but we can discuss Elston Howard a bit and take maybe one last look at Wally Schang and Ernie Lombardi, for instance.

At 3rd base, meanwhile, we probably have 10 more serious candidates in play. Might help all of us to let them duel it out and see if some of us get comfortable with a pecking order. And if not, well, then they all knock each other off in perpetuity. That's ok, too, if we get stuck.

The MMP rules discussion thread is still sitting buried down deep among the week-old Newsblog items. I put a new comment on it, but it stubbornly refused to pop up on Hot Topics. No idea why, other than it is cursed with the curse of the When We Catch Up: The Hall of Merit After 2007.

2007!

We have been talking about what to do next almost as long as the Hall of Merit was active.

Oh well. Nothing lasts forever.

Even the November rain?

Even cold November rain.

So... Here's my preliminary preliminary ballot. A consideration set, I suppose. I've got a working system now, and this is what it spit out for MLB players. Basically, using a combination of stats (some uber, some not), my program went through each year since 1871 and assigned players points based on things like "Who was the best player in MLB? Who were the best players at each position?" etc. I also gave points for each prime year (which I define as an above average year). So my ballot will be based pretty strongly on prime, positional dominance, and rank amongst one's peers. I'm not comparing career stats at all, or seasonal stats beyond the year in which they occurred, so players with similar career numbers, or similar WAR totals, might rank quite differently, based on how they fit into their era.

The last column is how many points each player accrued. Let the gnashing of teeth commence. I expect my placement of Bonds, Canseco and Rice, at least, to draw some ire.

1 Bobby Bonds RF 277
2 Rafael Palmeiro 1B 276
3 Pie Traynor 3B 270
4 Dale Murphy CF 258
5 Ed Konetchy 1B 252
6 Jack Clark RF 250
6 Jim Whitney P 250
8 Tommy Bond P 249
9 Jose Canseco DH 246
10 Jim Rice LF 243
11 Frank Chance 1B 239
12 Tommy Leach OF 235
12 Fred McGriff 1B 235
14 Dave Bancroft SS 233
15 Jack Fournier 1B 231
16 Rick Reuschel P 228
17 Buddy Bell 3B 224
17 Dave Concepcion SS 224
19 Bill Mazeroski 2B 221
20 Hal McRae DH 220
21 Fred Dunlap 2B 219
21 Cesar Cedeno CF 219
21 George Foster LF 219
24 Albert Belle LF 217
24 Rocky Colavito RF 217
26 Bucky Walters P 216
27 Fred Lynn CF 215
28 Bill Bradley 3B 209
29 Carl Mays P 207
29 Kirby Puckett CF 207
31 Bob Elliott 3B 206
31 Tommy John P 206
33 Dwight Gooden P 203
34 Jesse Barfield RF 200
34 Harlond Clift 3B 200
34 Gene Tenace C 200
34 Bernie Williams CF 200
38 Denny Lyons 3B 199
39 Don Baylor DH 197
40 Al Rosen 3B 196
41 Ned Williamson 3B 194
42 Davy Force SS 193
42 Bob Johnson OF 193
44 Wally Schang C 192
45 Norm Cash 1B 191
46 Orel Hershiser P 190
46 Dave Parker RF 190
48 Vern Stephens SS 189
49 Bobby Bonilla 3B 188
49 Ron Guidry P 188
49 Tim Salmon RF 188
49 Dizzy Trout P 188
53 Toby Harrah 3B 185
54 Harold Baines DH 184
54 Cy Seymour OF 184
54 Orlando Cepeda 1B 184
57 Greg Luzinski LF 183
58 George Uhle P 182
58 Robin Ventura 3B 182
60 Larry Doyle 2B 180
60 Amos Otis CF 180
62 Burleigh Grimes P 179
62 Camilo Pascual P 179
64 Darryl Strawberry RF 178
65 John Olerud 1B 177
66 Willie Kamm 3B 176
66 Ken Singleton RF 176
68 Jack Clements C 174
68 Pedro Guerrero 1B 174
68 Joe Tinker SS 174
71 Ken Caminiti 3B 173
71 Gil Hodges 1B 173
73 Hugh Duffy OF 172
74 Tony Fernandez SS 171
74 Jimmy Ryan OF 171
76 David Cone P 170
76 Eric Davis CF 170
76 Fernando Valenzuela P 170
76 Tony Perez 1B 170
80 Mike Griffin OF 169
80 Gil McDougald 2B 169
80 Don Newcombe P 169
83 Dennis Martinez P 168
84 Travis Jackson SS 166
85 Brian Downing DH 165
85 Johnny Pesky SS 165
87 Don Mattingly 1B 164
88 Johnny Callison RF 163
88 Jim Fregosi SS 163
88 Dom DiMaggio OF 163
88 Rabbit Maranville SS 163
92 Schoolboy Rowe P 162
92 Mickey Tettleton C 162
92 Tom York OF 162
92 Curt Welch OF 162
96 Ron Cey 3B 161
96 Arlie Latham 3B 161
98 Phil Rizzuto SS 160
98 Hack Wilson OF 160
100 Tony Lazzeri 2B 159
100 Red Lucas P 159
100 Steve Rogers P 159
100 Jim McCormick P 159
104 Kenny Rogers P 158
105 Jim Kaat P 157
105 George Burns OF 157
105 Fred Tenney 1B 157
108 Addie Joss P 155
108 Jesse Tannehill P 155
110 Rico Carty LF 154
110 Kiki Cuyler OF 154
112 Jay Bell SS 153
113 Chuck Klein OF 152
113 Tony Mullane P 152
113 Orator Shaffer OF 152
116 Del Pratt 2B 151
116 Bobby Veach OF 151
116 George Van Haltren OF 151
119 Dick Bartell SS 150
119 Wally Berger OF 150
119 Lave Cross 3B 150
119 Rico Petrocelli SS 150

So, like I said, this is just what the system spit out, without me making any subjective changes yet. I imagine I will. For instance, Bond and Whitney's high placement is due to the uber stats really liking 19th centry pitching. It considers both the best player in baseball at least once (Bond in 1878, Whitney in 1881/1882), which carries the highest amount of points with it (20 per year). I think that's an overvaluation of the worth of pitching back then on a per inning basis, so they may well be adjusted downward.

To do list:

1) Go through the NeL candidates and try to estimate their amount of points.
2) Figure out if I want to give a bonus to any positions (Cs would be the primary one... CFs playing pre-1954 might be another, as they all get lumped into an "OF" position, but I'm not sure if CF really had that big of a defensive bonus back then), and if so, how much.
3) Go through each candidate in... probably the top 100, and figure out if any need credit for play outside MLB (I don't give war credit, per se, but I will give credit for play in the military, among other things).
4) Give each candidate a final review to see if I'm missing anything on them that would require further subjective adjustment.

At 3rd base, meanwhile, we probably have 10 more serious candidates in play. Might help all of us to let them duel it out and see if some of us get comfortable with a pecking order. And if not, well, then they all knock each other off in perpetuity. That's ok, too, if we get stuck.

Well, my system sees Traynor as the best available 3B candidate, without it being particular close. He's never gotten much support here, it seems like, though I'm not completely sure why. He wasn't a mistake HOF pick, in the sense of the VC picks where, even at the time, given what they knew, the selections were dubious. Instead, his reputation was as the greatest 3B ever, until Mathews came along, so it's clear why he was voted in. But was his reputation overblown?

Well, yes, to some extent, at least. It was built on two things, his fielding and his penchant for hitting .300. BA was hugely revered at the time, but he didn't walk a ton, and he didn't have much HR power during a power hitting era. Clearly his hitting, while above league average at a defensive position, wasn't as good as his contemporaries thought it was.

But as for his fielding, and the value of 3B defense at the time... that's a lot harder to quantify. Looking at BR, it doesn't like his defense at all. He's at -3.6 dWAR for his career. That's completely at odds with his career reputation, though that doesn't make it wrong. I'll have to look at some other defensive systems for him, when I get the chance, to see if they like him more.

Of course, it was a weak time for 3B. Even while hating his defense, the only contemporary rWAR sees as better is Hack, who sailed into the HOM, if I remember correctly. That's what my system likes about him (though it doesn't use rWAR), that he was fairly dominate for his position and era.

Wow, that's a very interesting preliminary list. Here are some of my thoughts on Nate's list:
Pie Traynor--If he was actually a +150 fielder rather than a -32 (putting him among the top fielding 3B of all time), Traynor would come in around 55 rWAR and would have a strong case for the HOM. I don't particularly buy the "dominance of position and era" argument, because some positions and some eras are harder or easier to dominate than others--look at 1950s CF, with Mays, Mantle, Snider, and Ashburn, for example. So I'm not entirely convinced that Traynor's dominance of 3B in his era is reflective of his greatness rather than reflective of 3B's weakness at this time. Also, how did you account for Negro Leaguers when determining the best player at each position? HOMers Beckwith and Wilson make the 3B position quite a bit stronger in this era, reducing Traynor's dominance.
Ed Konetchy, Frank Chance, Jack Fournier--Another group of players from a weak position/era, who don't really stand out when you look at their career totals. I think that Ben Taylor might be the answer here (you might also want to take a close look at Julian Castillo, a power-hitting Cuban firstbaseman from this era).
Jose Canseco--I don't buy this at all. Nate, how did you treat the DH in your analysis? Does being the best DH count the same as being the best 1B/2B/etc.? Since there can only be half as many DH's as MLB teams (and since many teams don't have a regular DH), being the best DH should not be equivalent to being the best at a particular postion. I see that you have a bunch of DH's rated highly (Canseco, McRae, Baylor) so I'm worried that you might be overrating DH's somehow. Jim Rice could also be affected by the DH problem.
Luis Tiant--I don't see Tiant on here anywhere.

I'm also interested in the details of your system. Which stats did you use to rank players? Does being the best player in 1880 receive the same weight as being the best player in 2010? What did you do to evaluate pitchers vs position players?

Overall, though, I think this is a good start, even if I disagree with the particulars of the rating, and I'm glad that you're participating in both the HOM and MMP projects, Nate.

Pie Traynor--If he was actually a +150 fielder rather than a -32 (putting him among the top fielding 3B of all time), Traynor would come in around 55 rWAR and would have a strong case for the HOM.

