Let's take this argument all the way to its most absurd extreme. Ron LeFlore. IIRC, LeFlore was recruited by Billy Martin while the would-be CF was in prison in Michigan. One might suggest that LeFlore's character prevented several would-be teams from winning games while he was in jail!

And was LeFlore connected to Tim Raines's cocaine usage in the early 1980s? (they were teammates around this time.) If so, would LeFlore be once again penalized?

And then is Tim Raines to be penalized for his coke usage since the Expos were a contending team during that period? It all spirals out from there....

TomH, I'm not really trying to pick on you, nor do I mean any disrespect because I'm mostly just kidding around for kidding's sake with my absurd example. But I do think that it points out that character is a really difficult issue to encapsulate, especially since a lot of guys' character seems to change as their life/career moves along. Then when you add in the difficulty with Negro League information for guys like Beckwith, it's downright Heinsenbergian.

What made Bill James famous is that he realized that major league baseball is a heck of a lot closer to tabletop baseball than most people had thought. What made him wise, however, is that he realized that--as close as it may be--major league baseball is not identical to table top baseball. Just as it is ignorant to think that "hustle" and "heart" make someone a good player, it is arrogant to think that personal characteristics and dynamics don't affect performance in ways that are hard to discern in the statistics. Intangibles aren't going to keep Rogers Hornsby out of the HoM or elect Enos Cabell, but in marginal cases like Frank Chance or Albert Belle they have to be part of a holistic assessment.

IMO, if a player is such a pain to be around that teams constantly get rid of him despite his production, then that is a factor in determining his value. There are some players who pollute the lockerroom (this is not just about bad shower shoes or "lucky" underwear, Mr. Grimsley).
For example... This is from another sport, but the San Diego Chargers could not wait to get rid of David Boston and a defensive lineman whose name I cannot remember (Coleman?? went to Dallas last year) because they were so bad for the team's attitude.
Maybe attitude means more in sports where players have to play as teams - football, hockey, basketball - but I think some players are so noxious they can affect a team's performance.

The trouble is recognizing the effect. The individual player's performance may be good so there is no effect there. Is it by constant changing of teams? Well Reggie Sanders changes teams every month, but he seems to have a good rep, so that factor is not always proof. Did his team underachieve? Lot's of teams underachieve. Do we use reporter's accounts? Then we let the views of only a few people who may have been jerks themselves influence us. And then there are the players who change, mellow, mature, etc. Are they downgraded for the immature part of their career?
Maybe it is a weight of the evidence thing. If there are so many stories from different sources (as opposed to several variations of the same story) that there has to be some basis in truth.

Just babbling to stay away from homework.

Denny McLain is probably the other side of the line.

McLain was conning his own teammates, so that's a cancer I would definitely want to rid my team of.

And so, given a close call, I'll take Collins or Morgan as my all-time 2B guy if I were builidng a team.

If I have two guys who look identical stat-wise, but one has the baggage, while the other doesn't, then I have no disagreemnt with you about picking the guy without the rep over the other guy.

What if in 2008, Albert Pujols is a free agent, and refuses to sign with team X, despite their $50M overtures, because he publicly states that he won't play with player Y. Doesn't the hindrance of drawing the best player in the game reflect on player Y's value to team X?

Depends. If Pujols doesn't want to play with the player because he's abusive, I'll listen. But if it's because of the player's halitosis, then that's a different story. :-)

Seriously, it depends on the situation. I would refuse to accept Pujol's stand without finding out if it's creditable to me or not.

Albert Belle? Sammy Sosa? Nomar Garciaparra? They've all been traded for less than full value at some point.

Albert Belle was never traded.

I know what you mean, though. Sometimes a player benefits from a change of scenery.

it is arrogant to think that personal characteristics and dynamics don't affect performance in ways that are hard to discern in the statistics.

I don't disagree with that, Andrew. But it depends on what that player is doing.

BTW, I thought the A's and Yankee teams of the seventies won because of their mutual hate for each other. :-)

I think looking to stats for the answer to the character question is probably barking up the wrong tree. Here's an example. Let's say that Port Ruppert goes 81-81 in year X. Then they trade away their best player, a reknowned clubhouse cancer named Allenbert Beckberry. In year X+1, Port Ruppert goes 70-92. Everyone wails that they couldn't trade the malicious Allenbert Beckberry for decent value and that a return of that master of hustling, utility man Rex Ecksrose, and two A-ball arms wasn't enough.

BUT... In year X, they played +8 wins relative to pythagorus and in year X+1, they played exactly as well as pythagorus suggests. Perhaps the trade only netted them two lost wins, not 11, and those two lost wins could be explained in numerous other roster spots than the one where Allebert Beckberry used to reside.

