Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Saturday, August 02, 2008

The Book Blog: Tango: Edgar

Tingo Tango! A HOF case for Edgar Martinez…one of these winters.

In short, in order for Edgar to compile a batting career like Wade Boggs, he’d have to hit for 3 more seasons at far below average.

Is it that important to “compile” data in order to prove your overall worthiness of a hitter?  DH or no DH, if Edgar got to 3000 hits, he’d be in.  And if Edgar sucked for 3 years, he’d reach the equivalent of Wade Boggs.

Edgar got his 1000th PA 3 years after Wade Boggs got his.  The only difference between the two is that Boggs proved he had the better glove, but Boggs also managed to play a large portion of his career at Fenway Park.

To deny Edgar the HOF is to admit igorance.  Then again, Edgar will have good company with Tim Raines.

Repoz Posted: August 02, 2008 at 11:28 AM | 212 comment(s) Login to Bookmark
  Tags: hall of fame, history, mariners

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

Page 2 of 3 pages  < 1 2 3 > 
   101. JPWF13 Posted: August 04, 2008 at 10:01 PM (#2889783)
The evidence suggests to me that in 1989 Seattle wanted to give the job to Edgar - but he didn't grab it. It also suggests that the Mariners are getting a bit of a bad rap here; it wasn't 100% clear that Edgar was, indeed, their best option at 3B until about the time he actually took the job over for good.

-- MWE



100% clear is an absurdly high hurdle. I saw Presley play- he was a terrible player, absolutely awful, he kept getting as many PAs as he did by living off that one 100 RBI season.
No Edgar didn't "grab" that 1989 job, but if you are only going to give a job to someone who sets the world on fire those first 100 at bats...
   102. CrosbyBird Posted: August 04, 2008 at 10:32 PM (#2889822)
I'm going to post the guys on Tango's list (referenced in post #98) that I think are better HOF candidates than Martinez:

Bonds, Rickey, Ripken, Sheffield, Frank Thomas, Gwynn, Boggs, Raines, Piazza, Biggio, Alomar, Larkin, Sandberg, Kent, Trammell, Bagwell, Palmeiro, McGwire, Sosa, Walker.

Walker is probably the only one on that list that I don't consider a HOFer, and it's questionable.

For pitchers, I'd have Clemens, Maddux, RJ, Glavine, Smoltz as clear inductees, with Schilling, Brown, and Mussina as questionable. Of the three "questionables," I would put Schilling in (postseason and strikeouts give him the edge), not Brown (but I could be convinced), and not Mussina (although another season or so of what he's doing now and he passes Brown in my book. which I think is likely).

That puts me around 27 HOFers born between 1958-1968. It looks like my estimate of 15-20 per decade is a little low, but I think this happens to be an uncharacteristically strong 11 year period and that it will taper off a little bit.

Modern-day relievers really are a new position so they don't count for the comparison... those that get in (and that's a whole separate discussion) are really a separate category. I'm not sure any modern relievers merit HOF induction but if they do Hoffman certainly deserves it.

Where I tend to disagree with the Edgar supporters is comparing him to guys like Palmeiro, McGwire, Sosa, and Walker. I think they are all clearly better candidates. McGwire is a much better hitter. Sosa and Palmeiro have more career value (and not padded with crap seasons) and more defensive value (Palmeiro much more), and both have more value on the bases (Sosa much more). Walker is not as good a hitter as Martinez, but he's close, and he's a great defensive player with excellent speed.
   103. Ray (RDP) Posted: August 04, 2008 at 10:51 PM (#2889851)
For pitchers, I'd have Clemens, Maddux, RJ, Glavine, Smoltz as clear inductees, with Schilling, Brown, and Mussina as questionable.


For starting pitchers I have all of those, no questionables, with Mussina bringing up the rear and clearing the hurdle.

I don't see how one has Smoltz in but Schilling and Brown out (or "questionable"). (Mussina is not that much different, but is a half-notch below.)

I'm not sure any modern relievers merit HOF induction but if they do Hoffman certainly deserves it.


That's a funny way to spell Rivera.
   104. dlf Posted: August 04, 2008 at 10:57 PM (#2889861)
I'm not sure any modern relievers merit HOF induction but if they do Hoffman certainly deserves it.


That's a funny way to spell Rivera.


That's a funny way to misread. The list of candidates was specifically limited to '58 to '68; Rivera isn't in that group.
   105. Ray (RDP) Posted: August 04, 2008 at 11:18 PM (#2889885)
   106. Ray (RDP) Posted: August 04, 2008 at 11:18 PM (#2889886)
Yeah, missed that.
   107. Los Angeles Waterloo of Black Hawk Posted: August 04, 2008 at 11:22 PM (#2889891)
At what point did (not does) Albert Pujols cement his position as a HOF (the 10-yr rule notwithstanding)?

Well, I don't know that he has. I know BPro's WARP is substandard, but they have him in the 80s. I'm not sure where most 1B are in the HoF, but I suspect just about all of them are north of 100. I don't know where AROM's more reasonable WAR measures, or Dan Rosenheck's, may have him, but I would bet he's pretty close to 60 in AROM's formulation. I would suspect he's at most a season or two away from amassing the number of wins we tend to see from HoFers.

However with respect to the hall of fame it is exactly the wrong question to ask since "value" is the wrong measurement. What matters is not who is more valuable but who is the better player. It is axiomatic that the 150 WAR in 140 games is a better player and that is what should matter.

I don't know that I agree with any part of this. "Value" is demonstrated talent. I don't think it's axiomatic that a 5 WAR player in 150 games is better than a 5 WAR player in 140 games; the latter actually has a higher rate of contributing to wins. Winning games is what matters.
   108. Tango Posted: August 05, 2008 at 02:26 AM (#2890247)
Hawk: Extending the Pujols question, at what point did Wayne Gretzky and Bobby Orr cement their position as not only HOFers, but as 2 of the 4 greatest players of all time?

Pujols has about 5000 career PA. If you find the best 5000 stretch of PA for every player who has ever played (at any point in their careers), how many players are better than Pujols? And of those players, are all of them basically "no-brainers". Are the guys just below him also no-brainers? If you do something that can only be accomplished by someone of super great talent, then guess what that makes Pujols.

Dan: re Giambi. His 3 years, while great, does not necessarily imply a stretch of performance that could only have been done by a HOF-type player. It might qualify him, I don't know. I'd have to look. But, this is the question we are trying to answer. His 3 years could be similar to say Kevin Mitchell or Kal Daniels or lots of other players. If that's the case, then it won't really meet the definition, any more than someone pitching a perfect game.

Does throwing 58 scoreless innings qualify? I don't know. How much talent do you need to do that? Could it be possible that a good pitcher with a few lucky breaks do that? How about a 39-game hitting streak? Benito Santiago did that, right?

***

Anyway, go back to my hockey question and answer that. Hint: Bobby Orr is currently considered one of the 3 greatest hockey players of all time, and his career was over by the age of 26.... just 9 seasons. Does that mean after 8 seasons he'd be considered top 5? After 7 seasons top 10? After 6 seasons top 30?

See where I'm going here? What level of talent would produce a career like Bobby Orr?
   109. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 05, 2008 at 02:53 AM (#2890306)
The only 1B with a better 3 straight years than Giambi were Foxx and Gehrig, according to my WARP.
   110. The District Attorney Posted: August 05, 2008 at 02:53 AM (#2890308)
So you're essentially arguing for "signature significance": that if you're able to have a couple of Edgar Martinez hitting years or a couple of Sandy Koufax pitching years, that means you must have been greater than a Paul Molitor or a Don Sutton, who was never able to reach that level at any point? I don't like that argument at all. It's not the Hall of Demonstrated Talent at a Certain Point. I see no reason to believe that the players with the highest peaks intrinsically made more impact on the game, contributed more wins, or whatever else you might want to go by... not without considering other things besides highest peak, anyway.

I understood you previously to be making a "replacement value in a HOF context should be very high" argument. That would lead to many of the same conclusions, but it's not quite the same thing, and it's IMO a far more plausible starting point. But again, I'd like to see just what results such a system would come up with (in a way that reveals more of the process than just listing in/out names.)
   111. Tango Posted: August 05, 2008 at 03:11 AM (#2890377)
that means you must have been greater than a Paul Molitor or a Don Sutton, who was never able to reach that level at any point


Not quite. Sutton doesn't have to have 6 years like Koufax. It's possible that Sutton's 12 best years are more significant than Koufax's 6 best years. I don't know.

It's possible that Nicklas Lidstrom's 1400 games are more significant than Bobby Orr's 700 games.

It is a peak discussion, but a "floating" peak.

While Giambi's 3 straight great years may be very high, it is possible that it simply doesn't pass the significance test.

We want to know how many standard deviations from the baseline level he is. And it's possible that even 3 super great season won't be able to pass enough standard deviations to satisfy that. Unless probably seasons by Pedro or Ruth.
   112. Gaelan Posted: August 05, 2008 at 03:13 AM (#2890382)
I don't know that I agree with any part of this. "Value" is demonstrated talent. I don't think it's axiomatic that a 5 WAR player in 150 games is better than a 5 WAR player in 140 games; the latter actually has a higher rate of contributing to wins. Winning games is what matters.


You misread me. The 5 WAR in 140 games is the better player.
   113. Tango Posted: August 05, 2008 at 03:25 AM (#2890396)
Btw, of course average seasons have value. To a team, to fans, to everyone. But, do they necessarily have value to determining the best players of all time? Does Pete Rose's career in the 1980s really bolster his case (peripheral issues notwithstanding)? Or choose whoever you want. Clearly they have value in winning games. But they do nothing to establish what we already knew of his talent level.

Is there any player that you think "man, he's borderline, but if he can just be an average player for 3 more years, then he's a hall of famer!" Does that make much sense? But, this is the argument people are trying to make.