Michael Humphreys' fielding book Wizardry arrived yesterday from Amazon. He actually has his appendixes, including seasonal DRA data, online at http://www.oup.com/us/companion.websites/9780195397765/appendices/?view=usa. It looks like the data is available to anyone (though, unfortunately for our purposes here, he only calculates DRA from 1893 onwards). DRA quite likes Traynor's defense, much more so than BR WAR. Not to a +150 level though. He breaks fielders up into various eras, and Traynor was the 2nd best fielding 3B of the "Live Ball Era", according to him (though, as you'll see, this is a career measure, not a rate one, so Traynor's long career boosts his totals). Though only 16th all time, as what he calls the "Modern Era" (basically the 70s and 80s) had a lot of great fielding 3B, for whatever reason. Here's a snippet of his Live Ball Era career totals, for context:



(MCS is Mid Career Season. Runs is defensive runs saved above average.)

And here's what he has to say about Traynor:

Pie Traynor was for decades almost universally regarded by baseball fans as the greatest ever third baseman, mainly because his career batting average was .320 and he was believed to be a good fielder, which he in fact was. Bill James argued for many years, however, that Stan Hack was actually a much better overall player because he drew more walks and had a much higher on-base percentage, especially relative to his peers. Without going through a season-by-season analysis, it is clear that Traynor's approximately 150-run advantage on defense more than makes up for Hack's advantage on offense. Kell, Traynor and Hack were all very fine players, but, based on the best measures of overall value, not Hall of Fame impact players of the caliber of Ron Santo.

Kell and Santo are mentioned as he was previously discussing them in the 3B chapter. For whatever it's worth, DRA doesn't like Kell's defense (-50 for his career), as he was above average early on and then had a long decline phase. So, yeah, that last sentence isn't exactly a ringing endorsement of Traynor's HOM candidacy, but then, he doesn't think Hack belongs in the HOF either, and the HOM voters disagreed on that. And he thinks Traynor has more overall value.

Well, Hack's career rWAR is 54.5 and .3 of that is dWAR. Traynor's rWAR is 37.1 and his dWAR is -3.6. If we substitute DRA in, and use the shorthand of 10 runs per win (not completely accurate, but probably close enough), Hack's career rWAR would become 48.5, and Traynor's 49.9. Of course, this is career value, and I'm not a career voter myself, so that would hardly be definitive for me, but I still think it's clear, that if you buy Hurmphreys' defensive analysis, that the HOM electorate has underestimated Traynor. (Though my placement of him at #3 was before I had a chance to look at DRA. So I'll have to think about how I want to incorporate this data.)

(Also of interest here is that DRA absolutely loves Tommy Leach. He has the 7th highest career defensive runs above average at 3B, in spite of playing only half his time there. Of course, that short time at 3B kinda works in his favor, as he didn't have his defensive decline phase there, but DRA also likes his play in the OF, as well. At the end of the chapter he has on each position Humprheys' names a defensive top 10 all-time, which is a more subjective measurement than just his career value lists, as he takes into account things like peak. Leach ends up #9 on this list.)

More to follow on the rest of your comments later.

DRA quite likes Traynor's defense, much more so than BR WAR. Not to a +150 level though. He breaks fielders up into various eras, and Traynor was the 2nd best fielding 3B of the "Live Ball Era", according to him (though, as you'll see, this is a career measure, not a rate one, so Traynor's long career boosts his totals).

I'm replying to myself here, but I don't like how I phrased this. While Billy Werber accumulated his 105 defensive runs above average in much less fielding innings than Traynor accumulated the 92 that earned him second place for his era, it's not Traynor's long career that boosts his totals to somewhere near Werber's, while if you transformed this data into a rate stat (say, DRA/fielding innings) they wouldn't be anywhere close, it's his long defensively productive career that does so. DRA isn't a counting stat in the sense of hits or RBIs; someone who's hanging around on the field when they're below average reduces their career value, rather than adding to it. Now, neither Werber nor Traynor was a regular after their mid-30s (though Traynor hung around a bit while Werber was just done), and neither had much of a defensive decline phase, but Traynor got started in MLB much earlier and played more during their comparable seasons. Traynor had two bad defensive seasons, according to DRA; one at the beginning of his career (1922) and one at the end (1934). If you take just his defensive prime (1923 - 1933) he accumulated 117 defensive runs in more innings (13,931) than Werber played in his whole career. Werber only had one year that was below average defensively (1936), and it came in the middle of his career, so he doesn't have so neat of a prime. He was still a productive fielder in his last year (persumably it was his hitting that lead to losing his job, as he put up a 63 OPS+ in his last season), but I doubt he could have maintained his defensive runs lead over Traynor while playing as many innings as Traynor did.

FWIW, DanR's FWAA (which is basically an aggregation of BP's old FRAA and Fielding Win Shares) has Traynor at +91 over his career. So using either that figure or the DRA figure, Traynor comes in around 50 WAR, which places him right in the middle of my consideration set but not all that close to my ballot.

(Also of interest here is that DRA absolutely loves Tommy Leach. He has the 7th highest career defensive runs above average at 3B, in spite of playing only half his time there. Of course, that short time at 3B kinda works in his favor, as he didn't have his defensive decline phase there, but DRA also likes his play in the OF, as well. At the end of the chapter he has on each position Humprheys' names a defensive top 10 all-time, which is a more subjective measurement than just his career value lists, as he takes into account things like peak. Leach ends up #9 on this list.)

DRA has Leach at +227 at 3B and CF, while DanR's FWAA has him at +173 for his career. Both are well above Leach's career TZ of +67; in my updated HOM rankings, I replaced TZ with FWAA, resulting in Leach leaping to #4 in my rankings. Leach's favorable DRA rating only reinforces my new placement of him, and at this point I think he's very likely to wind up in the top half of my ballot.

I don't particularly buy the "dominance of position and era" argument, because some positions and some eras are harder or easier to dominate than others--look at 1950s CF, with Mays, Mantle, Snider, and Ashburn, for example.

I knew that aspect of my system would be controversial. The important question, as always, is how to measure performance from players of different eras against one another? It goes without saying that one can't simply compare the HR totals of a player from the 80s to one from the 90s to determine who was the better slugger (though some of the BBWAA apparently fail to realize this). Which is why most of the uber stats incorporate league average hitting into determining a player's offensive worth. Which over the course of a career (or prime) is, really, a measure of how dominant a player was for his era (though not position). Because we accept that if players as a whole hit much better in the one era than another then there was a systemic reason for it, even if we can't always fully explain what that reason was. And that influences our perception of a player's value. If we have two above average players, both with the same true talent level, but Player A plays in a time period when the true talent level of a league average hitter is higher (or the gap between the average and replacement player is smaller), then he'll accumulate less career value than Player B in most uber stats due simply to the misfortune of playing in a harder to dominate era. (I know some systems, such as DanR's WARP, attempt to correct for this by using standard deviations, but it's not completely clear to me that they succeed.) But, an argument goes, that is correct. Baseball teams don't care about a player's true talent level per se (particularly in a historic sense), they care about how much better they're going to be with one player than they would be with another. In that sense, player B derived more value from his talent level because he played in an easier to dominate era, regardless of whether the era was easier to dominate due to historical accident, or systemic issues.

And I think it's the same when adding position to the issue. I realize players at a position for a year is a smaller sample size than all hitters for a year, and thus more prone to lucky talent gluts or droughts, but, well, both are prone to those to some extent. The question is, what's the difference in how prone they are, and how big would it have to be for luck to be acceptable in the performance of the population of hitters, but not in the performance of the population of players at a position? And for that matter, how do we separate what's a lucky talent glut/drought from systemic issues that congregate talented players at a position (or make the players at a position look better than their true talent level)?

I really don't know the answers to any of that. So, for the purposes of my system, I'm taking it as an axiom that players who play at a position/era that is harder/easier to dominate generate less/more value. I'm sure others will disagree on that point. But I do mitigate this, to some extent, by just caring about the rankings rather than the raw totals. My system has Traynor as the best MLB 3B in 1923, 1925, 1927 and 1932. It doesn't matter if he was 1% better than the second best guy in those years, or 20% better, he gets the same amount of points. Another thing I'm sure others will disagree on the appropriateness of. But to me it represents an acceptable compromise, and a way to compare players across era without having to correct their statistics to put them all on even footing.

So I'm not entirely convinced that Traynor's dominance of 3B in his era is reflective of his greatness rather than reflective of 3B's weakness at this time. Also, how did you account for Negro Leaguers when determining the best player at each position? HOMers Beckwith and Wilson make the 3B position quite a bit stronger in this era, reducing Traynor's dominance.

I haven't accounted for them. I'm not sure that I will, in that sense. I do plan on going through the NgL candidates and attempting to figure out where they would rank, and then giving them points for that, but I'm not completely decided on whether to then deduct points from white players to compensate. I'm leaning towards not doing so. There's two issues, one practical, one... well, also practical, sort of. The first is, I give points for both what I would term best seasons and top seasons. That is, Traynor earns 48 points (12 per year) for his best at 3B seasons, but he also earns 66 points (6 per year) for his top seasons at 3B (being among the best 20% of MLB players at the position... so in 1933 3 3Bs get top at position points, while in 2003 6 do). And yes, best years are double counted (earning both best and top points). So to fully integrate the two populations of players, I wouldn't need to just figure out if any NgL 3B was the best at the position in, say, 1932, I'd also need to determine if any was the 3rd best, as that would remove 6 points from Joe Stripp (my #3 3B that year). Obviously that's pretty much an impossible task to do by hand, and I'd rather not do it haphazardly (that is, just going through the HOMers and strong candidates to see if they would claim points from white players, but ignoring the rest of the NgL players). The only other way I could see to deal with that is to apply an across the board penaly to pre-integration players; possibly a small penalty in the 19th century when few black players were active, increase it for 1900 to 1920, and then increase it again for 1921-1948 or whenever, when the NgLs were strongest. But I'd really rather not do that, as I feel it paints with too broad a brush.