Or spin it the other way. The Rupes deal away A-Beck in the same lopsided deal after year X in which they went 81-81. In year X+1, they go 90-72 with Rex Ecksrose. The scribes say huzzah for hustling, but pythagorus says:
X= -8 pythag wins
X+1 = +0 pythag wins

Is Beckberry's obnoxiousness really to blame for eight wins of negative variance?

To actively mark down Allenbert Beckberry for character issues any more than TomH and John Murphy recommended (that is choosing between one equal player or another) seems to strongly favor the notion that we can pin down teamwide sag on one player's behavior. And while I do respect an individual voter's right to take this stance, the more I think about it, the more I believe that unless a voter can use retrosheet to somehow produce game-by-game evidence of those negative effects on team performance, I'm not willing to say that we can explain it all or even much of it with personality. Unless he was brutually abusing other players or threatening them or exposing them to underworld gambling agents.

Especially when there's 24 other guys on a team, each with their own personality quirks that we may simply not have been made privy to.

karl,

As a Red Sox fan what is your beef with Theo Epstein? One would think that having the ablity to celebrate your first title in 86 years would make you give the guy a break. I know that as a Yankees fan I wish that Boston Management thought of Theo's reign as a circus.

I think Theo is a somewhat above average GM, though not as good as Duquette, who I thought the new management shoved out the door unnecessarily quickly. The fanboy reverence for him gets up my nose, though, and he makes at least his share of mistakes, while appearing (never met the guy, so what do I know) to have more than his share of arrogance. The WS is lovely, but he got lucky; with the same breaks, it could have happened in '95, '99 or '01 (probably not in '98) Let alone in '46, '75 or '78 when we wuz robbed.

Also Epstein's 25 years younger then me. Yech!

It is arrogant to think that personal characteristics and dynamics don't affect performance in ways that are hard to discern in the statistics.

I'm not going to argue against this, because I think it's accurate, but I think it's even more arrogant to assume that we, sitting outside the clubhouse in many cases decades removed from the situation, can accurately discern who is/was helping and hurting the teams in those intangible ways.

One of the names you cited, Albert Belle, is perhaps the greatest example of this problem. His teammates almost to a whole respected him. The media, most certainly, did not. This was, of course, mostly a mess of his own making but the point stands: He's widely thought of as some sort of cancer because the media hated him, not because his teammates hated him. Same can be said of Rickey Henderson, whose past teammates have nice things to say about him, but whose reputation gets tarnished by his incompetant public speaking and outright urban legends repeated by the likes of Peter Gammons.

We are semi-capable (at least as capable as we can be) of judging the statistical record when voting for the HOM. The statistics simply are what they are; how we judge them is what we're doing him. I would argue that except in extreme circumstances (Black Sox and...er....well, other obvious game throwers), we are simply incapable of being able to figure out who was a clubhouse winner and loser through all the noise.

I think it's a net neutral, even if you are correct Tom (which I'm not necessarily granting).

Sure it's a negative to the trader, but it's a positive to the team trading FOR the player. I don't see how it impacts the player's value at all.

Wow, I'm thinking of redoing my ballot again (from scratch) and I came up with 81 players that I would consider having a reasonable case for selection. Wow.

Did I mention 'wow!'?

Second cut has it to 22 that I want to vote for, with another 17 as 'questionable'. Here are the lists.

Would like to vote for: Mackey, Suttles, Beckwith, Monroe, Moore, Schang, Beckley, Leach, Jennings, Sewell, C.Jones, Roush, Ryan, Van Haltren, Averill, F.Jones, Cravath, Rixey, Lyons, Griffith, Hubbell, Ferrell.

Questionable: Mendez, Taylor, Lundy, Sisler, Bresnahan, Chance, Lazzeri, Bancroft, Bartell, Williamson, Traynor, Griffin, Duffy, Willis, Waddell, Luque, Mays.

Still sorting the whole thing out.

Whoops, forgot to include Lave Cross on the questionable list.

I think the top 9 will be something like . . .

1. Hubbell
2. Mackey
3/4. Rixey/Lyons
5/6/7. Cravath/Jones/Griffith
8. Beckley
9. Leach

Can I just put the other 13 in a 13-way tie for 10th-15th? That would give them each 3.92 points . . .

Sure it's a negative to the trader, but it's a positive to the team trading FOR the player. I don't see how it impacts the player's value at all.

Joe, I didn't even think of it in that way, but your logic is inescapable. I think that's the first time that I have seen it expressed that way. Good job, Joe!