From 1997-2003, Pedro was 118-36. Does he really need to go 10 seasons of 14-14 for you to say "yup, because of those seasons, he brought his case up to Jim Palmer, Bob Gibson, Juan Marichal levels".

Does anyone believe that? Pedro is a HOF based only on 1997-2003. Everything else is filler. He doesn't need those seasons to be equal to the other 3 guys. He's already demonstrated a talent level better than those 3 guys based on those 7 seasons alone.
   114. Los Angeles Waterloo of Black Hawk Posted: August 05, 2008 at 06:35 AM (#2890509)
Hawk: Extending the Pujols question, at what point did Wayne Gretzky and Bobby Orr cement their position as not only HOFers, but as 2 of the 4 greatest players of all time?

Pujols has about 5000 career PA. If you find the best 5000 stretch of PA for every player who has ever played (at any point in their careers), how many players are better than Pujols? And of those players, are all of them basically "no-brainers". Are the guys just below him also no-brainers? If you do something that can only be accomplished by someone of super great talent, then guess what that makes Pujols.


I agree that Pujols has demonstrated "HoF-level talent". I disagree (or, rather, don't take as a given, I'm open to the argument) that he has already demonstrated "HoF-caliber longevity", which is an integral part of the definition of "Hall of Famer". Achievement matters. Performance matters. Wins matter. There is a 10-year minimum for a reason.
   115. Los Angeles Waterloo of Black Hawk Posted: August 05, 2008 at 06:54 AM (#2890516)
And I'm not sure where this ends. Herb Score, in his first two seasons, demonstrated "super-great talent". Two seasons not enough? Dwight Gooden, in his first five seasons (which actually includes his first decline year), demonstrated such talent. Is five seasons enough?
   116. CrosbyBird Posted: August 05, 2008 at 08:19 AM (#2890530)
I agree that Pujols has demonstrated "HoF-level talent". I disagree (or, rather, don't take as a given, I'm open to the argument) that he has already demonstrated "HoF-caliber longevity", which is an integral part of the definition of "Hall of Famer". Achievement matters. Performance matters. Wins matter. There is a 10-year minimum for a reason.

This is exactly what I'm thinking. When I consider a HOFer, I take into account peak performance, career performance, and longevity. All of them matter. If a hitter put up a .500/.750/1.500 line over two full seasons and then died in a tragic motorcycle accident, he's not a HOFer. If a pitcher plays 25-30 seasons pitching like Adam Eaton in 2007 on good teams and sneaks up on 250-300 wins, he's not a HOF either.

Pedro retiring after the 2003 season is even worse than Koufax as a comparison to Edgar Martinez, because he put up the sort of performance within his peak that defied the perceptions of how well a player could perform. The kind of seasons where no other player in baseball could touch him. Compare Pedro 2000 to the second-best pitcher in the league in ERA, in K/9, in WHIP. 3 CY awards in 4 years, 6 top 3 finishes in 7 years.

Martinez basically has one offensive season that makes it into the top 100 of all time, despite playing in one of the best offensive eras in baseball. That's pretty unimpressive for a guy whose claim to fame is hitting.
   117. Blackadder Posted: August 05, 2008 at 10:48 AM (#2890549)
Dan, how many players, period, had better three year consecutive stretches than Giambi? And are any of them not viewed as all-time greats?
   118. Tango Posted: August 05, 2008 at 11:16 AM (#2890558)
Hawk: I don't look for "reason" with the 10yr rule. Bobby Orr played 9 full seasons, then a couple of 10 game seasons here and 20 game season there. Imagine keeping, at the time, the greatest hockey player ever out of the HOF on a technicality like that. Similar if Ted Williams retired after 9 seasons. Didn't Jim Brown only play 10 seasons?

As for 2 seasons or 3 seasons or 5 seasons being enough: like I said, whatever is significant. What kind of talent level would be required to produce what Gooden produced in his first 5 years (aside: probably not the best example, as his ERA+, outside of that 24-4 season is nothing terribly exciting). Maybe the best example of the short career is Maddux from 1994-1995. That is, would it be enough of a statistical significance test if Maddux came into the league in 1994, pitched those 2 seasons, and was done for his career? That performance would probably be say 10 standard deviations from the mean. From that standpoint, he would almost certainly qualify in my book. It's a level of actual performance that could only have been achieved by someone of the talent level of Gibson, Seaver, Clemens, etc. Terry Mulholland and Oil Can Boyd would not have put up 2 years like that. Well, it's "possible", at the probability of whatever 10 standard deviations implies. Then again, Bob Gibson's career in its entirety might also be 10 standard deviations away from a .500 pitcher. This is what I'm trying to say.

If Lebron had played one year in college, and you had all these other typically good players who played 4 years, Lebron would be considered the greatest player ever in that college, even if you had a "2 year min rule", because the performance line he would have put up was so far out that only someone with a very high talent could have put up those numbers.
   119. BDC Posted: August 05, 2008 at 12:11 PM (#2890569)
Tango's arguments are really convincing. It's part of the nature of baseball that pitchers in particular can establish that they are truly great and then, because of injury, disappear or become greatly diminished. It's been happening from Joe Wood to Dwight Gooden.

Here's an extreme example: Jack Quinn came up to the majors in 1909, which was Wood's first full season. Quinn had a very long career, and a good one (247-218, 114 OPS+). Given your choice of the two and control of their entire careers, with perfect foresight, you might have chosen Quinn, who certainly had more career value (281 WS to Wood's 193).

But Wood, by any reasonable definition, was a far better pitcher. Quinn never, in his dreams, could have done what Wood did in 1911-12 (and not for lack of opportunity). And Wood's being "better" does not reflect potential, but actual clear demonstration thereof; he dominated a major league and won three games in a World Series.

Now it's perfectly fair to say that Wood is no Hall of Famer, because we expect a Hall of Famer to have contributed more than even that. It's not a Hall of the Best Pitchers Who Got Hurt and Became Journeyman Outfielders. But somewhere between Wood and Koufax, there's a line (an arguable line, of course) where career length completely ceases to matter, and a Hall of Fame becomes absurd without someone like Koufax in it. And Koufax had only 194 Win Shares, less than Bob Friend or Claude Osteen; but in that kind of case career value becomes entirely beside the point.

Sorry for belaboring the obvious, but I really like the way that Tango has put things here, and I wanted to think aloud a little on the topic ...
   120. Tango Posted: August 05, 2008 at 03:56 PM (#2890793)
For those interested, I have tried to translate my words and ideas into numbers so that you can see the various implications.
   121. Greg Maddux School of Reflexive Profanity Posted: August 05, 2008 at 04:03 PM (#2890801)
there's pretty much no player that could possibly accumulate 600 HR right now and not merit induction based on performance.

Close your eyes and try to imagine...Sammy Sosa.

His career basically consists of one peak Mel Ott season (2001), four years of peak Bobby Abreu ('98-00, '02) that people mistakenly believe were on par with 2001, and 13 years that were collectively about a win-and-a-half above average. Brian Giles has as many WAR. The Brian Giles of not-a-regular-'til-he-was-28 fame. I can only hope the writers get this one right for a reason as preposterous as using a translator in front of Congress.
   122. Los Angeles Waterloo of Black Hawk Posted: August 05, 2008 at 04:25 PM (#2890844)
Hawk: I don't look for "reason" with the 10yr rule. Bobby Orr played 9 full seasons, then a couple of 10 game seasons here and 20 game season there. Imagine keeping, at the time, the greatest hockey player ever out of the HOF on a technicality like that. Similar if Ted Williams retired after 9 seasons. Didn't Jim Brown only play 10 seasons?

Well, I don't know how other sports compare in this regard. I know practically nothing of the average length of a hockey career, but I know that the career of a productive NFL running back is less than that of a productive major league baseball player. And longevity is certainly considered as far as the Football Hall of Fame goes -- Art Monk was never as spectacular a WR as Terrell Davis was a RB, but Monk is in the Hall and Davis may never be, because the voters recognize the difficulty of maintaining performance over time.

Maybe the best example of the short career is Maddux from 1994-1995. That is, would it be enough of a statistical significance test if Maddux came into the league in 1994, pitched those 2 seasons, and was done for his career? That performance would probably be say 10 standard deviations from the mean. From that standpoint, he would almost certainly qualify in my book.

Okay, I see what you're saying, but the Hall as constituted asks us not only to evaluate whether or not a HoF performance is demonstrated, but that it is maintained.

If Lebron had played one year in college, and you had all these other typically good players who played 4 years, Lebron would be considered the greatest player ever in that college, even if you had a "2 year min rule", because the performance line he would have put up was so far out that only someone with a very high talent could have put up those numbers.

Well, I simply don't see this as true -- maybe it depends on the college. It, obviously, depends on how you'd define things. (Incidentally, a relatively low percentage of the greatest college basketball players of all time have played four years. Freshmen were ineligible until 1972, and of course over the last decade we've seen players leave early, and the one-year rule is only going to exaggerate that trend.)
   123. Los Angeles Waterloo of Black Hawk Posted: August 05, 2008 at 04:30 PM (#2890859)
Does Dutch Leonard's 1914 season make him a HoFer?
   124. Barry`s_Lazy_Boy Posted: August 05, 2008 at 04:42 PM (#2890878)
If you favor peak, does a guy like Sutton make the HOF? How about Nolan Ryan?
   125. Mike Emeigh Posted: August 05, 2008 at 04:56 PM (#2890895)
I think you are holding the Mariners to a standard to which no team would have been held at the time if you pillory them for not handing Edgar the job before they did.

Presley hit 24 HR in 1987. How many teams in the major leagues would have replaced a player who was (a) only 26 and (b) coming off two straight 20-HR years, for anything other than the next A-Rod? Edgar had a good year in 1987, true - but as I pointed out before, it was on the heels of two relatively poor season at Chattanooga, it was in Calgary, in a hitters' hotbed, and he didn't hit for power (.473 SLG and 10 HR). I suspect that the MLEs for that park and environment would have been right in line with the MLEs from Chattanooga (.258/.353 BA/SLG in 1985; .264/.390 BA/SLG in 1986). .329/.473 BA/SLG in the PCL didn't shout "success". Yes, Edgar hit a ton in his ML debut, but how many teams (outside of the Pirates) hand somebody a job on the basis of a 13-game trial, with that kind of minor league record? I don't think it's reasonable to argue that Edgar should have had the job handed to him at the start of 1988.