My other issue is, I'm pretty skeptical of the ability of MLEs to accurately represent value. I've crudely done a few for the MMP project, but the rules that were decided on there require non-MLB players to be considered in an MLB context. That's not the case here. MLEs represent an approximation of a performance, while with MLB players we know their exact performance (if not always how to interpret it). That's not comparing like with like. Even without regression (which I believe some of the MLEs use and some don't, depending on who did them and when) it dampens the magnitude of peak/fluke seasons. Because if a player is translated at .87/.82 (the factors that I believe were used for the MLEs) that's an average of how players who made the transition performed, and there's no guarantee that any individual player would have played at that rate for his career, much less individual seasons. It looks like Beckwith's best year was 1925 when he hit .406 and lead his league in HR. If he could have played that year in MLB, what percentage of his performance could he achieved, given the same number of games? 87% may (or may not) be the correct average percentage, but what is the chance he could have achieved 93% (or 81%) of his NgL batting average? For that matter, if we could simulate him playing that year in MLB 1000 times, I suspect he'd out-do his NgL performance (that is, translate at over 100%) at least once.

A player's performance is not exactly equal to their true talent level. Not for a season, and not for a career. Yes, as the sample size grows our confidence that there's a strong correlation between results and ability increases, but it never becomes absolute. And for plenty of players the sample size never grows large enough for us to have confidence in a strong correlation, because they were unlucky enough to blow their arm out, or get hit in the head with a baseball, or whatever, and couldn't stay on the field. The players that we honor are the lucky ones; not just lucky to be born with their talents, and not to have been injured that much, but to have gotten great performances out of the talent. How many players are in the HOM now? 230 something? What's the real difference in talent between the 230th best player in baseball's history and the 250th? I imagine it's quite small, if it exists at all; instead of ability what we are honoring is that the 230th best player out-lucked the 250th best. But in the use of MLEs, we're discounting the chance that Beckwith could have been very lucky in 1925 and twinned what was undoubtedly a lucky hitting performance (since it was his best) with a 92%, 95%, or even a 98% translation (or gotten unlucky and translated at 80%). That strikes me as wrong.

But then, what are the alternatives? I don't trust player reputation, for all the usual reasons. And trying to determine NgL players' rankings in their own league (pretty much impossible to do systemically without a DB of NgL stats) and then slotting them against MLB players by gut isn't an alternative unless I want to completely abandon my normal approach. So I'm back to using a combination of the MLEs and their raw stats to try and figure out where they'd rank in MLB for each of their years. Which I figure is the best that I can do. But given my uncertainty, I'll give NgL players the benefit of the doubt that they could achieve at least what their MLEs say, but I'm not prepared to deduct from MLB players for something that may or may not have happened. In an integrated league, maybe Traynor would have been the best at 3B in 1925, maybe Beckwith would have, I can't say one way or the other with certainty. Given that I have the conflicting goals of not wanting to credit or debit players for things that didn't happen, but only for what they achieved, but also wanting to reward NgL players for their performance, I think I've settled on giving NgL players "extra credit". Yes, that means the population of players in 1935 will earn more points than the population of players in 1965, everything else being equal (that is, ignoring expansion). And some MLB players will be overrated compared to what they would have achieved in a hypothetical integrated league. But I think I'm OK with that. And, if nothing else, it represents that during the height of the NgL more players were playing at the top level available to them than immediately before or after. And those games had value.

I think I'll do something similar with wartime. That is, figure out where, say, Phil Rizzuto would have ranked for his play in the military in 1943, but not deduct the points that did go to players that year.

Jose Canseco--I don't buy this at all. Nate, how did you treat the DH in your analysis? Does being the best DH count the same as being the best 1B/2B/etc.?

Yes.

Since there can only be half as many DH's as MLB teams (and since many teams don't have a regular DH), being the best DH should not be equivalent to being the best at a particular postion. I see that you have a bunch of DH's rated highly (Canseco, McRae, Baylor) so I'm worried that you might be overrating DH's somehow.

There's one area where I've definitely been too kind to DHs. I didn't account for the lower population of DHs in calculating top positional seasons. That is, as I said above, in 2003 the top 6 3B got 6 points each. I also gave out 6 points each to the top 6 DHs. That was an oversight on my part and clearly wrong. I'll correct my system on that score when I get a chance.

I could also reduce the points awarded to DHs for the various categories. But I'm not inclined to do so at this time. DHs are already somewhat disadvantaged, as a group, in my system as they don't have the ability to generate fielding worth (of course for individual players this may benefit them, if they would have below average fielding value at a position). No DH has ever been the best player in MLB in my system, and only a handful have been among the top 2% of players. Frank Thomas came closest to the top in 1991, finishing 2nd, but fielding ability (and positional adjustments) pushed Cal Ripken past him. Additionally, I give out points for being the best fielder, a top fielder, best fielder at position, and a top fielder at position. DHs can't, obviously, receive those points.

In the case of Canseco, limiting the population according to AL teams, rather than all MLB teams, would remove 5 top seasons at position for him (1994, 1995, 1999, 2000 and 2001, all of which he ranked either 4th or 5th) and 5 top offensive seasons at position (same years), for a total of 50 less points. That would put him well off-ballot at around 40th (and about 250th all time). That's still higher than something like rWAR would rank him, but is probably more reasonable. Rice, on the other hand, wouldn't lose any points, since in 1977 he was the best DH in MLB, and that was the only year he played more DH than another position till 1988-89, when he wasn't good enough to qualify even with too large of a DH population.

So I'll make the correction and take another look at DHs before contemplating any further changes. Thanks for getting me to take a closer look at them.

Luis Tiant--I don't see Tiant on here anywhere.

Yeah, I noticed that he's way down my list too. He's at 139 points, which is 418th all time. I planned to take a closer look at him, given his placement on the last HOM results. I should have put another item on my to-do list in #39, which is look at top 15 (maybe top 30?) players on last year's results which aren't in my top 100 to see why, but I forgot to include it. Glancing at Tiant, he just doesn't rank well compared to his peers. (And pitchers do have a larger population than position players to determine how many should be credited with being top 20% at their position.) He gets 65 points for a 13 year prime, 36 points for being top 20% at position 6 times (1968, 1972, 1973, 1974, 1976 and 1978), another 30 points for being a top pitcher (just considering pitching performance, not a player's all around value at position) 6 times (same years), 8 points for twice being a top 20% offensive player at his position (1967 and 1969), and that's it. He was never best at his position, or within the top 2% of all MLB players, which hurts him.

So, rWAR sees him as the 13th best pitcher 1960-1990, right behind Jim Palmer and ahead of Tommy John and Jerry Koosman (heh, has he ever gotten a vote?). So to compare him to John, who I have ranked ahead of him (206 points), John gets 85 points for 17 prime years, 10 points for once being a top 2% player (8th best player of 1979), 12 points for one time being best at his position (1979 again), 24 points for being top 20% at his position 4 times (1970, 1977, 1979, and 1980), 20 points for being a top pitcher 4 times (same years), 4 points for being a top 4% fielder once (1969... rare event for a pitcher, which is why it's 4% of the number of position players rather than 2% of the number of the whole player population), 18 points for being the best fielder at position 3 times (1969, 1972, and 1973) and 33 points for being a top fielder at position an amazing 11 times (1967, 1968, 1969, 1972, 1973, 1977, 1980, 1981, 1982, 1983, and 1984). Interestingly, he doesn't get points for being the best pitcher in 1979, because while he was the best player at his position that year, it was due to his above average fielding performance lifting him over some other pitchers with better pitching performances but below average fielding. So while Tiant had more excellent years, John had a longer prime, a higher peak (if you consider 1 year a peak), and was a much better fielder. Most people seem to discount pitcher fielding out of hand, so I suspect that won't be a popular line of argument.

I'm also interested in the details of your system. Which stats did you use to rank players?

It depends on the category. Any category that measures overall value or compares pitchers and position players, I rely on 3 uber stats, Win Shares Above Bench, WAR (both from thebaseballgauge.com) and a rudimentary WAA system I put together from wOBA (minus ROE) for offense, Palmer's Fielding Run formulas for fielding, Base Stealing Runs (a linear weight formula) for baserunning, the difference between RA and league RA for pitching, and a positional adjustment based on the difference between league and position wOBA. (I'm impressed with DRA so far though, so I'm contemplating switching to using it instead of Fielding Runs, but I would need to decide what to do with pre-1893 players and pitchers. Either I'd have to keep using FR for them, or try to calculate their values myself. Hmm.)

I do like baseball gauge's metrics, but the real reason I'm using them is convenience. They let one download their full dataset, 1871-2010, in csv format, which means I could import it into my database and run custom queries against it. BR doesn't allow that. With BR's Play Index it's possible to answer the question "Who was the best 3B in 1971?" (at least according to rWAR), but not "Which 3B was best in MLB the most times?" (Eddie Mathews with 8 such seasons, according to my system), which I found frustrating. I would like to add more uber systems into the mix, but BG's were the only ones I could find that allow such a full download, and downloading and importing each year from 1871 on would drive me crazy. (DanR does make his WARP spreadsheet available, but with no pitchers or pre-1893 players, it's not useful for me to use it in a systematic fashion.) So, basically to determine the best and top players, and best and top players at position, I do something similar to the charts I posted on the 1961 MMP discussion. I rank each player based on the 3 stats, sum their total and then re-rank that number. A player had to have played at least 40% of the team median games for that season at a position to qualify for that position (only 15% for pitchers, due to the low game totals of modern starters).

For categories where I'm not looking at total values, I use a broader assortment of stats, but employ the same technique. For instance, to find the best offensive players each year, I rank all players on Extrapolated Runs, Runs Created, ISO, SECA, wOBA, OPS+, Equivalent Runs, RAA (the batting runs plus base stealing runs I calculate for WAA), Marginal Lineup Value, and offensive WAR and offensive WSAB (all but the last two calculated by me). I them sum the ranks and re-rank based on that figure.

Does being the best player in 1880 receive the same weight as being the best player in 2010?

Yes. There's an argument that that overvalues 19th century players, when it was easier to be the best player at a position than it is now. But, well, being the best in baseball or the best at a position is important enough (I think) that I wanted to give it a lot of weight. And modern players have more opportunity to gain points for being top players, etc. I suppose one alternative would be to increase the number of points assigned based on the number of teams. I did consider that, but my current list seemed like a good mix of old and new players, so I didn't want to weight it more heavily towards current ones.