Joe, I'm with you on having about 22 players I'd like to have on my ballot and another 20 questionable. I can rank a top 10 pretty cleanly, but after that it gets extremely difficult to make meaningful distinctions.

Just to avoid the NgL WS est thread from sprouting too much so it's easy to search...

I don't recollect off the top of my head, Chris Cobb please comment at will... But I think a couple guys might benefit from a review of their MLEs or WS due to more stats becoming available via Gary A/Gadfly and due to their previous estimates being compiled before we'd explored the question of conversion rates in the depth we understand it now.

Specifically, I'm thinking about the NNL portions of the careers of Moore, Poles, Heavy Johnson, and Taylor, as well as the total careers of Judy Johnson, Ollie Marcell, and perhaps some of the other lesser OFs like Jules Thomas, and Jimmy Lyons.

I don't know whether a recheck before posting the "definitive" WS estimates on the NgL WS thread will help/hinder some of these guys at all, but in the cases of Moore, Poles, and Taylor, I think itcould be crucial to their candidacies since they are absolutely on the bubble.

Also I seem to remember that there was some question about whether Redding's and Winters' WS est ought to be changed one way or another for procedural/conversion reasons, and a recheck prior to posting them might be appropriate as well.

I know that what I'm saying basically falls into Chris Cobb's lap (or laptop as it might be...), but I think we'd all be happy to help in some way to make sure the info is as accurate as we understand it can be before it posts to the thread where everyone will likely be getting their NgL WS info from.

Good idea Dr. Chaleeko. I posted the numbers for Mackey, Suttles, Beckwith and Bell, but if you think I should take them down until they can be confirmed, just say so . . . Bell impressed me much more than from when I read the thread. There's some peak there (5 seasons between 25-28 WS) and the career value is immense. But if those numbers need to be deflated, that's important to know.

This theory is a sibling of TomH's (abstracted below).
Suppose a player's baseball strengths and weaknesses, or simply his skills S, are subtle or misunderstood in his time (Luis Aparicio?). He is likely to be misused on the field, and that aspect has negative value. Suppose his skills S are obvious or well understood in his time (John Kruk?). He is unlikely to be misued on the field, and that aspect has positive value.

--
TomH #89 #94 #99:
when it comes to the Dick Allens of the world, I will assess whether the player’s conduct was a contributing factor to his team’s ability to keep his talent (or at least get comparable talent in a trade) on that team.<i>

<i>All I have to do is find one player who was traded for less than "stats" value because his team didn't want his persona.
. . . the fact that the Cardinals traded Dick Allen in 1970 to the Dodgers for comparatively little hurt the Redbirds, and this is relevant to the Value (meritability) of Dick Allen.

if you create an atmosphere in which a GM is likely to engineer a trade for less-than-market value, isn't that a detriment to the team?

Tiboreau - When you say 10% discount, does that mean I should be discounting all of the offensive numbers 10% or something? Or are these totals good to go? The updates I've made are using the numbers above.

From what I've gathered from the discussion between Chris Cobb and some of the others is that Bell's playing time early in his career isn't in line with the offensive production shown in his OPS+ numbers. The 10% discount was a suggestion (actually made by Chris elsewhere) to get what would probably be a more accurate picture of his career value. Looking at the years in question . . .

. . . we see the obvious effects of Cool Papa's education as a switch hitter, but he started most every game for his teams during this time. While an excellent defensive center fielder with a decent average (with little patience and no power) would still be a starter, he probably wouldn't play quite so often, so adjusting playing time during 1927 - 31 wouldn't hurt ('27 - '29 especially).

Here's a rundown on the status of the NeL MLE projections, to lay out clearly where they all stand at the moment, as far as I can recall.

1. Cool Papa Bell's playing time should be reduced overall by 10% for his career: his seasonal rate stats should remain the same. As Tiboreau notes in the preceding post, a rigorously accurate handling of that reduction would take a proportionately larger amount from Bell's offensive downturn during the switch-hitting transition.

2. John Beckwith's and Mule Suttles' win shares, as posted on the Win Shares thread, are not current with their offensive statistics posted there. Both stand to modestly benefit in their win shares, esp. in the 1930s, from revisions to the conversion system. Since the conversion of offensive statistics is (partly) mechanized, it goes much faster than the win shares, where every comparison is done by hand.

3. Aside from the players whose totals Tiboreau has posted on the Win Shares thread, the only other player for whom I have done win shares using the same exact system that led to the MLEs for Beckwith, Bell, Mackey, Suttles, and Wilson is Dick Lundy. His win shares need an update for the same reason as Beckwith's and Suttle's, and I am not sure I have posted his full data in the form that Tiboreau has reported it, but I have all that data and could so post it.