Presley got off to a slow start in 1988, and Edgar did come up for a week or so in May when Mario Diaz got hurt. Edgar didn't set the world on fire, and Presley's bat showed some signs of picking up, so back Edgar went. I can't find any references to it, but I believe he got hurt shortly thereafter; he played in only 95 games for the Cannons in '88 and he wasn't up for very long in May. Again, I think this is defensible.

In 1989, as I also mentioned, the Mariners gave the 3B job to Edgar out of ST. Presley played some 3B, some 1B. On April 27, Edgar was hitting .176/.259/.176. Presley wasn't doing much better at that point, but he started to pick it up right after that, and by mid-May he was hitting over .300. Edgar's playing time dwindled as Presley moved back into the lineup at 3B. Again, I think that was a reasonable decision on Seattle's part; you have a guy who had been your regular hitting, and the heir apparent wasn't.

On June 13, Presley got hurt. Edgar, who had finally been sent out a couple of weeks earlier (hitting .229/.304/.286), was recalled and put back into the lineup. Over the next two weeks, he went on a mini-tear, boosting his line to .277/.336/.366 on June 29 - and even though Presley was back playing, he couldn't get back into the lineup regularly. Then Edgar cooled off again. By the All-Star break he was down to .246/.311/.321, while Presley was at .275/.314/.410. Both players struggled over the next couple of weeks, but when Presley hit a two-run, pinch-hit HR off Bryan Harvey to cement a six-run rally which gave Seattle a win over the Angels on July 31, he got the job back, and Edgar was optioned back to AAA two days later (hitting .235/.316/.302). Presley's game winner was, more or less, his last hurrah; he slumped horribly after that, and by the end of August was down to .234/.271/.372. Darnell Coles took over at 3B; Edgar was recalled in September, but started only four games that month, while Presley was also limited to spot starts.

If you limit yourself what we knew at the time: Edgar could, obviously, hit in Calgary. So could other players; Calgary was typically among the PCL leaders in BA and in hits allowed by their pitchers, and their team BA during Edgar's three seasons there was .287, .290, and .289. Calgary's hitters led the league in HR in 1987 and 1988 and they were third in 1989 (Colorado Springs joined the league that year). Calgary was an excellent environment for hitters in a league full of them. We simply didn't know with any reasonable degree of certainty - in 1989 - whether Edgar was a real hitter or someone whose stats were inflated by the environment. And the evidence we were presented by his major league performance and his minor league performance outside of Calgary suggested caution. If we were faced with a prospect of Edgar's age, and his stats as of 1987, in today's environment, I think that there would be few minor league analysts who would suggest that today's Edgar had earned the right to compete for the starting job.

So I don't concur with the majority that Edgar was "blocked" by a recalcitrant Mariners' management. I think that Seattle acted reasonably and mostly fairly.

-- MWE
   126. Tango Posted: August 05, 2008 at 05:19 PM (#2890934)
Okay, I see what you're saying, but the Hall as constituted asks us not only to evaluate whether or not a HoF performance is demonstrated, but that it is maintained.


It should go without saying that we are applying our personal definitions here, and not whatever the suits are writing. That is, if the HOF didn't exist, what would we do?

As for career length in hockey, I'd say it's fairly close to baseball. Number 50 in career games played in the NHL is 1282 games, and with the league being 70-82 games for the majority of the new age, that implies around 16 or 17 seasons. Number 50 in MLB is 2499 games, and with 154-162 games being a season, that makes it around 16 seasons. Not necessarily the best way to do it, but I kinda expected the career lengths to be similar.

So, what do you do with Bobby Orr, or a Ted Williams that say may have either been killed in a war or out for injury after 9 years? After 7? 5?
   127. Tango Posted: August 05, 2008 at 05:27 PM (#2890945)
Does Dutch Leonard's 1914 season make him a HoFer?


Check out the link I provided. If you use the comparison point as a below-replacement level player, you will see that that single season is not enough standard deviations away to define greatness.

I provided several examples in that link, so check it out, and let me know what you think.

If you favor peak, does a guy like Sutton make the HOF? How about Nolan Ryan?


As noted, if you have a "floating peak" as I'm describing it, then both would likely qualify.
   128. David Nieporent (now, with children) Posted: August 05, 2008 at 05:29 PM (#2890949)
Pujols has about 5000 career PA. If you find the best 5000 stretch of PA for every player who has ever played (at any point in their careers), how many players are better than Pujols? And of those players, are all of them basically "no-brainers". Are the guys just below him also no-brainers? If you do something that can only be accomplished by someone of super great talent, then guess what that makes Pujols.
Someone of super great talent. But is the HOF for mere "talent"?

If Barry Bonds came up from the minors, in his first season had his 2001 (or 2002, or 2003, or 2004) season, and then was tragically murdered by Kevin, would he be a HOFer in your formulation? I'm pretty sure that an OPS+ of 250, even in a single season, is something that can only be accomplished by someone of super great talent. So is that single season -- putting away any steroid consideration -- enough, to you?

What if Kevin was intercepted by the mental health authorities before he got to Barry, so Bonds was able to keep playing -- but then never had a year anything like it again? Is it enough that he had the "talent" to do that?
   129. JPWF13 Posted: August 05, 2008 at 05:35 PM (#2890958)
Presley hit 24 HR in 1987. How many teams in the major leagues would have replaced a player who was (a) only 26 and (b) coming off two straight 20-HR years, for anything other than the next A-Rod?


Not many, basically:

1: An incredibly cheap team that wanted to unload the established player before he got expensive

2: When the established player is traded for something the team wants even more- as part of a deal for a "star" for instance
   130. Tango Posted: August 05, 2008 at 05:41 PM (#2890965)
If Barry Bonds came up from the minors, in his first season had his 2001 (or 2002, or 2003, or 2004) season, and then was tragically murdered by Kevin, would he be a HOFer in your formulation?


Please read the link in post 120. It's all there. Choose the scenario that makes the most sense to you, and you'll get your answer.
   131. David Nieporent (now, with children) Posted: August 05, 2008 at 05:45 PM (#2890969)
Presley hit 24 HR in 1987. How many teams in the major leagues would have replaced a player who was (a) only 26 and (b) coming off two straight 20-HR years, for anything other than the next A-Rod?
Presley was a lousy player in 1987. Now, your argument, I guess, is that he was a lousy sabermetric player, not a lousy player by stats in common use at the time, I guess. But he hit .247, struck out 157 times, played defense like a wounded buffalo, and even if teams didn't really understand contextual adjustments, they knew that 24 HR in 1987 wasn't too impressive, given that everybody and his grandmother was hitting HRs that year. While his 27 HR the previous year put him vaguely near the top 10, his 24 HRs in 1987 didn't even look interesting.
   132. David Nieporent (now, with children) Posted: August 05, 2008 at 05:53 PM (#2890981)
Please read the link in post 120. It's all there. Choose the scenario that makes the most sense to you, and you'll get your answer.
I did read the link (though after I posted my question). But you didn't answer the question. You asked it:
So, is that enough? Is it enough to say that your performance is 8 standard deviations from the league mean, in order for your Observed performance to infer great talent?
But you didn't answer it:
I don’t know.
I wanted you to do so, to see how far you took your view.

That link also doesn't answer the second question I posed: do you view a player who puts forth a 260 OPS+ in one season differently than a player who puts forth a 260 OPS+ in one season and a (say) 120 OPS+ in another season? If all you're looking at is peak ability, then it shouldn't -- and from your rhetoric, you've implied that you wish to disregard any season which isn't a "great" season -- but it would seem like a very strange notion.
   133. Tango Posted: August 05, 2008 at 05:59 PM (#2890990)
David, I laid out scenarios so that you get to choose the one that makes the most sense. I gave different examples, how far a performance is from a .340 wOBA, .300 wOBA, .260 wOBA, .220 wOBA.

To me, it seems that the number of standard deviations from the .260 wOBA probably makes the most sense. That's why I laid out those scenarios. I'm not giving you the right answer. I'm giving you the framework so you can end up with your own answer.

And I've never intended to imply "that you wish to disregard any season which isn't a "great" season". What I am always saying is that you look at the performance and infer what kind of talent level would produce that performance.

And, in the link I provided, I even give a scenario in which someone who is a league average player for 28 seasons (aged 18 to 45 let's say) could qualify as being a borderline candidate, because he'd be far enough away from a specific baseline. Again, it all depends on which baseline you settle on, be it .340 or .220 or whatever floats your boat.
   134. Los Angeles Waterloo of Black Hawk Posted: August 05, 2008 at 06:06 PM (#2891002)
Tango, your examples at that link are instructive, but as you mention there's a strong element of personal definition here, we are essentially at an impasse. To me, duration is part of greatness.
   135. Tango Posted: August 05, 2008 at 06:10 PM (#2891007)
As I said in my blog, if you subscribe to the WAR theory, such that a 60-70 WAR player is borderline, then do you accept the implication that Pedro having a career solely based on 1997-2003 and Ted Williams having a career solely based on his first 6-7 years (or being unanimous MVP for his first 6 seasons), or Orr or Gretzky or Jordan having a career that only spans 6 or 7 years will make them all borderline candidates?

I'm fine with whatever you say, as long as you realize the implication of your position.