Overall, though, I think this is a good start, even if I disagree with the particulars of the rating, and I'm glad that you're participating in both the HOM and MMP projects, Nate.

Thanks. I'm having fun here already, and I haven't even voted for anything yet.

And I think it's the same when adding position to the issue. I realize players at a position for a year is a smaller sample size than all hitters for a year, and thus more prone to lucky talent gluts or droughts, but, well, both are prone to those to some extent.

The problem is that players aren't accruing value against their positional peers--they're accruing value in relation to all other players in MLB that year. That's because positions can change, unlike eras--you can't move Pie Traynor to the 1970s, but you could definitely move him to 2B or SS or 1B or OF or whatever, and then he would be competing against a different pool of positional peers. Or you could move other guys to 3B--Mel Ott and Arky Vaughan, just to name two stars of the era, probably could have handled 3B without completely embarrassing themselves. So I would rather compare players to all MLB players, or at least compare them within "positional pools" as Tom Tango has proposed (IF, OF/1B, C), where skills are more easily transferred between positions.

Most people seem to discount pitcher fielding out of hand, so I suspect that won't be a popular line of argument.

I think that considering pitcher fielding is a valid approach, as long as you're not double-counting it or over-weighting it (i.e. considering other stats that already have pitcher fielding rolled in). What you're doing with John and Tiant seems to be a pretty reasonable approach.

I'm contemplating switching to using it instead of Fielding Runs, but I would need to decide what to do with pre-1893 players and pitchers. Either I'd have to keep using FR for them, or try to calculate their values myself.

BR has TZ available all the way back to 1871 (actually it looks like they're having some kind of database error, as all the fielding numbers pre-1950 are listed as 0, but you can still find TZ on www.baseballprojection.com).

15 Jack Fournier 1B 231

He may be due some MLE credit, according to his thread--he spent 3 years of what should have been his prime in the PCL.

The problem is that players aren't accruing value against their positional peers--they're accruing value in relation to all other players in MLB that year. That's because positions can change, unlike eras--you can't move Pie Traynor to the 1970s, but you could definitely move him to 2B or SS or 1B or OF or whatever, and then he would be competing against a different pool of positional peers.

That's certainly a valid point. But while players can move, we can't know how'd they perform in the new position until they actually make the move. Players generally perform better in the field when moving leftward on the defensive spectrum, but not always. Alex Rodriguez, for example, seems to be a player who was a better defender at SS than 3B, even though 3B is the easier position to play for most players. So I'd prefer to compare players at what I think is the smallest meaningful level of granularity, their position.

He may be due some MLE credit, according to his thread--he spent 3 years of what should have been his prime in the PCL.

Thanks. I'll take a look at his thread.

I'll just say, Nate, that I very very strongly disagree with your approach. Being the best 1B in 1935/CF in 1956/SS in 2000 is not the same as being the best 3B in 1925/1B in 1955.

Let's just return to first principles here. Teams' job is to win pennants. The ingredients of pennants are wins. The number of wins needed to win a pennant is determined by the standard deviation of team wins (in a high-stdev league, you might need 100 W; in a low-stdev league 93 might do it).

Holding the distribution of talent among teams constant, the standard deviation of team wins is in turn the product of the standard deviation of overall aggregate player performance above positional average. The spread of team records is the exactly same if every position is distributed the same, or if all the 3B are exactly average and the CF are spread out twice as much as normal. Thus I don't see any any valid empirical basis for using only the standard deviation at a player's own position to adjust his wins, rather than the leaguewide standard deviation. A team in 1927 with the best 3B in baseball and a league-average RF would obviously not be as good as a team with a league-average 3B and the best RF.

I am a little concerned that you could be using the same standards for pitchers as for other positions. Teams don't play just one pitcher like they play just one 1B. The pool of pitchers should be approx 5 times larger for baseball post 1950.

Your 2nd ranked modern pitcher is Tommy John and he ranks behind Hal McRae. That doesn't seem right at all. A GM would have been smart to have traded Hal McRae's career for Tommy John's career.

It seems to me we compare a player to his position all the time in a way that effects his wins when we evaluate fielding. The most common way to determine a player's fielding worth is to determine how many plays he made, determine how many plays an average fielder at the position would have made that year in the same playing time, and then convert the net plays into runs. Pretty much all the fielding systems do this (except for fielding win shares, but that has its own set of problems), even though they differ on the particulars of how to implement that method. If a player is at a position that is experiencing a star fielding glut, his net runs will be lower than what he'd gain with the same true talent during a different era, and (probably) lower than what he'd gain at a position that doesn't have a glut.

An example Humphreys' points out in his book (on my mind since I was reading it yesterday) is Duke Snider. Snider was probably a below average CF just on talent alone, but his net play numbers are made worse by the fact that 1/4th of the players at his position in his league were historically great fielders (Ashburn and Mays), thus raising the average level of play. That, in turn, effects the number of wins we estimate he provided to his team.

I'm obviously using a different method and applying it to a player's total worth, not just his fielding worth, but like fielding worth how I rank players is effected by gluts and droughts. And I'm OK with that.

Being the best 1B in 1935/CF in 1956/SS in 2000 is not the same as being the best 3B in 1925/1B in 1955.

True.

However, the worth of the positions are, more or less, equal to each other throughout baseball history (excepting P, which is its own case, and maybe C). That's the entire basis of the defensive spectrum. So if one position is significantly out performing another in total worth over a year or several years, why is that? Coincidence? Playing conditions that make it easer to excel at the seemingly stronger position? Mangerial incompetence in who is played at the seemingly weaker position? It could be any of those or more. I think it's impossible to tell with complete certainity, and would obviously vary from case to case. So, in ranking players at their positions, either one can weight the rankings based on how the positions performed that year, or one can hold the worth of the positions steady. If I had perfect knowledge I might choose to give positions more weight when they performed above the average position for that year if that was based on a coincidental concentration of talent, but not when they benefited from playing conditions favoring their position. But since I don't, I've chosen to hold the worth of the positions steady.

Thus I don't see any any valid empirical basis for using only the standard deviation at a player's own position to adjust his wins, rather than the leaguewide standard deviation. A team in 1927 with the best 3B in baseball and a league-average RF would obviously not be as good as a team with a league-average 3B and the best RF.

Has there ever been a year where a team would be better off with the best C in baseball and an average 1B rather than vice versa? I ask because plenty of people here give Cs a bonus, but, well, if you're using a strict "how many wins did he provide to his team?" criteria, what's the empirical basis for that bonus? It seems to me that while C may be as important as the other positions (though there is an argument to be made that it isn't, if one doesn't think C fielding is all that valuable), individual catchers are almost invariably less valuable, on average, then other players, on a wins per season basis. Gary Carter, for example, probably accumulated enough career value to be in the HOM regardless, but without a bonus it's hard to argue for him on a (especially) peak or prime basis. Just looking at his raw rWAR values, he only had one season (1982) that was close to MVP quality, and even that fell short. And he's a consensus top 5 catcher. And sure, it's due to the demands of the position that catchers fall short in wins, but so what? Either a player provided his team with wins or he didn't. But the catcher position is part of the game, and people want to honor it, regardless if catchers aren't as important as other players.

The same thing with RPs. Mariano Rivera is creeping up on #200 all-time on the rWAR list, so he has a decent argument for the bottom quarter of the HOM for career voters who use that stat. But for peak voters there's probably 500 guys who have a better case, if you're soley looking at wins. And this is the best RP of all time. But people consider RP an important part of the game and want to honor those who excelled at it, even if they didn't help their teams win as many pennets as the best RFs or whatever.

Point being, voters consider things beyond wins all the time.

I am a little concerned that you could be using the same standards for pitchers as for other positions. Teams don't play just one pitcher like they play just one 1B. The pool of pitchers should be approx 5 times larger for baseball post 1950.

I am using a larger pool of players for pitchers than for other positions. I'm not at home, but I believe I start off using 2x the normal position population in the 19th century and scale up to 5x in the modern era. I'll post the exact years I bump the population up at when I get a chance. They were mostly guesses, so could probably be improved.

Your 2nd ranked modern pitcher is Tommy John and he ranks behind Hal McRae. That doesn't seem right at all. A GM would have been smart to have traded Hal McRae's career for Tommy John's career.

McRae may be benefiting from the the problem I had overrating DHs (by overestimating their population) that I mention in #46. I haven't corrected the problem yet, or looked at McRae's specific case, so I'm not sure, but I'll take a look later.

I do have some modern pitchers ranked very high though, just not these ones. I believe both Maddux and Clemens are in my top 25 all time.

> scale up to 5x in the modern era

We're closer to 5.5 or 6 at the current time with 12 man pitching rosters.

OK, so here are the multipliers:

1871 - 1892: x2
1893 - 1903: x3
1904 - 1974: x4
1975 - 2010: x5

We're closer to 5.5 or 6 at the current time with 12 man pitching rosters.

I suppose. I pretty much ignored RPs, and just based the population on my guesses for how many SPs were used regularly per team, as so few RPs come close to being worthy.

As for McRae, it looks like when I make my correction to DH population he'll lose 26 points, which would put him below John.