4. The win shares posted for Dobie Moore were done using a very early version of the system now in place (he was the first player I tried to project from scratch, actually), so his numbers are in need of updating. If I did win shares for Heavy Johnson, they are from the same generation of analysis.

5. Earlier projections for earlier position players -- Pop Lloyd, Ben Taylor, Spotswood Poles, Jules Thomas, are based on discounted versions of the i9s projections, with the degree of discount determined by some conversion tests of my own. I think Cristobal Torriente was done this way also, but I can't recall off the top of my head. Their numbers could be revisited, but the data from seasons prior to 1920 is so limited that systematic MLEs generated by the system I've used from 1920 on cannot be created, at least not with the same degree of confidence.

6. For pitchers, virtually every one for whom I have estimated win shares was handled somewhat differently, and these projections are much more tentative. From what I learned working on Nip Winters and Bill Foster, I can say that the win-share projections for Jose Mendez and Dick Redding are almost certainly too high for their projected raw statistics, though those statistics themselves _may_ be alright. Bill Foster's analysis is the only one I am truly satisfied with, among the pitchers.

In conclusion, while a lot of statistical analysis has been produced, it is at present far from complete and far from regularized. Of the players for whom win shares projections are available, the only ones I think it might be better for Joe not to use at this point are those of Mendez and Redding. For the rest, I think they're either as good as they're going to get or close enough to give a fair view of the player. However, everyone should understand that they may change as we get more and better data and as I catch up older projections to the later, more accurate methodologies.

Let me add that I'm tremendously grateful for all the work that so many on the list have put into these projections and the analysis of each player that has surrounded them.

Must get back to work: I'll try to post more later today about ways in which we might speed the work of revisiting old MLEs and getting MLEs calculated for more players. I'll try also over the weekend to create a player-by-player list of the source and status for the MLEs I've done so far.

I'll try to post more later today about ways in which we might speed the work of revisiting old MLEs and getting MLEs calculated for more players.

Chris, I'm sure I'm not alone in saying sign me up for any work that I can do to help facillitate this process.

Here's a complete run-down of the MLEs and WS for NeL players that I have done so far and that are posted somewhere (usually in the player's thread) on the HoM site. They are grouped by the projection method used.

The methods are

1) Full MLE projection. This is the current system, based on my/our estimates of playing time, translations done from the best available NeL stats, OPS+ calculations from DavidFoss, and my win shares estimates based on that (for pitchers leave out the OPS+).

2) Earlier version of MLE projections. Some form of translation from NeL stats, with win shares estimates based on those translations. Playing time for these is sometimes drawn from the i9s project. These would need to be updated to account for improvements in the methodology and the data.

3) Win shares estimated from i9s MLE projections. These are usually discounted 5% or so. These may or may not be updatable, since they often cover players whose pre-1920 careers do not provide sufficient data for the current methodology to be applied.

Full MLE projection
Cool Papa Bell
Biz Mackey
Bill Foster
Jud Wilson

Earlier MLE projection
John Beckwith -- stats current, WS need updating
Dick Lundy -- pt from i9s
Jose Mendez -- pt from i9s
Dobie Moore
Bruce Petway -- projections from BA only; pt from i9s
Dick Redding -- pt from i9s
Joe Rogan -- pt from i9s, no pitching WS
Turkey Stearnes -- stats current except BB, no WS
Mule Suttles -- stats current, WS need updating
Cristobal Torriente -- projections from BA only; pt from i9s
Joe Williams -- Career MLEs no WS
Nip Winters

WS based on i9s MLE projections
John Henry Lloyd -- Variable discount to i9s
Spotswood Poles -- 5% discount to i9s
Louis Santop -- 5% discount to i9s, reworking of pt
Ben Taylor -- 5% discount to i9s
Jules Thomas -- 5% discount to i9s

Just a quick process question...I hadn't updated my voters list in a while. Who was the voter briefly known in the 1930s as George Sisler's Best Friend?

That was me.

2) Earlier version of MLE projections. Some form of translation from NeL stats, with win shares estimates based on those translations. Playing time for these is sometimes drawn from the i9s project. These would need to be updated to account for improvements in the methodology and the data.

Chris Cobb, in post #123 just above, you wrote this point about a group of players that includes Dick Lundy. When you update your MLEs or WS methods, how do you anticipate it will effect Lundy's MLEs and ultimately his WS?

More or less career?
More or less peak or prime?

Thanks, Chris.

Lundy's MLEs will go up a bit, mostly for the 1930s.

That will add career value, and perhaps extend his prime somewhat. I don't believe that his peak in the 1920s will be much affected.