***

The framework I'm providing is a way to combine peak and longevity in such a way as to not have the Koufax-exception, and without creating a hodgepodge way of mixing peak and longevity.
   136. GGC don't think it can get longer than a novella Posted: August 05, 2008 at 06:10 PM (#2891008)
Tango, about 1.3% players make the HOF. Doesn't that actually make the threshold less than 8 standard deviations?
   137. JPWF13 Posted: August 05, 2008 at 06:14 PM (#2891015)
Presley was a lousy player in 1987.


and in 1988, and 1989 and 1990...

the problem with MWE's statement above
If we were faced with a prospect of Edgar's age, and his stats as of 1987, in today's environment, I think that there would be few minor league analysts who would suggest that today's Edgar had earned the right to compete for the starting job.


is that now, generally speaking, statheads and analysts SCREAM to have players like Presley replaced- even if only by AAAA players.

Presley from 1987 to 1990 was a complete trainwreck- awful at every single baseball skill- their as literally nothing he did well (by MLB standards) the closest analogy I can think of in this century is Angel Berroa:
Berroa's 2003 matches Presley's 1985-86, and Berroa's 2004-2008 matches Presley's 1987-1991
   138. JPWF13 Posted: August 05, 2008 at 06:17 PM (#2891017)
I did read the link (though after I posted my question). But you didn't answer the question. You asked it:

But you didn't answer it:

David, I laid out scenarios so that you get to choose the one that makes the most sense. I gave different examples,


I'm having a flashback to a thread discussing MGL's inability/unwillingness to communicate to non-statheads...
   139. Los Angeles Waterloo of Black Hawk Posted: August 05, 2008 at 06:17 PM (#2891018)
As I said in my blog, if you subscribe to the WAR theory, such that a 60-70 WAR player is borderline, then do you accept the implication that Pedro having a career solely based on 1997-2003 and Ted Williams having a career solely based on his first 6-7 years (or being unanimous MVP for his first 6 seasons), or Orr or Gretzky or Jordan having a career that only spans 6 or 7 years will make them all borderline candidates?

I'm fine with whatever you say, as long as you realize the implication of your position.


If the players in question produce wins similar in total to other HoF players, sure, they're in consideration. I'm fine with that.

Looking at the Pedro example, I don't know how many WAR he had in 1997-2003. Assuming it was in the 60-70 range, and assuming 60-70 is borderline (I don't know that either of those things are true, but for the purposes of discussion I'll stipulate that they are), yes, those years alone make him a borderline candidate. I'm fine with that. But seven years is a long time to maintain such dominant performance, so I don't think that example contradicts anything I've said.
   140. Tango Posted: August 05, 2008 at 06:18 PM (#2891025)
GGC: I don't think you are comparing the right things. The .380 observed wOBA being 8 SD from the .340 wOBA after 16 seasons is correct. What we don't know (or haven't shown anyway) is what is the distribution of talent in the league. If the distribution of talent is fairly wide, then being 8 SD from the mean is no big shakes. If the entire population of MLB was all a true .335 to .345, then having a .380 career is enormous.

So, the 1.3% has nothing to do with what I'm saying.
   141. Tango Posted: August 05, 2008 at 06:26 PM (#2891042)
I'm having a flashback to a thread discussing MGL's inability/unwillingness to communicate to non-statheads...


I'm certainly willing. Am I able to? I thought I am. I already said I don't know the answer. I'm not going to pretend that I do.

But, more important than my answer, which, really, is irrelevant and useless, is that I provided a framework to allow you to create your own answer. Isn't that more important, or at least intriguing?

Perhaps your or David can word or rephrase your questions a bit differently so I can give you a better answer.

If you insist on an answer, then... 42.

***

...yes, those years alone make him a borderline candidate.


Fantastic. Thanks for asserting the implication.

Given that position, say someone who plays for 7 years and has a 60-70 WAR total and 60-70 is borderline, then in order for him to be pushed above that level, anything he does above replacement level, say 8 seasons of being a .450 pitcher will now make him HOF material.

Again, there's nothing necessarily wrong with that. You've got your framework (WAR), you've got your position, and you are not making an "exception" that contradicts your framework.

Good...
   142. PreservedFish Posted: August 05, 2008 at 06:33 PM (#2891056)
If a hitter put up a .500/.750/1.500 line over two full seasons and then died in a tragic motorcycle accident, he's not a HOFer.


He'd get my vote
   143. JPWF13 Posted: August 05, 2008 at 06:37 PM (#2891064)
But, more important than my answer, which, really, is irrelevant and useless, is that I provided a framework to allow you to create your own answer. Isn't that more important, or at least intriguing?


Well that's what gave me the flashback, MGL insisted upon giving his framework to a guy who had zero interest in a framework- he wanted an "answer" even if just an opinion- the thread devolved into a minor flame war from there.

If you insist on an answer, then... 42.



"Forty two?!" yelled Loonquawl. "Is that all you've got to show for seven and a half million years' work?"

"I checked it very thoroughly," said the computer, "and that quite definitely is the answer. I think the problem, to be quite honest with you, is that you've never actually known what the question is."
   144. CrosbyBird Posted: August 05, 2008 at 06:54 PM (#2891101)
But, I want to see the WAR supporters acknowledge this fact, that Ted after 6 or 7 years would have been debatable for the hall and that Pedro 1997-2003 would also be on the threshhold.

You got it. The production you’re talking about probably is pretty close to the minimum total production where I would start considering inducting a guy with a 6-7 year career.

A short career is a mark against a player in HOF qualification. An especially short career is a serious mark. And 3-4 years of replacement-level ability adds nothing to either player’s case (in my personal HOF, as opposed to the real HOF, where it’s the difference between even being eligible for induction).

For a current example, hit by a truck tomorrow, Pujols is not a HOFer. Pujols with an injury tomorrow that renders him a replacement level player for the rest of his career is not a HOFer either.

Given that position, say someone who plays for 7 years and has a 60-70 WAR total and 60-70 is borderline, then in order for him to be pushed above that level, anything he does above replacement level, say 8 seasons of being a .450 pitcher will now make him HOF material.

I don't know that the line is that narrow, where 5-10 WAR by itself is enough to change someone from "clearly out" to "clearly in." At the margins, my personal HOF equation is not purely objective, nor do I believe it would be appropriate for it to be so. That's where milestones, MVP/CY voting, all-star selections, perception of "dominance" while playing, etc. come into play.

I would personally not include any modern relievers in the HOF, because I do not feel they have enough career value. That includes Gossage and Rivera.
   145. The Bones McCoy of THT Posted: August 05, 2008 at 06:59 PM (#2891113)
128 OPS+ from a corner OF in a hitter's park

Isn't the fact of being a hitter's park taken into account in *OPS+?


D'OH!!!!!!!!!! (smacks palm to forehead)

Me so silly.

Best Regards

John
   146. Los Angeles Waterloo of Black Hawk Posted: August 05, 2008 at 07:05 PM (#2891138)
I don't know that the line is that narrow, where 5-10 WAR by itself is enough to change someone from "clearly out" to "clearly in." At the margins, my personal HOF equation is not purely objective, nor do I believe it would be appropriate for it to be so. That's where milestones, MVP/CY voting, all-star selections, perception of "dominance" while playing, etc. come into play.

Incidentally, I agree with this. Using WAR as a "benchmark" in the above discussion was basically for simplicity's sake.
   147. David Nieporent (now, with children) Posted: August 05, 2008 at 07:09 PM (#2891147)
To me, it seems that the number of standard deviations from the .260 wOBA probably makes the most sense. That's why I laid out those scenarios. I'm not giving you the right answer. I'm giving you the framework so you can end up with your own answer.
But I already know my answer; I certainly didn't need to ask you that. I wanted your answer. And while you start to answer the question by saying that you choose the .260 threshold, you don't come right out and answer the one I'm asking: namely, do you really mean that if a player has a good enough single season (for whatever value of "good enough"), you'd call him a HOFer. And the second question: does it really not matter to you if, outside that great season, he did well or badly or not at all.


I'm certainly willing. Am I able to? I thought I am. I already said I don't know the answer. I'm not going to pretend that I do.

But, more important than my answer, which, really, is irrelevant and useless, is that I provided a framework to allow you to create your own answer. Isn't that more important, or at least intriguing.
Not really, since I don't agree with your framework at all. (I think it's answering the wrong question -- I do think the Hall is about value, not "talent.")

You seem to be worried that I'm asking your answer because then I can say, "Well, so-and-so's a HOFer (or not) because Tango says so." But I'm asking the question because I'm trying to explore how far you take this "framework."
   148. Mike Emeigh Posted: August 05, 2008 at 07:09 PM (#2891148)
But he hit .247, struck out 157 times, played defense like a wounded buffalo, and even if teams didn't really understand contextual adjustments, they knew that 24 HR in 1987 wasn't too impressive, given that everybody and his grandmother was hitting HRs that year.


I don't think anyone, including the Mariners, thought that it was anything but a bad year. But I also don't think anyone, including the Mariners, would have thought that Presley should have been replaced after *one* bad year - even by a player who had just had his first GOOD year in the minors. Presley's situation after 1987 was not unlike Aramis Ramirez's after 2002, or Matt Williams's after 1992 - he was a young regular who had his first truly bad season since becoming a regular. I'll stipulate that neither the Pirates nor the Giants had an Edgar available to them, but I seriously doubt anyone would have been making the argument that either deserved to be replaced at that stage of their careers even if a player with Edgar's career minor league stats to that point HAD been available to them - two poor seasons at AA, one decent but not great season given the context at AAA. What reason did we have to think that Edgar would have been significantly better at that point, going in to 1988? What reason did we have to think that Presley's wouldn't bounce back, before 1988? Presley at the level of 1985-1986 was an adequate offensive player, and Edgar before 1988 didn't project to be better than adequate when you account for the context at Calgary. I just don't see how - given what we knew in 1988 - anyone can say that Edgar deserved to be playing in the majors ahead of Presley that season. Sure, if we had known how their careers were going to develop subsequently, we could argue that Edgar should have been playing earlier. But we didn't know that after 1987.