And here's the top 100 or so again, with the DH population reduced:

1 Bobby Bonds RF 277
2 Pie Traynor 3B 270
3 Rafael Palmeiro 1B 266
4 Dale Murphy CF 258
5 Ed Konetchy 1B 252
6 Jim Whitney P 250
7 Tommy Bond P 249
8 Jack Clark RF 246
9 Jim Rice LF 243
10 Frank Chance 1B 239
11 Tommy Leach OF 235
12 Dave Bancroft SS 233
13 Jack Fournier 1B 231
13 Fred McGriff 1B 231
15 Rick Reuschel P 228
16 Buddy Bell 3B 224
16 Dave Concepcion SS 224
18 Bill Mazeroski 2B 221
19 Cesar Cedeno CF 219
19 Fred Dunlap 2B 219
19 George Foster LF 219
22 Rocky Colavito RF 217
23 Bucky Walters P 216
24 Fred Lynn CF 215
25 Albert Belle LF 211
26 Bill Bradley 3B 209
27 Carl Mays P 207
27 Kirby Puckett CF 207
29 Bob Elliott 3B 206
29 Tommy John P 206
31 Dwight Gooden P 203
32 Jesse Barfield RF 200
32 Harlond Clift 3B 200
32 Gene Tenace C 200
32 Bernie Williams CF 200
36 Denny Lyons 3B 199
37 Jose Canseco DH 196
37 Al Rosen 3B 196
39 Hal McRae DH 194
39 Ned Williamson 3B 194
41 Davy Force SS 193
41 Bob Johnson OF 193
43 Wally Schang C 192
44 Norm Cash 1B 191
45 Orel Hershiser P 190
46 Vern Stephens SS 189
47 Bobby Bonilla 3B 188
47 Ron Guidry P 188
47 Tim Salmon RF 188
47 Dizzy Trout P 188
51 Toby Harrah 3B 185
52 Cy Seymour OF 184
53 Greg Luzinski LF 183
54 George Uhle P 182
54 Robin Ventura 3B 182
56 Larry Doyle 2B 180
56 Amos Otis CF 180
58 Burleigh Grimes P 179
58 Camilo Pascual P 179
60 Orlando Cepeda 1B 178
61 John Olerud 1B 177
62 Willie Kamm 3B 176
62 Dave Parker RF 176
64 Jack Clements C 174
64 Pedro Guerrero 1B 174
64 Joe Tinker SS 174
67 Ken Caminiti 3B 173
67 Gil Hodges 1B 173
69 Hugh Duffy OF 172
69 Ken Singleton RF 172
71 Don Baylor DH 171
71 Tony Fernandez SS 171
71 Jimmy Ryan OF 171
74 David Cone P 170
74 Eric Davis CF 170
74 Tony Perez 1B 170
74 Fernando Valenzuela P 170
78 Mike Griffin OF 169
78 Gil McDougald 2B 169
78 Don Newcombe P 169
81 Dennis Martinez P 168
81 Darryl Strawberry RF 168
83 Travis Jackson SS 166
84 Johnny Pesky SS 165
85 Don Mattingly 1B 164
86 Johnny Callison RF 163
86 Dom DiMaggio OF 163
86 Jim Fregosi SS 163
86 Rabbit Maranville SS 163
90 Schoolboy Rowe P 162
90 Mickey Tettleton C 162
90 Curt Welch OF 162
90 Tom York OF 162
94 Ron Cey 3B 161
94 Arlie Latham 3B 161
96 Phil Rizzuto SS 160
96 Hack Wilson OF 160
98 Tony Lazzeri 2B 159
98 Red Lucas P 159
98 Jim McCormick P 159
98 Steve Rogers P 159

Canseco, McRae, etc. drop. Palmeiro loses 10 points as he no longer gets any points for 2003, when he was the 6th best DH.

Having said all of that, my system is ranking pitchers lower than most other voters probably would. As way of comparison, here's the pitchers it would put in the top 100 all-time:

12 Cy Young P 665
19 Greg Maddux P 548
22 Roger Clemens P 533
25 Walter Johnson P 516
28 Pete Alexander P 482
33 Tom Seaver P 427
34 Christy Mathewson P 424
39 Al Spalding P 402
45 Kid Nichols P 387
66 Warren Spahn P 340
67 Lefty Grove P 339
68 Steve Carlton P 337
74 Bob Gibson P 333
83 Pedro Martinez P 316
88 Carl Hubbell P 308
90 Randy Johnson P 302
99 Tom Glavine P 290

(The number on the left is the all-time rank.) That's 17/100. I'm not sure what the number ideally should be, but that seems a bit low. Pitcher value compared to the other positions is difficult to determine though. While I think the other positions (with the possible exception of C) are equal in value to each other in the long-term, I'm not sure the same is true of Ps.

(Maddux over Clemens is probably going to be seen as an odd choice, so I'll just say right now it's based on the points he gets for fielding.)

I would say between 27% (3 of 11) and 33% pitchers is appropriate depending on the era. Walter Johnson at 25th is way too low. I have him at #3.

I have many posts outlining why I think the HoM has been too tough on pitchers and why we're currently underrepresenting pitchers (26%). Ideally this election would induct two or three pitchers to get the number back up.

How is your system handling war credit? I ask because Rizzuto is way down at the bottom and you have Glavine ahead of Bob Feller.

Ben Taylor should bubble sort to the top of your Konetchy / Chance / Fournier group. There are several years where he is the best 1B and not them. That should bump them down and put him in your top 15. Likewise Traynor loses in comparison to Jud Wilson.

Edit: When are you going to integrate the ballot with Negro League ballplayers? It is especially odd not to see any pitchers in your top 100 players or anyone in your consideration set.

I would say between 27% (3 of 11) and 33% pitchers is appropriate depending on the era.

Checking a larger sample, my top 250 is somewhat better, as it has 55 pitchers in it. So substituting that for a pHOM (which I don't have), that would be 22% pitchers I would have voted for using this system without any subjective adjustments (of course, some of the guys in the top 250 aren't eligible yet, but the same is true of batters). That's still probably a bit low, so I may tweak the points I'm assigning to each category to bump pitchers up a bit. I'll have to think about it some more.

Walter Johnson at 25th is way too low. I have him at #3.

Johnson is underrated compared to Young due to the (over) dominance of 19th century pitchers in the uber stats. That's something I'd have to correct manually (and I probably won't bother for players who are already in the HOM). Of course, that would lower Young but not raise Johnson. That could be seen as more evidence that I'm underrating pitchers in general a little bit. Of course, if I raise the points I'm giving them, Bond and Whitney might very well end up at the top of my eligibles list. But I'm probably going to give them a subjective downward adjustment when I get around to evaluating my top players anyway.

How is your system handling war credit? I ask because Rizzuto is way down at the bottom and you have Glavine ahead of Bob Feller.

It's not, so far. I have to go through the top eligibles and determine whether they deserve any credit for play outside MLB (MiL, play in the military, etc.). It's on my to-do list.

Ben Taylor should bubble sort to the top of your Konetchy / Chance / Fournier group. There are several years where he is the best 1B and not them. That should bump them down and put him in your top 15. Likewise Traynor loses in comparison to Jud Wilson.

Edit: When are you going to integrate the ballot with Negro League ballplayers? It is especially odd not to see any pitchers in your top 100 players or anyone in your consideration set.

Going through the NgL players is tops on my to-do list, once I finish with these system tweaks. I probably won't complete integrate the populations (that is, Ben Taylor might get credit for being the best 1B several years, but he won't take points away from Chance). See the 2nd half of post #45 for a (long, rambling) reason why.

Nate -

Just a few notes, albeit from someone who does not always turn in a ballot close to consensus:

I found Wikipedia's article on Dick Redding to be very helpful, because it printed a lot of different sources' stat lines, and shows how wildly they disagree. There are very strong Redding supporters here and people like me, who pretty much have written him off. I imagine it depends on whose stat lines you use.

Chris Cobb has done a LOT of work on Negro League MLEs. He participates here, and is the best NgL resource I know of. Listen to him.

Paul Wendt does not vote here, but does comment. He is, as far as I can tell, the strongest expert we have on 19th century baseball. Listen to him, too.

Ed Konetchy is a very good pick for best deadball era 1B, if you don't include NgLers or George Sisler. Good call by your system.

Traynor really only had the huge rep in the 1940s-60s. Before then, it was usually Jimmy Collins, who invented a lot of what we know as modern third base play, sometimes Home Run Baker (who is the correct answer) as to who was the best 20th century 3B before Mathews. Heine Groh also comes to mind. A lot of Traynor's rep came from batting in a RBI spot in a Pirates lineup that featured the Waner brothers and Glenn Wright in front of him. Defensively, he was probably a very good major league shortstop, maybe a great one, who was blocked by Wright when he came up, and ended up at third. Modern opinions of his defense, as you have already found out, vary wildly.

Don't forget to include hitting value when looking at early pitchers, especially 19th century ones. Also, you will have to make a decision as to how much deduction to make for the American Association years compared to the NL, and each year needs its own deduction.

When I first got involved in this project, which is only 3 or 4 years ago, the best idea I probably had was to go read the discussion and ballot threads for the PREVIOUS year's ballot. It gave me a lot of perspective as to who was looking good to the consensus and why. I disagree with the consensus a lot, but it's a very good place to start, and, frankly, probably still comes up with just as good results as I do, if not better. I always feel queasy when I disagree with the consensus.

DL is right. You have to come up with a way to deal with war credit for WWI, WWII, and Korea, and also some way to deal with minor league credit. The consensus here seems to be to treat war years as full credit, trying your best to estimate what the player would have done if he had played. There is a lot more diversity of opinion on minor league credit.

If you really want a hard test case to check your system with, I suggest Don Newcombe. He has everything from NgL credit to war credit to being the best-hitting pitcher of his time, to all sort of things. He's the hardest adjustment case I've ever seen.

Don't try to do everything in one ballot year. The workload is impossible, which is why it's a good idea to look at the threads from last year. When I started here, I had already done this sort of project twice before, and I'm still finding new guys to check out or recheck (Vic Willis currently, along with a long comment that should be titled "Will White, Deacon Phillippe, Sam Leever, Kevin Brown, and David Cone).

- Brock Hanke

I agree - don't try to do it all. Conversely, make sure you've covered the top 20 returning players.

Thanks for the advice. I appreciate it.

I found Wikipedia's article on Dick Redding to be very helpful, because it printed a lot of different sources' stat lines, and shows how wildly they disagree. There are very strong Redding supporters here and people like me, who pretty much have written him off. I imagine it depends on whose stat lines you use.

I haven't taken an in-depth look at him yet, but he doesn seem like a tough guy to get a handle on, given that half his career took place before the organized NgL formed. I'll take a look at his wiki page.

Don't forget to include hitting value when looking at early pitchers, especially 19th century ones.

That's definitely something I'm considering. For instance, Rynie Wolters is my best player for 1871, and I think it's pretty legit, unlike the normal 19th century pitcher overdominance of the uber stats due to their superficially impressive IP totals, because while his ERA+ wasn't impressive, it was above average, he lead the NA in IP, and he was something like the 4th or 5th best offensive player that year while playing almost every game. Even if pitching was less valuable back then on a per inning basis, when someone is a top offensive player and contributes a bunch of slightly above average innings, that's a very valuable combo.

(Which reminds me, this is probably a long-shot, but anyone know anything about his pre-NA play? He has no thread here, and has only been mentioned a few times as far as I can tell. 1871 was his last (only?) good year, but wiki lists him as playing 6 years prior to 1871 in the NABBP, without providing any details. With only one good year in MLB he's not close to my top 100, but if he had been able to put up similar numbers in the NABBP for 2-3 years, he could probably get that high.)

DL is right. You have to come up with a way to deal with war credit for WWI, WWII, and Korea, and also some way to deal with minor league credit. The consensus here seems to be to treat war years as full credit, trying your best to estimate what the player would have done if he had played. There is a lot more diversity of opinion on minor league credit.

I don't plan on giving war credit, but I will credit players for play in the military. I read through a lot of the previous threads, so I know that's a minority (and controversial) position, but there have been voters in the past who felt the same way.

If a player is at a position that is experiencing a star fielding glut, his net runs will be lower than what he'd gain with the same true talent during a different era, and (probably) lower than what he'd gain at a position that doesn't have a glut.

No, that's not right at all. His fielding wins above average will be lower, but so will the fielding wins above average of the replacement player he is compared to. There should thus be no effect whatsoever on his WAR.

However, the worth of the positions are, more or less, equal to each other throughout baseball history (excepting P, which is its own case, and maybe C). That's the entire basis of the defensive spectrum.

Is the aggregate WAR of the other 7 positions in baseball history really roughly the same? I'd find that very surprising. 3B in particular I would think would be way behind.

And that most certainly is not the "entire basis of the defensive spectrum." The basis of the defensive spectrum is that every team has to put a player at every position, so guys should be compared to the freely available talent level at their own position. It would be wrong to count Traynor's wins above a replacement 1B, because you can't replace Traynor with a replacement 1B.

But that's not what you're doing. If Traynor is 5 wins above the replacement 3B and Gehrig is 10 wins above the replacement 1B, you're arguing that Traynor is still equal to Gehrig, just because there doesn't happen to be a Gehrig-caliber player at Traynor's position. That doesn't pass the smell test.

The difference between 1920s 3B, say, 1970s SS is that while 1970s SS hit badly as a whole, there were huge gaps between the best ones (Concepción, Campaneris, Harrah) and the worst (Rob Picciolo) that won real games for their teams. By contrast, there were plenty of guys floating around the fringes of MLB who could hit a bit and play a decent 3B in 1925 (e.g. Howard Freigau). Again, you want to adjust for the positional replacement level but the leaguewide standard deviation. Using within-position standard deviations is saying that a win at one position isn't worth the same as a win at another. In that case, why are we bothering to measure value at all?

Has there ever been a year where a team would be better off with the best C in baseball and an average 1B rather than vice versa?

Sure, much of the 1950s, no?

Your point on C and RP is well-taken. The difference is that C and RP have real constraints placed on their value by playing time restrictions. No such limit exists for 1920s 3B or 1950s 1B.

Defensively, he was probably a very good major league shortstop, maybe a great one, who was blocked by Wright when he came up, and ended up at third.

Minor nitpick - Wright didn't come up until 1924, by which time Traynor was already solidly at third. Rabbit Maranville (an all-time defensive great) was Pittsburgh's SS when Traynor came up.

-- MWE

Mike -

Thanks for the nitpick. I haven't looked that one up in a long time because I "knew" it was Wright. Actually, Maranville makes more sense; Traynor was probably a better defender than Wright was.

Thanks again, - Brock

No, that's not right at all. His fielding wins above average will be lower, but so will the fielding wins above average of the replacement player he is compared to. There should thus be no effect whatsoever on his WAR.

It depends on how one calculates WAR and if the runs between an average player and a replacement player are held constant or not, and if not, then how the difference is calculated. For rWAR, I believe the difference is held relatively, though not completely, constant.

Here's some players that had 674 PA in a year and how many runs they accrued due to the difference between an average and replacement player, according to rWAR:



It's not completely steady, as a league strength bonus is factored in, according to the description of Rrep (the idea being that if a league is stronger, then replacement players in the league played better), but it's pretty close. Certainly closer than empirically setting the difference per position would generate, I believe. So the defensive runs above average of a replacement player isn't even considered in this method.

So my point stands, for some widely used versions of WAR. I imagine you'd say this is why the above method shouldn't be used. But that's irrelevant, I wasn't talking about how player value should be determined, but how it often is.

And even if it hadn't, fielding runs/wins above average are often used to make comparisons about players' fielding worth, without reference to replacement level. I've seen it done recently on this site (and not just by me). So, again, how other players play at a player's position influences how we determine their value, depending on how one defines value, and what methods one uses.

Is the aggregate WAR of the other 7 positions in baseball history really roughly the same? I'd find that very surprising. 3B in particular I would think would be way behind.

I don't know. I was speaking axiomatically, not empirically. But it's an interesting question, so let's check. Here's the totals, by position, of Baseball Gauge's WAR (it's much easier for me to do this for BG WAR than rWAR, since I have the former fully imported into a relational database). I have the year by year figures, but it's too much to paste here, so I'll just do the totals. Outfielders are lumped together as OF till 1954, DH starts in 1973.



Obviously there's some caveats with using this data. First, I assigned every player to a single position per year, rather than prorating their WAR by games played in the field. I think that evens out over the long term, except in the case of DH, as the relatively low number of full-time DHs means DH comes in lower, compared to the other positions. Secondly, I haven't made any attempt to correct for player population, which makes it difficult to compare P and DH (even beyond the problems of the first issue) to the other positions. And finally, not having individual OF positions prior to 1954 makes it a little harder to compare OFs to IFs.

Having said all that, 3B holds its own quite well. It's consistently ahead of 2B until I cut the range down to the most recent period, when they're just about dead even. They're both well ahead of SS, which consistently lags behind. Now, I figure you'll say that's because BG WAR is flawed, as it doesn't use empirical positional replacement levels (instead replacement level is defined as 67 wins per 162 games, and a replacement player is one which, if a team had solely such players, they would win at replacement level), and if I did the same exercise with your WARP the results would be different. But my thinking is, if one position lags behind another in a version of WAR/WARP then how do we know it's a real gap in value, and not an artifact of the system? You might have complete faith in your system, but I, on the other hand, don't. So I'd rather hold that, as a matter of a priori knowledge, all positions are equal in the long term (excepting P, but including C, even though I'm skeptical of its real value).

And that most certainly is not the "entire basis of the defensive spectrum." The basis of the defensive spectrum is that every team has to put a player at every position, so guys should be compared to the freely available talent level at their own position.

I disagree with your characterization of the defensive spectrum. The defensive spectrum holds that, if a position, as a whole, is worth less on offense than another position, then it must be worth more on defense. The implication here is that, the total worth of the positions' are equal, or close enough to make no difference. This is an axiom, based on assumptions on why managers play the players that they do, as it can't be empirically demonstrated.

But that's not what you're doing. If Traynor is 5 wins above the replacement 3B and Gehrig is 10 wins above the replacement 1B, you're arguing that Traynor is still equal to Gehrig, just because there doesn't happen to be a Gehrig-caliber player at Traynor's position. That doesn't pass the smell test.

No, I'm not arguing that Traynor and Gehrig were equal in 1927 (a year in which I have them both as the best at their positions). I'm arguing that they were equal in one aspect of what I wish to measure (that is, they were both the best at their position). Positional dominance is just one aspect of my system, though a very important one. As I said, I also give points for being the best player in MLB, within the top 2% of players, etc. In the end, Gehrig ends up doing much better than Traynor over the length of their careers, even given his lead in games played. I have him as the 17th best player ever (though I'm still tweaking the weights for the categories to get a better mix of position players and pitchers, so that might change slightly), while Traynor is much lower down.

The difference between 1920s 3B, say, 1970s SS is that while 1970s SS hit badly as a whole, there were huge gaps between the best ones (Concepción, Campaneris, Harrah) and the worst (Rob Picciolo) that won real games for their teams. By contrast, there were plenty of guys floating around the fringes of MLB who could hit a bit and play a decent 3B in 1925 (e.g. Howard Freigau).

I'll note that while your WARP loves Concepcion, there's not exactly a consensus for him over Traynor.



(BP isn't currently showing WARP values for any players prior to 1950, for whatever reason, so I didn't include it.) So, in spite of Concepcion's longer seasons, more years played, and TZ (probably) underrating Traynor's defense, he trails in 3 of the 5 systems. If you substitute DRA for TZ in rWAR and fWAR, and adjust for the shorter seasons in Traynor's time, I'm sure he'd pass Concepcion in fWAR, making it only your system that sees him as a better player.

Sure, much of the 1950s, no?

Possibly. I didn't look. The point being, even if there are years where the best C is better than the best 1B, it's a very distinct minority of them. But most voters want to honor a comparable number of Cs, if not exactly an equal amount.

It's not completely steady, as a league strength bonus is factored in, according to the description of Rrep (the idea being that if a league is stronger, then replacement players in the league played better), but it's pretty close.

I messed this up. The idea behind the league strength bonus is that if a league is stronger, then the gap between average and replacement player is bigger, giving the player more Rrep runs (what I described would have had the opposite effect). That's what I get for posting at 2 in the morning. Nevertheless, my point that the difference between average and replacement player is relatively steady still holds.