-- MWE
   149. Tango Posted: August 05, 2008 at 07:47 PM (#2891256)
namely, do you really mean that if a player has a good enough single season (for whatever value of "good enough"), you'd call him a HOFer. And the second question: does it really not matter to you if, outside that great season, he did well or badly or not at all.


Theoretically, it's possible that one season a HOF makes. It hasn't happened yet, not even close.

And correct, once someone has established his HOF creds, he can't lose that cred.

(I think it's answering the wrong question -- I do think the Hall is about value, not "talent.")


To be clear, I'm specific in my definition of talent. It's not just some guy who has the tools or whatever. But, the guy whose actual performance can only have been done by someone of a specific level of talent.

That said, that's my question as to what makes a "good" HOF.

You seem to be worried that I'm asking your answer because then I can say, "Well, so-and-so's a HOFer (or not) because Tango says so." But I'm asking the question because I'm trying to explore how far you take this "framework."


I'm not worried. Since the framework is quite flexible, it allows one person to say his benchline is performance from .340 and another to say performance from .220, such that in one case, one Bonds season is enough, and in another case, it requires 5 Bonds seasons.

It's a way of talking about peak and longevity without making it a hodgepodge solution.

Honestly, I thought that my explanation in the blog was quite clear, since I give multiple scenarios and the implications for each scenario. If you subscribe to the .260 scenario, then you know exactly how many PA and at what performance level constitutes borderline.

Perhaps I didn't make it clear enough...
   150. David Nieporent (now, with children) Posted: August 05, 2008 at 07:55 PM (#2891281)
Honestly, I thought that my explanation in the blog was quite clear, since I give multiple scenarios and the implications for each scenario. If you subscribe to the .260 scenario, then you know exactly how many PA and at what performance level constitutes borderline.

Perhaps I didn't make it clear enough...
No, it was clear -- but the way to test any methodology is to look at border cases. If I say, "My criteria for the HOF is 70 WAR. Period." then it's a fair question to say, "Do you mean if someone has 3 WAR for 25 years, you'll put them in?" If I say, "Yes," you know I really meant "Period." If I say, "No, the player has to be better than that," then you'll know I'm really implicitly mixing in some sort of peak criterion there.
To be clear, I'm specific in my definition of talent. It's not just some guy who has the tools or whatever. But, the guy whose actual performance can only have been done by someone of a specific level of talent.
Right, I understand your definition of talent in this context; I was just using "talent" as shorthand. I'm still disagreeing with it.
   151. Tango Posted: August 05, 2008 at 08:02 PM (#2891316)
If let's say you subscribe to the .260 wOBA level as described in my blog, here is an extended view of the performance level required and number of years to be considered around the bubble level for HOF:
wOBA seasons WAR WAA
0.800 0.8 22.0 20.4
0.700 1.2 26.5 24.1
0.600 2.0 33.2 29.1
0.500 4.0 44.3 36.0
0.480 4.7 47.3 37.4
0.460 5.7 50.8 38.8
0.440 7.1 54.7 39.9
0.420 8.9 59.1 40.4
0.400 11.7 64.0 39.6
0.380 15.9 69.1 35.9
0.360 22.9 73.7 25.9
0.340 35.8 74.7 0.0

All of these performances are at the same number of standard deviations from .260.

wOBA is equivalent to OBP, with the corresponding expected SLG. So, a .400 OBP implies a .500 SLG, etc.

Bonds' 4 seasons were probably in the .500 range, so by this assumption (the .260 level) that's enough to get him in.

WAR is wins above replacement and WAA is wins above average.

As you can see, the WAA is a pretty good stand-in for what we want.

That said, if you don't like the numbers in this table, perhaps you prefer a .220 comparison point (where you'd need say 5 Bonds seasons) or a .300 comparison point, as shown in the blog entry. That's why this is pretty flexible, in the way that say WAR would be flexible by simply lowering or increasing whatever you think the replacement level should be.
   152. Srul Itza Posted: August 05, 2008 at 08:04 PM (#2891323)
I've gone through the thread, and the links, and aren't we right back where we started?

It is still the old peak vs. prime. We have a "new framework" for discussing it, but not the numbers to come to conclusions. Without the new framework, we are still left with going by "feel" to answer the questions:

How high a peak do you need to see, to make up for a short or shorter career?

How long a good and useful career do you need to see, to make up for the lack of a very high peak/prime?

Tango's answer probably puts him at the edge of opinions -- if the peak is high enough, he does not need to see more, provided the peak convinces him that he was looking at a rare and extraordinary talent.

But for most people, raw talent is not enough, however breathtaking it may be. They need to see some durability and consistency, in seasons and over time.

Still, he raises an interesting question. Koufax is in the real Hall of Fame for a very specific run of excellence. If that was all he ever did, would he deserve to be in the Hall of Fame if you were the one defining it? If 1997-2003 were the sum total of Pedro's career, would he deserve to be in the Hall of Fame if you were the one defining it?

What is the absolute minimum number of years of truly stellar performance that would be sufficient for you to name somebody a Hall of Famer, if you were the one defining the Hall?

What is the absolute minimum number of years of absolutely impossible performance (ERA+ of 300 in 250 innings, or OPS+ of 250 in 700 PA) that would be sufficient for you to name somebody a Hall of Famer, if you were the one defining the Hall?

Or are you stuck on 10 years?

Having re-thought it, I lean a little toward Tango on this issue for my Personal Hall of Fame, but I would not change the real Hall of Fame to match. I would want at least 5 years of spectacular, or even impossible performance, for my Hall. But for the real Hall of Fame, I want more.
   153. CrosbyBird Posted: August 05, 2008 at 08:24 PM (#2891435)
Koufax is in the real Hall of Fame for a very specific run of excellence. If that was all he ever did, would he deserve to be in the Hall of Fame if you were the one defining it? If 1997-2003 were the sum total of Pedro's career, would he deserve to be in the Hall of Fame if you were the one defining it?

For Koufax, maybe, and for Martinez, definitely. I'd say around 1500 IP is probably the bare minimum length for any pitcher in my personal HOF, which is why I wouldn't induct Rivera right now. I think Pedro's 1997-2003 is more impressive than Koufax's 1961-1966.

What is the absolute minimum number of years of truly stellar performance that would be sufficient for you to name somebody a Hall of Famer, if you were the one defining the Hall?

I'd say between 6-10 seasons, depending on how stellar. Bonds/Pedro peaks, maybe 6 is enough.

What is the absolute minimum number of years of absolutely impossible performance (ERA+ of 300 in 250 innings, or OPS+ of 250 in 700 PA) that would be sufficient for you to name somebody a Hall of Famer, if you were the one defining the Hall?

I don't know how to really approach this question. Maybe one season of a pitcher starting every game, and throwing a 27 K CG shutout in each one going 34-0 or a hitter getting either a HR or BB in each of 600 PA might be enough. A player demonstrating talent so far above any player in MLB history is outside the realm of what any metric I'd consider for the HOF discussion is designed to deal with. I'm not sure there's much value in designing a metric that gets so specific at the high end... it's like asking how you'd approach a player who put up league-average results for 50 consecutive seasons... I guess that guy has to be a HOFer too with 7500+ H.
   154. BDC Posted: August 05, 2008 at 08:28 PM (#2891455)
Koufax is in the real Hall of Fame for a very specific run of excellence (Srul, #152)

Who's the best example of someone who established himself in the majors as a really superior talent, but didn't come close to sustaining it long enough to build a decent Hall of Fame case? Lefty O'Doul? People cite Pete Reiser, but even when he was briefly great Reiser didn't stay in the lineup well. His injury-prone behavior was part of his limitation as well as part of his greatness. O'Doul, OTOH, could seriously hit at a Pujolsy level, and was durable, and aged well, but didn't do his thing for very long in the major leagues.

If you favor peak, does a guy like Sutton make the HOF? How about Nolan Ryan? (BLB #124)

Depends on how large the Hall. Sutton was one of the half-dozen best NL starters for several years in the early 1970s. He's far from the weakest "peak" pitcher in the real Hall.

Ryan is something else altogether. He defines extremes of possible performance at peak and career levels as well as the single-game "signature" level. He's so unusual that he defines the Hall, in a sense, rather than the Hall defining him
   155. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 05, 2008 at 09:27 PM (#2891643)
Blackadder: total 3-straight-year runs as good as Giambi's since 1893: Aaron, Banks, Bonds, Cobb, Collins, Foxx, Gehrig, Hornsby, Joe Jackson, Jennings, Lajoie, Mantle, Mays, Morgan, Musial, Pujols, Ripken, Jackie Robinson, A-Rod, Ruth, Schmidt, Speaker, Vaughan, Wagner, Williams, Yount. So of the comparison group, 100% are Hall of Merit inductees, and, leaving Pujols out, 75-80% are what I would call the Hall's "inner circle" (everyone but Banks, Jackson, Jennings, Robinson, Yount, and Vaughan is right on the edge). By the "demonstrated talent" standard, Giambi's 2000-02 should put him without a doubt among the elite, chemical enhancement aside.

Greg Maddux School of Reflexive Profanity, Sosa's '01 (11.2 of my WARP) was far superior to anything Ott did (Ott's best was 8.7). Yes, his '98-'00 and '02 were similar to Abreu's best, but that's a damn high standard! And you're being too harsh on the side years--he played at a clear All-Star level in both strike years, and was doing the same in 1996 before he got hurt. Sosa really had a whale of a peak. For his career, I have him at 58.8 WARP, which is definitely light for the Hall, but the peak definitely gets him in for me. Giles doesn't have quiiite as many WARP, he's mid-50s, but he's still playing, and he definitely has more career if you give him "blocking" credit. Giles lacks an astonishing year like Sosa's '01. I support Giles for the Hall.
   156. Tango Posted: August 05, 2008 at 10:05 PM (#2891684)
By the "demonstrated talent" standard, Giambi's 2000-02 should put him without a doubt among the elite


As I've shown, depending on the baseline you choose, you could need 4 Bonds-like seasons to make him borderline, so the 3 Giambi season simply won't be enough. Again, it all depends on the baseline number you choose.
   157. Blackadder Posted: August 06, 2008 at 11:00 AM (#2892438)
I was thinking about this last night, and I believe when you actually examine the math there is much less difference between Dan's and Tango's positions are not that different.