The following list is NOT A DRAFT BALLOT

What it is, is a [still] preliminary working out of a method I've been working on for quite some time (including on the 2011 ballot). The numbers represent runs above average, - if a player was above average for a year that year counts- if not I throw the year away. That said, I've given no one minor league credit, and only 1 player WWII credit (Rizzuto- I just haven't gotten around to others yet), and I include only MLB stats and haven't made any league quality adjustments. I did some tweaking with regard to pre- 60'6" pitching stats, and here goes:

1 Phil Rizzuto 335
2 David Cone 332.4
3 Tony Mullane 331.1
4 Travis Jackson 322.7
5 Frank Chance 321.5
6 Mike Tiernan 319.9
7 Rick Reuschel 316.7
8 Dave Bancroft 316.3
9 Babe Adams 314.7
10 Carl Mays 310.8
11 Bob Johnson 307.7
12 Tommy Bridges 306.8
13 Urban Shocker 306.6
14 Luis Tiant 306.2
15 Denny Lyons 305.7
16 Rafael Palmeiro 305.4
17 Bill Joyce 304.3
18 Bert Campaneris 300.8
19 Kevin Appier 299.7
20 Wilbur Cooper 297.6
21 Burleigh Grimes 295.9
22 Jim Fregosi 294.8
23 Vic Willis 294.3
24 Sal Bando 292.7
25 Luis Aparicio 289.1

Phil shoots to the top via WAR credit, without it he's not in the top 25, as it is there is a long list of not yet eligibles above him. Not far from his 2011 ballot finish.

David Cone... no explanation necessary...

Mullane- OTOH very high WAR #s for someone not in the HOM, OTOH, may have to keep tweaking those pre 20th century pitching numbers.

Travis Jackson- I'm not sure what to make of this- basically had no 2-3 WAR seasons padding his WAR totals (which give no value in my system) No votes in 2011- but is in the HOF

Frank Chance - He is someone who I do think the HOM has been underrating- career shape similar to Travis'
Tiernan- short career, high rates...

Big Daddy... pretty much where HOM voters seem to see him

Bancroft- the big IF is his defense (I'm working off BBREF WAR which has no fielding #s for him)- if he was an above average defensive player he just may be HOM worthy- just average- he's at the high end of HOVG- and if he really was the "Ozzie Smith" of his day, which has been alleged (by someone I suspect did not see Bancroft play) then he really should have been in awhile ago

Babe Adams- just looks too high to me, gotta look at it
Carl Mays- ditto

Dropping down some- Raffy- my system doesn't like him- but it really hates Dave Winfield (though to be fair to my system- it's my including BBREF fielding runs that really kills him)

Give Tommy Bridges and Urban Shocker war credit and they'll move up. What you have is a high-replacement career system. It is going to give some goofy results if you don't moderate it with a true-replacement career system. My spreadsheet is a hybrid of a wins above positional average and wins above positional replacement.

It is going to give some goofy results if you don't moderate it with a true-replacement career system.

I did that deliberately- If someone had 60 WAR in 7500 PAs and someone else had 60 WAR in 9000 PAs I want to elevate the 1st guy over the second.

More importantly let's say someone had 60 WAR in 9000 PAs Since about 2 WAR per 600 PAs is simply getting from replacement to average, that guy is only about 30 wins above average- and 60 WAR seems to be a pretty strong HOM indicator to date. Let's take someone with 50 WAR in 9000 PAs - that's a guy who is 20 wins above average- its' also someone not likely to get much HOM support

Now let's take someone who has 50 WAR- in 6000 PAs- that guy is 30 wins over average- IMHO he's as good (or should be as good) a HOM candidate as the guy with 60 WAR in 9000 PAs

Give Tommy Bridges and Urban Shocker war credit and they'll move up.

Bridges definately- about 2 years?
Shocker? 1/2 to 3/4 a year? Some I guess.

Mullane- OTOH very high WAR #s for someone not in the HOM, OTOH, may have to keep tweaking those pre 20th century pitching numbers.

Mullane also had a number of peak years in the early, very weak American Association, 1882-1884. BBREF WAR applies a blanket adjustment to the AA, that does not factor in how much weaker the earlier AA was from the later AA. I originally had Mullane up near the top of my consideration set, but after dinging him for the AA, he fell way down out of the top 50.

Bancroft- the big IF is his defense (I'm working off BBREF WAR which has no fielding #s for him)- if he was an above average defensive player he just may be HOM worthy- just average- he's at the high end of HOVG- and if he really was the "Ozzie Smith" of his day, which has been alleged (by someone I suspect did not see Bancroft play) then he really should have been in awhile ago

Bringing up DRA (and Humphreys' book Wizardry) again, as I think it's the best pre-Retrosheet, publically available, defensive metric out there (which is kinda damming with faint praise, but...), it likes his defense, but as far as being an all-time great? Well, maybe, depending on how you look at it. It has him at +145 defensive runs over his career. That's tops of the Live Ball Era (1920 to 1946), over Appling (+141), Boudreau (+129) and Maranville (+114). He's sixth all-time since 1893 (after Tinker, Art Fletcher, Dahlen, Belanger, and Smith).

However, in the book, Humphreys' argues that the context of early (particularly Deadball, and, presumably, pre-1893, though he doesn't calculate figures for that time period) shortstops was quite different than today and that fielding variance was much bigger the earlier you go, giving the good to great shortstops much more opportunity to rack up defensive runs above average than would be available to a shortstop today, when the standard deviation for the amount of plays made by the average shortstop in a league is quite smaller. So he comes up with a second career ranking, TPAR (time pool adjusted runs), which is essentially applying a timeline to the career defensive runs list, to try and put modern and old school players on the same scale. Bancroft suffers quite a bit by this metric, as it's designed to bring the top Deadball era shortstops down to earth, and they had huge numbers compared to him (Tinker +283, Fletcher +275, Dahlen +254), so he falls to #22 (+56 TPAR). Finally, in Humphreys' more subjective top ten plus honorable mentions list (where he goes through each player's record, compares them to the other available metrics, like TZ, and tries to determine if DRA is really capturing the player's ability within the context of his time) Bancroft is nowhere to be found.

I understand Humphreys' reasons for constructing a timeline, but this, at least partially, gets back to the old ability vs value question. Yes, it's a bit weird that if you just take defensive runs above average, the 3 best defensive shortstops of all time played in Deadball or earlier (and by a huge margin, as no one else breaks +200, though Belanger gets close at +197), when most subjective (players are faster, bigger, etc) and objective (players make less errors, standard deviations on number of plays made are lower) measures tell us otherwise. But the flip side to that is, fielding was simply more valuable back then, both because of the bigger variance in defensive runs, and because more balls were put in play. So voters are going to come to different conclusions on how to deal with these issues. I try to put every player within the context of his time with my points system, but that's an unusual approach.

So, if you buy DRA at all, the range on Bancroft's fielding value was well above average to all-time great. Now, was he the Ozzie Smith of his day? Perhaps, but I think someone with a better claim to that title might be Tinker (even though my system likes Bancroft more). He hit almost as well as Bancroft and fielded significantly better than him. Though, admittedly, it was a different day, so they both could have been, I suppose.

Now let's take someone who has 50 WAR- in 6000 PAs- that guy is 30 wins over average- IMHO he's as good (or should be as good) a HOM candidate as the guy with 60 WAR in 9000 PAs

Not if the 60/9000 guy also had 50 in 6000. In that case he matched the first guy, then added on 10 more WAR as a slightly below average ballplayer. That makes him better in my eyes, he was able to provide value when the first guy was done. That's what I mean by moderating your results.

---

Bridges gets 2 years, Shocker gets about 1/2 a year in my spreadsheet.

Not if the 60/9000 guy also had 50 in 6000. In that case he matched the first guy, then added on 10 more WAR as a slightly below average ballplayer. That makes him better in my eyes, he was able to provide value when the first guy was done. That's what I mean by moderating your results.

My system throws out years where the guy was not above average, in which case if the 60/9000 guy was 50/6000 in his best seasons then my system would see him as a 50/6000 and not a 60/9000 guy.

he was able to provide value when the first guy was done.

10 war in 3000 PAs? I don't think thats "value" for HOM purposes.

10 war in 3000 PAs? I don't think thats "value" for HOM purposes.

It's Omar Vizquel's career post-Cleveland. Lots of players are able to hang around and help teams win at a reduced rate. I think that has more value than retirement. How about a pitcher who moves to the bullpen like Smoltz or Eckersley? They're no longer "above average" but they're providing value.

Do you throw out years where a player is exactly average? That doesn't make sense to me, average players have value. Even below average players have value. In your system does a player who is 1 win above average this year and 1 win below the next year have more value than a player who is exactly average 2 years?

Do you throw out years where a player is exactly average?

yes

average players have value

yes

Even below average players have value.

yes

In your system does a player who is 1 win above average this year and 1 win below the next year have more value than a player who is exactly average 2 years?

sort of, I'm not looking for "value" per se- I'm looking for greatness. I don't an average season should help push someone TOWARDS the hall

How about a pitcher who moves to the bullpen like Smoltz or Eckersley? They're no longer "above average" but they're providing value.

Terrible examples, both men certainly had above average seasons in the pen.

Based on WAR no reliever is "above average" for a pitcher. The top relievers only have seasons of 3 WAR. Are you considering "above average" as a rate stat? If so, players who are platooned will gain an advantage - see the last several years of Lou Whitaker's career. That doesn't make the player more talented, it is a mirage based on usage patterns. Measuring above replacement helps tease these things out.

I absolutely believe that comparing two players, one with 50 WAR in 6000 PA and one with 48 WAR in 6000 PA, the guy who puts up 5 WAR in the next 1000 PA should be ranked ahead of the guy who retires.

RE: Johnny, #67 -

"What it is, is a [still] preliminary working out of a method I've been working on for quite some time (including on the 2011 ballot). The numbers represent runs above average, - if a player was above average for a year that year counts- if not I throw the year away. That said, I've given no one minor league credit, and only 1 player WWII credit (Rizzuto- I just haven't gotten around to others yet), and I include only MLB stats and haven't made any league quality adjustments. "

If I understand this right, along with the other stuff you've written in your conversation with DL, what you seem to be doing is developing a method for ranking primes. Not peaks, since you include all above-average years. Not careers, since you discard below-average. Just primes, without any need to concentrate those prime years together (which most people insist on when ranking primes). This certainly is not a useless thing to do, but I wouldn't use it as an overall ballot method. I'm also not sure that this kind of method should be making tweaks to early pitchers. You're not counting up the number of years anyone needed to do what they did; you're just counting the total of above-average years. Early pitchers should have very high individual years, but not very many of them, so it should wash out - for primes. Peaks and careers respond differently to the early conditions, so you can't go by prime alone. - Brock Hanke

So, if you buy DRA at all, the range on Bancroft's fielding value was well above average to all-time great. Now, was he the Ozzie Smith of his day? Perhaps, but I think someone with a better claim to that title might be Tinker (even though my system likes Bancroft more). He hit almost as well as Bancroft and fielded significantly better than him. Though, admittedly, it was a different day, so they both could have been, I suppose.