What, precisely, is Tango's method? He starts with a hypothetical player, say, Joe Smith, of a certain level, say average, or replacement, or AA. Then he "rates" the actual baseball player, Tom Awesome, by asking how likely it is that Joe Smith could have put up the performance that Tom Awesome did. Since the probabilities are obviously going to be incredibly small if Tom Awesome is even a marginal hall of fame player, it is easier to express them in terms of standard deviations. In other words, Tango rates players by asking how many standard deviations their performance is from the mean performance of some baseline player.

The stat Tango uses is wOBA, which is basically just a version of OBA that gets the relative weights of the offensive events right. It has the nice property that it is linearly related to the number of runs a player adds, hence the number of (offensive) wins. If a player has a fixed wOBA and has PA plate appearances, the standard deviation of his sample wOBA is

SQRT(wOBA*(1.1-wOBA)/PA)

(I am not sure why it is 1.1 instead of 1, but I think it is because wOBA is rescaled.) To make things readable, let's assume that we use the baseline of .300 wOBA, which is a replacement level hitter. Then if Tom Awesome has a given wOBA in a given number of PA, his Standard Deviations above what the .300 wOBA guy does is

(wOBA-.300)/(SQRT(.300*(1.1-.300)/PA))

Now, since we are only interested in comparing players, we can ignore multiplicative constants in this formula (they won't change any ordinal rankings). In particular, we can ignore the 1/SQRT(.300*(1.1-.300)), since we are comparing everyone to a .300 hitter. The formula becomes

(wOBA-.300)*SQRT(PA)

wOBA-.300 is proportional to batting runs above replacement per plate appearance. This, in turn, is proportional to batting wins above replacement per full season, assuming seasons of fixed length. By taking any player's fielding, positional, and baserunning value and "reassigning" it to his hitting, wOBA-.300 can be taken to be proportional to WARr, the players WAR per season. PA is proportional, by definition, to the number of seasons. So the formula becomes

WARr*SQRT(N)

Where N is the number of seasons player. Squaring this, which won't change any relative rankings, we get

N*(WARr)^2

As Tango's definition of a players value: the number of seasons played, times the square of his rate of accumulating WAR. But this is REMARKABLY similar to Dan R's old salary estimator! Indeed, if you apply Tango's value formula to each individual season (so that N<=1) and add up the results, you precisely get Dan R's formula. Applying Tango's formula to a player's entire career, or prime, is not exactly the same as applying it season by season and adding; the later method will, for example, favor players who were more inconsistent in their value. Still, the errors are not that great.

So what Tango's method with the .300 baseline does is essentially rate each season by the square of its WAR. Weighting seasons non-linearly in this manner has been fairly common in HOM discussions as way to seamlessly blend peak and career considerations. Dan R, as I mentioned, used an exponent of 2, and now uses 1.5. Joe Dimino, in his Pennants Added method, uses something more like 1.25. Although no matter the baseline Tango's methods are always at root quadratic, they are well approximated by other exponents.

For instance, consider Tango's method with a .260 wOBA baseline. .040 points of wOBA is about two wins over a whole season, so running through the same calculation as before with the new baseline we get the formula

(WARr+2)^2*N

for a players value. In the range [4,11], which is the relevant range of WARr values for HOF contenders (anyone less is probably not a reasonable candidate, and there aren't many people higher) the function

(x+2)^2

is very well approximated by

4.5*x^(1.5)

Thus, the .260 baseline is very close to using an exponent of 1.5 to weight seasons, which Dan R currently does (there are other differences here; Dan currently considers in-season durability, which this does not). The .220 baseline is about an exponent of 1.2, which is not that far from Pennants Added.

Thus, instead of thinking "what is my baseline", you can think "with what exponent do I weight each season"? Personally, I like the 1.5 exponent, or the .260 baseline. But for people who want to rank players solely by career WAR, there is a baseline (in fact, infinitely many) that, in the [4,11] WAR range, are well approximated by a linear function.

So Tango is not really that much of a heretic after all!
   158. Tango Posted: August 06, 2008 at 11:55 AM (#2892444)
{clap clap clap}

Fantastic! G-dd-mn it, here we are talking about the same d-mn thing, and all we needed was a translator.

Well done, sir.
   159. Blackadder Posted: August 06, 2008 at 12:42 PM (#2892463)
Thanks Tango!
   160. Chris Dial Posted: August 06, 2008 at 01:12 PM (#2892478)
I second. That was a terrific explanation, Blackadder.

But I am not sure it pushes Srul's point aside (it settles one part but not the other).

I would have some questions because I don't know the WAR answers - what about Charlie Keller? Should he be a HOFer (assuming #151)? Benny Kauff?
   161. Tango Posted: August 06, 2008 at 01:49 PM (#2892504)
One tiny correction which I noted at my blog, "(wOBA-.300)" would itself need to be multiplied by PA to make it proportional to WAR. So, the effect is that the "N*(WARr)^2" really becomes cubic. Nonetheless, just a matter of playing with the baseline to see what Dan's 1.5 and Joe's 1.2 aligns to.
   162. Tango Posted: August 06, 2008 at 01:53 PM (#2892512)
Then again... if you consider a "season" to really be "PA/650" or some such, then Blackadder is right. Since most of the HOF we are talking about is lined up so that they all have about the same number of PA per season, I guess what Blackadder has is correct. I'm sure he can set us straight on this issue...
   163. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 06, 2008 at 02:15 PM (#2892530)
Blackadder, you're the man.
   164. Blackadder Posted: August 06, 2008 at 02:49 PM (#2892569)
Tango, I am not sure I agree with what you posted on your blog; I should have been more clear in defining my terms. Let me say how I see things, and we'll see what you think:

The basic formula was

(wOBA-.300)*SQRT(PA)

Now, SQRT(PA)=PA^(1/2)=PA*PA^(-1/2) and PA*(wOBA-.300) = WAR/1.15

So we can factor out the PA, and ignore the 1.15 (since again it just a factor).

So (wOBA-.300)*(PA^(1/2))=PA*(wOBA-.300)*PA^(-1/2) ~ WAR * PA^(-1/2)

Squaring, we get the formula

WAR^2/PA

Now, this is a weird formula; the dividing by PA looks funny. To make it look better, define

WARr := WAR*650/PA,

the rate at which WAR is accumulated per complete season. Also define

N = PA/650, the number of complete seasons played. Then we have

WAR^2/PA = (WAR/PA)^2 * PA = 650 * (WARr)^2 * N ~ (WARr)^2 * N

This is actually identical to the formula that Dan R used to use to value an individual season: he figured the WAR, figured the fraction of the season played SFrac, and the valued the season as SFrac*(WAR/SFrac)^2.

He and Joe compute value differently today; now, they just take each season's WAR and raise it to some power, and add up the totals. On the assumptions that

1)Each player plays a full season every year; and

2)Each player's performance level is static

This will EXACTLY agree with Tango's approach. Of course, neither is true, but 1) is pretty close to true for HOF players, as Tango mentions in the previous post and while 2) is false, the distortion introduced is not that severe, especially since players tend to "vary" around their mean performance to a roughly similar extent. There would be some players whose relative positions would change depending on the method used, but there shouldn't be anything too severe.

Does that clear things up? Are there any problems?

Chris: You are right, this does not clear up the peak/career issue. Nothing really could; it is really a choice one has to make. What this does do is establish a better framework for conceiving of these issues than the arbitrary "3 year peak" or "7 year prime" or whatever; if you want your rankings to be peak-heavy, use an exponent of 2 or higher; if you want them to be career-heavy, use 1.2 or lower; if you want something in between, use 1.5. I like 1.5, and by some scattered remarks it seems like Tango does as well.

Charlie Keller, as someone who was very very good for a very short amount of time, depends very heavily on your weight. By straight WAR he is not deserving, of course. In Dan's rankings using the 1.5 exponent, I think Keller was around the 110-120th best position player ever. If your hall is the size of the real one there is certainly room for Keller; however, I recall Chris being a "small-hall" guy, so he probably wouldn't support Keller anyway. FWIW, he is in the Hall of Merit.

I really should learn the virtue of brevity :)
   165. Chris Dial Posted: August 06, 2008 at 03:11 PM (#2892599)
Thanks, Blackadder. You recall correctly - I'm a very small Hall guy, but I would generally have room for the 110-120th best player. Plus Keller was awesome for my DMB teams. And I just have always been a fan.

But no, he has four great great seasons, two great ones, and well, he was pretty good. So, he wouldn't make it. I value longevity as well as peak (beause I would include Koufax, I think).

I really should learn the virtue of brevity

Disagree. Your posts have tremendous clarity - don't change a thing.
   166. Tango Posted: August 06, 2008 at 03:14 PM (#2892604)
Right, if you consider the "N" as PA/650, then we are ok.

I was thinking of situations where you have some seasons of 300 PA and others of 700 and so on. And for the players we are considering, this is really not an issue. So, if you are treating a season as really an "effective season" (PA/650), then we're good.

Great job!
   167. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 06, 2008 at 03:29 PM (#2892626)
I used to plug the WARP rate per season into my salary estimator; now I do the seasonal total. The reason is that using rate in the estimator had the effect of raising my replacement level above what is empirically accurate. Say one guy has 5 WARP in half a season, and another has 6 WARP in a full season. 10^1.5 / 2 = 15.8, while 6^1.5 = 14.7. So that meant I was valuing a 5-WARP season above a 6-WARP season, which just can't be right.
   168. Blackadder Posted: August 06, 2008 at 03:48 PM (#2892651)
Chris, thanks for the kind words. Dan's list is at post 422 here:

http://www.baseballthinkfactory.org/
files/hall_of_merit/discussion/
dan_rosenhecks_warp_data/P400/ (sorry, I don't want to break the page!)

Keller is at number 112, although that is with war credit; without it, he obviously drops considerably. He was really amazingly good; even with his super short career, he has almost as many Palmer batting wins as Sammy Sosa! I can see not wanting him, but he does have a Koufax-style case, albeit a much weaker one than Koufax.

Dan, that's a good point, but don't you think it is reasonable to give Bonds the MVP in 2003, despite being a tiny bit behind Albert Pujols in WARP, because of just how much better he was when he was when on the field? Now that I mention it, I think I got this example from you!
   169. Tango Posted: August 06, 2008 at 03:48 PM (#2892652)
What Blackadder is suggesting in your case is that since you have 11 WARP in 1.5 seasons, then you have a rate of 7.33 WARP per season at 1.5 seasons.

7.333 raised to the power of 1.5 (unfortunate coincidence) times 1.5 seasons gives you 29.8 "blackadder wins".

Two 6 WARP seasons (12 WARP in 2.0 seasons) gives you a total of 29.4 "blackadder wins".

From your standpoint, it still doesn't make sense right?

How would my approach see these two seasons? Let's say the 6 WARP season is a .400 wOBA player in 700 PA and the 5 WARP half-season is a .460 wOBA player in 350 PA.

If you have two full 6.0 WARP seasons (total of 12 in 1400 PA), that's 11.2 SD from the .260 baseline.

If you have a total of 11 WARP on 1050 PA (weighted wOBA of .420), that's 11.1 SD from the .260 baseline.

In both cases, be it the blackadder approach or my approach, there are very close. Except my approach gives a slightly "better" answer, in that you expected (or wanted?) the guy with the 12 WARP in 2.0 seasons to come out ahead of the guy with 11 WARP in 1.5 seasons.
   170. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 06, 2008 at 04:11 PM (#2892686)
Blackadder, yes, absolutely, the Bonds/Pujols '03 MVP vote was what I was looking at when I was first working on the salary estimator. Quite simply, I've changed my mind. No, I no longer think it is reasonable to give it to Bonds over Pujols. Otherwise, why do we bother calculating replacement level at all? I'm the one who always gets in nasty spats in HoM discussions saying that replacement level is not some theoretical/philosophical "voter preference" issue like peak vs. career, it's a hard and fast empirical fact. If I give the award to Bonds over Pujols, I am contradicting my own argument. Bonds plus a replacement LF for the games he missed in '03 is not as valuable as Pujols plus a replacement LF for the games he missed in '03.
   171. Tango Posted: August 06, 2008 at 04:55 PM (#2892736)
Some numbers so I (and the rest of us) can follow along: in 2003, Bonds had 550 PA and Pujols had 685. For the sake of argument, and to make the PA disparity wider, let's say you have Bonds at 500 PA and Pujols at 700 PA. Let's also say that their WAR are the same (say +10 WAR).

The entire argument of replacement level is that Bonds plus Joe Schmoe at 200 PA would match Pujols in 700 PA.

Anyway, this would imply a wOBA of .530 for Bonds and .464 for Pujols. In terms of "impressiveness", which one is an indicator of better talent? Using .260 as the baseline, the .530 comes out at being 12.9 SD while the .464 comes out as being 11.5 SD. In fact, only if you were to drop the baseline comparison all the way to wOBA = .100 does Pujols overtake Bonds.

Nonetheless, I have no problem awarding Pujols an MVP here over Bonds, or vice versa, since they were both 10 WAR. MVP is for actual performance. I (and some) see the HOF as something somewhat different.

It's the difference between the best-grossing film and the best-film of the year.

Anyway, my method, to get both players as 12 SD from the .260 level, would have Bonds at .511 wOBA in 500 PA and Pujols at .472 wOBA in 700 PA. In terms of WAR, that puts Bonds at +9.2 wins and Pujols at +10.5 wins. That is, in terms of "indications of true talent", being 10.5 WAR in 700 PA is the same as 9.2 WAR in 500 PA.

Using Blackadder's approach, 10.5 WAR in 700 PA, with the 1.5 exponent is straightforward: 34.0 "blackadder wins". The 9.2 WAR in 71.4% of the season is: 9.2 divided by .714, then raise to the power of 1.5, then multiply by .714, which is 33.0 "blackadder wins". If you put it back to the exponent of 1/1.5, you get Pujols back to 10.4 wins, naturally, and Bonds at 10.3 "blackadder unexponentialized wins".

In this particular case, if Blackadder had used an exponent of 1.65, they'd be even.

All said and done.... it depends.
   172. AROM Posted: August 06, 2008 at 05:11 PM (#2892758)
In both cases, be it the blackadder approach or my approach, there are very close. Except my approach gives a slightly "better" answer, in that you expected (or wanted?) the guy with the 12 WARP in 2.0 seasons to come out ahead of the guy with 11 WARP in 1.5 seasons.


Who cares what we want to see? I don't see any value in engineering a method to fit a conclusion we've already come to.

We're asking two differnent questions here: Who has demonstrated greatness and who had the most value? For the 2nd question, the 12 warp guy wins. But on the first question it's close, and if he had put up the 11 warp in one season and then did nothing the second, the methods laid out here would see him as the greater demonstrated talent.

A while back Chris Dial had a post on how to quanitify greatness. I didn't have any suggestions then, but the Tango and Blackadder methods seem to be very good approaches to the question.
   173. Chris Dial Posted: August 06, 2008 at 05:31 PM (#2892776)
was what I was looking at when I was first working on the salary estimator. Quite simply, I've changed my mind. No, I no longer think it is reasonable to give it to Bonds over Pujols. Otherwise, why do we bother calculating replacement level at all? I'm the one who always gets in nasty spats in HoM discussions saying that replacement level is not some theoretical/philosophical "voter preference" issue like peak vs. career, it's a hard and fast empirical fact.
Not to take this thread elsewhere, but replacement level is a "fact"? How are you defining it? I've tried in a specific research article here to define it, but AFAICT it is a "theoretical/philosophical" issue due to ill-defined data to create the hard fact.

AROM, I think you are right - that can define greatness. Which player would get into the HOF using (either) system? Is he the greatest player?
   174. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 06, 2008 at 05:57 PM (#2892809)
Chris Dial, I'm referring to the studies that aggregate the performance of actual replacement players (minor league free agents and the like). Nate Silver's Freely Available Talent figures are the ones I use. Yes, there are certainly angels-on-the-head-of-a-pin subtleties, but my contention in the HoM is that anything too far from 80% of positional average (like Win Shares or BP WARP) is just factually wrong.
   175. Blackadder Posted: August 06, 2008 at 06:01 PM (#2892815)
Those are very good points; of course we have to distinguish value from greatness. If not, it's hard to see how you don't just end up rating people by career WAR, since that is the total "value" that a player contributes over his career. So maybe it is right to say that Pujols was more valuable than Bonds, and hence deserved the MVP, while Bonds' season was greater, and counts for more than Pujols' season for the purposes of HOF deliberations (it is not obvious that this is so, but it at least seems reasonable. I don't have Dan's spreadsheet handy, but I recall both Pujols and Bonds being around 10 WARP that year, as Tango suggests, with Pujols a little higher and Bonds a little lower).

I think, by the way, that the fact that independent and seemingly totally distinct attempts to quantify the greatness of players ended up being almost equivalent is some sign that they are on the right track...
   176. Chris Dial Posted: August 06, 2008 at 06:03 PM (#2892818)
Chris Dial, I'm referring to the studies that aggregate the performance of actual replacement players (minor league free agents and the like).
Yes, I did that exact study here. I don't really know Nate's study (I compared to Woolner's work a decade before for VORP), but it'd be interesting to see how close our data matched. I also tried to look at replacement defense.
   177. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 06, 2008 at 06:13 PM (#2892831)
Blackadder, yes, Pujols was juust over Bonds, right around 10, that year. Some of that is credit for Pujols's defense though, which I imagine is deserved but would have to check against the PBP's.

Nate's study is at http://www.baseballprospectus.com/article.php?articleid=4891.
   178. Tango Posted: August 06, 2008 at 06:42 PM (#2892891)
I don't see any value in engineering a method to fit a conclusion we've already come to.


I think that is the exact thing we want to do! Greatness would be on par to pornography: we know it when we see it. Most baseball fans would consider Koufax a sure-fire HOF, even part of a small Hall where you have only half as many pitchers as currently in the hall.

So, the attempt here is to create a model that reflects that (perception of) reality. Once you are satisfied that the model works with the data points you can think of, then you let the model run on all the data available, past, and future. If it's a well-built model, then when new extreme cases come along, we have a tested model that is ready to answer the question.

This is no different than trying to figure out say your salary at work. You put in all the relevant parameters, including whatever "intuition" and "intangible" things you want to tangibilize, and you come up with a salary figure. When someone comes along to explain, complain or you have someone new coming on board, you'll be consistent in how you set the salary.
   179. Tango Posted: August 06, 2008 at 06:48 PM (#2892903)
Dan/177: I already had this talk with Dan and others at my blog. I'm not a big fan of the approach (just a small fan) as that method has selective sampling issues. Certainly the result, insofar as the SS, doesn't make sense to me.

Further proof that the method won't hold is if you try to repeat that process for pitchers. You will come away with the replacement level pitcher ERA to be barely worse than the league average ERA.

My method is to start with the positional difference as so:
+1.0 C
+0.5 SS/CF
+0.0 2B/3B
-0.5 LF/RF
-1.0 1B
-2.0 DH

And add in a replacement level of 2.25 wins per 162G.

The results will be somewhat comparable to what Nate has, but the SS is the big stickler here.