I felt I should point out that the career figure (+283 defensive runs) I cited for Tinker probably undersells him a tiny bit, as I just noticed that no DRA amounts were computed for the Federal League (not sure why). For what it's worth, TZ sees him as +18 defensive runs above average over those two years (side-stepping the issue of whether FL fielding numbers should be discounted). DRA has him at +7 runs in 1913 (in limited playing time) and +0 runs in 1916 (in very limited playing time), so it's possible Tinker was still an above average defender during his defensive decline phase, and thus those years would have added, rather than subtracted, from his fielding value.

as a matter of a priori knowledge, all positions are equal in the long term

I see absolutely no reason whatsoever to assume this is true. Are placekickers equal to quarterbacks?

On Concepción, my system sees him as a better player because it's the only one that takes into account how hard it was to find a SS who could hit in the 1970's! That's, like, one of the principal findings of my method, and why many HoMies prefer it to other approaches.

I see absolutely no reason whatsoever to assume this is true. Are placekickers equal to quarterbacks?

It's certainly more likely to be roughly true when comparing Baseball position players than comparing football players

I see absolutely no reason whatsoever to assume this is true. Are placekickers equal to quarterbacks?

No. But football has a lot of structurial issues when it comes to its positions that simply aren't present in baseball, outside of the pitchers vs position players question, and you'll note that I consistently exempt pitchers when holding that all positons are equal in the long term, for that exact reason. A more apt analogy, I think, would be: are centers equal to point guards? Which... well, I don't know. That would require a lot of study to answer definatively. But it's not an outlandish starting place, given that they both participate in all aspects of the game (not true of quaterbacks and kickers).

As to why assume baseball positions (again, exempting P) are equal, well... all positions are on the field for the same amount of time (not true of individual catchers, but true of the position as a whole). All positions have the same ability to influence the game via offense. Not all positions have the same ability to influence the game via defense, but that's where the defensive spectrum comes in. Since the positions that have more defensive worth are harder to play, that reduces the population that can man them while still hitting acceptably, meaning those positions produce less offensive worth in the long term.

When comparing, say, SS to 1B, does this balance out? Does SS's increased defensive worth balance its lack of offense compared to 1B? Honestly, I don't know. How could we even tell? By calculating the wins each positions generates, I suppose. But then how do we measure wins? rWAR? fWAR? Your WARP? BP's system? And so on. Each uber system probably produces a different answer. In the absence of one system that I have rock solid faith in to accuratly measure value throughout baseball history, holding the positions to be equal is the only approach that makes sense to me.

On Concepción, my system sees him as a better player because it's the only one that takes into account how hard it was to find a SS who could hit in the 1970's! That's, like, one of the principal findings of my method, and why many HoMies prefer it to other approaches.

Sure. But I prefer to take a broader, consensus based approach to using uber-stats, rather than relying on just one, and there Concepcion suffers. If your system is ever finished (adding in pitchers and perferably pre-1893 players as well), then I'd be happy to toss it into the mix, but its current gaps make it impossible for me to use systematically.

I've seen comments about catcher not being worth as much because they don't have as much playing time. I'd posit that catcher is by far the most physically demanding position after pitcher. The catcher is involved in every pitch. The "degree of difficulty" based on hitting ability is one aspect to consider. The other aspect to consider is the physical wear caused by catching. It might take just as much athletic talent to play C as it does to play SS but SS doesn't require the same amount of physical exertion. This is why the positional bonus for catcher above and beyond the defensive spectrum makes sense.

"On Concepción, my system sees him as a better player because it's the only one that takes into account how hard it was to find a SS who could hit in the 1970's!"

This is true if general managers were making optimal decisions.

But what if they overreacted to the reasonable notion that artificial turf would place a premium on greater range, thereby overcorrecting by placing 5-9, 160-pound whippets (I remember the era well) who saved fewer runs than lesser fielders would have created offensively?

If that's the case, then Concepcion gets bonus pts for how dumb GMs were.

And Concepcion is on my 15-player voting list, so I'm not a hater.

This is quite well-worn territory. My system measures value in context, and makes no pretense to do anything else. We of course don't know just why so many major league teams put out banjo hitters at shortstop in those years. I will vehemently argue with anyone who suggests that Concepción or Campaneris or Harrah's teams didn't win tons of games by playing them there while their rivals trotted out Mario Mendoza. (And remember--the correct measure of value is standard deviation-adjusted wins above Mendoza/Picciolo--not wins above positional average or simple ordinal rank at position, which are polluted by star gluts or droughts.) But the question of how to assign credit for that value is up to each voter's subjective definition of Merit.

I've been procrastinating on a couple of research projects this summer by working on my 2012 preliminary ballot, so let me throw what I have out there. Comments to come in a week or so, but main changes are using Michael Humphreys' DRA as my defensive metric of choice (specifially, a version that detrends changes in variance of performance within positions over time), and making a revised effort to convert MLEs for Negro Leaguers and minor leaguers to WAR.

2012 Preliminary Ballot

1. Hilton Smith
2. Dick Redding
3. Frank Chance
4. Buddy Bell
5. Tommy Leach

6. Roy White
7. Bobby Veach
8. Rafael Palmeiro
9. Luke Easter
10. Bill Monroe

11. David Cone
12. Rick Reuschel
13. Buzz Arlett
14. Thurman Munson
15. Vic Willis

James, where do you have Gavvy Cravath? At first glance, Cravath and Arlett seem to be a pretty similar type of player (defensively-challenged slugging corner outfielders), but Arlett had just one season in the majors, while the majority of Cravath's career was in the big leagues. I haven't thoroughly looked at Arlett's candidacy, but right now I feel that Cravath is the better candidate because he is less reliant on MLEs.

How about Rizzuto? Lots of "credit" cases in your ballot so I'm pretty certain it isn't a war credit issue.

Nate the Neptunian,

Thanks for mentioning Wizardry and DRA. However, I would like to clarify comment #72. As shown in chapter three of Wizardry, DRA appears to be more accurate than the version of TotalZone that uses the Retrosheet data available from 1950 (now 1948) through 1988 and from 2000 through 2002. And modeling 1989-99 and 2003-10 is trivial, because we have exact counts of ground balls and fly balls (including line drives and pop ups) by opponent batter-handedness.

Since the publication of Wizardry, I have created a complete play-by-play database of Retrosheet data from 1950 through 2010. Applying the best Retrosheet play-by-play data available to the DRA methodology will result in the least biased defensive runs ratings for all periods of major league history, including the present.

Thanks for commenting. I really liked your book. My statement that I thought DRA is the best pre-Retrosheet defensive metric available wasn't meant to imply anything about its worth in the Retrosheet era, but I can see that it might be ambiguous.

As long as you're here... When you mention all periods of major league history, have you calculated any DRA figures for 1871 to 1892? I know you use 1893 as a cutoff in your book, but I was curious if you had run anything for 1871-1892, and had an opinion on how well the methodology works for capturing value in that time period. I have the Lahman data imported into a relational DB, and I was toying with the idea of running your pre-Retrosheet formulas for that time period, as it might be useful for this project, but it's a lot of work, and the regression math is a bit over my head.

Nate, it would really be necessary to run regressions on the 1871-92 data to see how well the whole approach works. I could guide you through how to do it in Excel without any VBA programming. I would expect the noise in the resulting model to be quite large, because the noise in the 1893-1900 model is much larger than in the 1901-1919 model, which is greater than for the 1920-1950 model, which is more than the post-1950 model.

Thanks for the offer. I've sent you an email about it via the console.

Time to bump this back up to the top. The election should begin in just over a month.

My prelim hasn't really changed. I spent some time looking at Ned Williamson and Pie Traynor and they moved up but not onto my ballot.

As an aside - I just signed onto Google+ and would like to have a "Hall of Merit" circle if anyone else is interested.

I pointed this out in a different thread, re Traynor:

Traynor played SS in Birmingham in 1920. According to Allan Barra's book on Rickwood Field, Traynor airmailed a lot of throws to first base, and one of the locals was quoted as saying his arm was "too strong" for SS. I don't think I have ever heard "that" particular comment before!

-- MWE

My apologies for not participating in the MMP project. 60+ hours of work a week, assisting my wife with school, and nursing her during a tough year of sickness has made it challenging to dedicate the proper time that a project like the MMP demands.

2012 Early Prelim:
1 Reuschel, Rick
2 Rizzuto, Phil
3 Palmeiro, Rafael
4 Cone, David
5 Newcombe, Don
6 Campaneris, Bert
7 Cravath, Gavvy
8 Pesky, Johnny
9 Shocker, Urban
10 Smith, Hilton
11 Williamson, Ed
12 Monroe, Bill
13 Appier, Kevin
14 Willis, Vic
15 Bell, Buddy

Guys I have interest in that could make the ballot with arguments from the electorate:
Perucho Cepeda
Doc Gooden
Omar Linares
Thurman Munson
Dick Redding

A fine discussion about DRA has raised my curiosity: I do not have access to Humphrey's figures...can anyone give an assessment of who the big winners and losers are compared to the competing defensive metrics. Are Humphrey's numbers available anywhere that I can access on the net?

Thanks and keep up the good work.

Perucho Cepeda

He'd be an interesting guy to post a thread on.

-- MWE

http://www.baseballthinkfactory.org/files/hall_of_merit/discussion/perucho_cepeda

which didn't come up when I searched for it. Thanks!!

-- MWE

Cepeda had about everything you could want in a candidate- except I don't think he's eligible by our rules. He needed to play part of his career in the US (like Mendez or Oms) in order to be eligible but he didn't. Kind of like the Japanese greats.


Here's the relative clause from the constitution (emphasis added):

All major league players are eligible for the Hall of Merit. Also eligible are all “excluded” players, most notably Negro Leaguers, and pre-MLB players that played professional ball in the US.

A quick search of Wikipedia says Perucho Cepeda played briefly for the New York Cubans.