***

In a more general sense, Dan's point was that once you've established your replacement level, however and whatever that is, you can't then just discard it if you are not happy with the results. The case in point was the Pujols/Bonds discussion. As he said, why have replacement level as your framework if you are going to discard it when it suits you.
   180. Chris Dial Posted: August 06, 2008 at 07:39 PM (#2893016)
So, the attempt here is to create a model that reflects that (perception of) reality. Once you are satisfied that the model works with the data points you can think of, then you let the model run on all the data available, past, and future. If it's a well-built model, then when new extreme cases come along, we have a tested model that is ready to answer the question
Well, I do it slightly differently. I say "What constitutes 'greatness'?" High level of performance and for a long time. I wouldn't give the CYA to a pitcher that went 12-0 with a 0 era before the break, then missed the rest of the season.

So I design the framework based on what I think the definition of greatness is, then see who turns up. Sure, if Bob tewksbury turns up and Gibson doesn't, well, I have to re-think my criteria or how I worked the data, but it just has to pass the smell test.

I actually think ERA+ times IP give s a great idea of great seasons. Of course, I tend to start counting after the HR became vogue (1920 or so). I am not sure if OPS+ times PA works, but I suspect it does.
   181. Chris Dial Posted: August 06, 2008 at 07:41 PM (#2893018)
In a more general sense, Dan's point was that once you've established your replacement level, however and whatever that is, you can't then just discard it if you are not happy with the results. The case in point was the Pujols/Bonds discussion. As he said, why have replacement level as your framework if you are going to discard it when it suits you.
That's all well and good, but I don't really care for RL based ont eh wiggly baseline between people. That is - is your baseline the same as mine? Maybe, maybe not, but our "average" *is* the same. Yes, an average player has value for a team, but he isn't valuable against a player above average. As long as you draw the same lines, I think it is okay.
   182. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 06, 2008 at 08:03 PM (#2893071)
Tango, how were those positional weights derived? SS as equal to CF seems nuts to me...
   183. Tango Posted: August 06, 2008 at 08:42 PM (#2893126)
I use UZR for players who played multiple positions, 2003-07. I've also done it in the past for 1999-03 to similar effect.

The ones I stand completely behind are the OF ones, relative to each other. Not only do I have an overwhelming amount of data there, the skillsets required to play the three positions overlap to a great deal. And the result there is that the average fielder in CF is +1.0 wins better than the average fielder in the corners (who themselves are fairly even). Managers are smart here in that, over a long period of time, you will find that the offensive runs created are about +1.0 wins higher by the corner OF than the CF. So, we've got great equilibrium here (though not necessarily every year).

For the IF (2B,SS,3B), relative to each other, I'm not as strong on those, since selective sampling issues will rear its ugly head here. Guys go SS to 3B and SS to 2B, but they don't necessarily go 3B to SS. There's a period of adjustment (familiarity) since the skillset don't necessarily overlap. That said, 2B and 3B are very close, and any time we see how SS do at 3B or 2B, they do not standout as one would expect. The net result is about a 0.5 win gap between SS and 2B/3B.

Now, the dangerous part: comparing these three IF positions to the three OF positions. Since almost all moves are IF to OF, we have the familiarity issue. And, the skillsets required are not close to the same. And of course, selective sampling.

But, we do have one savior: firstbaseman. Alot of players move 3B to 1B and 2B to 1B. And the net result there is that those IF are +1.0 wins above 1B. And, we have alot of corner OF that move to 1B. The net result there is that the corner OF are +0.5 wins above 1B.

We do have some 2B/3B that move to LF/RF, and it's somewhat consistent with the above pattern.

However, let's say you don't buy it. You buy the CF/LF/RF (as you should), and you can kind of buy the SS/2B/3B. But, you think that 2B/3B/CF should be even. If you do that, you get these kind of relative rankings, setting SS as "+0.75":
+1.25 C (just forcing this one in)
+0.75 SS
+0.25 2B/3B/CF
-0.75 LF/RF
-1.25 1B

That could work. Now you have to accept that the 2B/3B is 1.5 wins ahead of 1B. The data doesn't support that, but that's what you get if you want to bring down the OF and up the other IF.

You have certain constraints that you have to deal with. The OF ones we know and we can live with and accept. The 1B should be worse than the corner OF. The SS should be a little better than the 2B/3B, but not that much better. Certainly not the gap as CF compared to LF/RF.

The last chart is really about as far as you can possibly go (catcher excluded).
   184. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 06, 2008 at 08:54 PM (#2893138)
I see, these are based on position-switchers. The problem is that you almost never see guys go the other way (from playing other positions to SS, rather than the other way around). When SS move to 2B/3B or to CF, it's for a reason--they weren't cutting it at SS. Therefore it's not surprising that they would only be .5 wins better, relative to positional average, after the move. But what would happen if you moved a 2B to SS, say? I suspect he'd lose FAR more than 0.5 wins a year, because he most likely simply doesn't have the arm to play the position (if he did, he never would have been moved off in the first place; the majority of MLB 2B are failed SS). Here are my numbers, derived from Nate's study but anchored to a moving average of the worst regulars at each position (see the thread on my WARP for details), measured in wins above average per 162 games compiled by replacement players at each position:

SS -2.6
C -2.4
3B -1.8
2B -1.7
CF -1.5
LF/RF -0.9
1B -0.3
DH 0.0

Note that these are based on 2001-05; it could be that the spectrum has shifted in the last three seasons. Back in the 80s, SS was WAY scarcer than C, and that gap has been closing ever since.
   185. Tango Posted: August 06, 2008 at 09:15 PM (#2893171)
If I add 1.5 from each of those numbers and flip the sign, you get:
+1.1 SS
+0.9 C
+0.3 3B
+0.2 2B
+0.0 CF
-0.6 LF/RF
-1.2 1B

That puts them on the same scale as mine (0 is average).

As you can see, we have fairly good agreement on all positions, except SS and CF. You are suggesting that CF is only 0.6 wins ahead of LF/RF, which really goes against the mountain of data here. The position-switchers are so bountiful, I don't see how you can justify going against that method and data.

You've got the SS as +0.85 wins ahead of 2B/3B. That's possible. It's stretching it, but you could be right.

As for SS/3B/2B compared to LF/CF/RF, you've got them as +0.9 wins ahead. I've got them as +0.3 wins ahead with my original chart and +0.8 wins with my second chart.

If you can knock down the SS by 0.3 wins and knock up the CF by 0.3 wins, we'll both be happy.
   186. Francoeur Sans Gages (AlouGoodbye) Posted: August 06, 2008 at 09:21 PM (#2893185)
2008 MLB SS hit .266/.321/.382, CF hit .266/.331/.416 (so far)
2007 MLB SS hit .275/.330/.407, CF hit .272/.338/.420
2006 MLB SS hit .276/.332/.408, CF hit .269/.335/.427
2005 MLB SS hit .270/.325/.394, CF hit .272/.334/.423
2004 MLB SS hit .271/.323/.407, CF hit .273/.337/.437

I stop there but I could go on. CF outhits SS every year by a nice margin. The positions are clearly not equal. Over the past 5 years, on average SSs give up 34 points of OPS to CFs. The data tells us what common sense suggests, that you need to give extra value to a guy playing shortstop than a guy playing centre. In fact SS is a consistently (well) below average position for offense, CF is consistently average or a little above.

I believe that the reason for your error is you are focused on players who move from one position to another. The problem is that players who throw lefthanded can never play infield positions, except first base, however well they can field. In fact I have read elsewhere that an element of what we think of as the "defensive spectrum" is simply the fact that lefties can't play the tough defensive positions.

34 points of OPS is somewhere between half a win and one win. Call it half a win to be conservative.
   187. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 06, 2008 at 09:31 PM (#2893204)
Tango, I am sure you're right that the position-switcher data yields one win for SS vs. CF. However, the freely available talent method yields 0.6. What this suggests is that teams are not allocating their outfielders optimally between CF and the corners--that they would be better off, on the whole, trading some offense for defense at LF and RF. A replacement CF, per my modification of Nate's work, will field roughly at the league average and hit 6 runs worse per season than a replacement LF/RF. If you move that guy to left or right, you lose 6 runs with the bat, but according to your position-switcher data, you pick up 10 with the glove, for a gain of +4. Is that right, Tango?
   188. CrosbyBird Posted: August 06, 2008 at 09:48 PM (#2893242)
Well, I do it slightly differently. I say "What constitutes 'greatness'?" High level of performance and for a long time.

That's a coarse description of my system. Here's my response to a post on Tango's blog.

It seems to me like you’re compartmentalizing HOF standards into binary answers with little to no differentiation beyond the yes/no answer. If I’m wrong, help me out, but how is your process not like this?

Does a player have a HOF peak - Yes/No?
Does a player have a HOF length - Yes/No?
Are there externalities to consider - Yes/No?


For the extreme cases, that is a fair description of my process. If the answer to the first two questions is yes, that’s a clear HOFer. If the answer to the first two questions is no, that’s clearly not a HOFer.

If the player has one or the other (like Palmeiro having a HOF career but not peak, or Martinez having a HOF peak but not career), then it requires further qualitative review.

If the player is at all close to being in at this point, I consider the third question and see if there’s enough to push him over the threshold.

For Palmeiro, the peak is questionable. It’s lower than what I would consider to be an average HOF peak, so he’s going to need something else to get in. He has not just a HOF career, but one that is exceptionally long and productive in comparison to other HOFers (15th in career PA, 17th in career RC). That’s enough for me so I don’t go to the third question, but if I needed to, I’d consider his excellent gray ink and his extreme in-season durability as some added value.

For Martinez, the lack of accumulated counting stats is an indicator that the career is questionable. Clearly, he has a HOF peak, but is it exceptional among fellow HOFers? I think it is pretty much the average peak for a HOF player (considering missed games, defense and baserunning). He’s still really close, because the career isn’t an embarrassment. The problem is that I can’t find anything to push him over the edge. That’s why he’s out of my personal HOF.

I am doing basically the same thing Tango suggests although my methodology is less mathematical, and that’s by design. I don’t believe there is a HOF formula that works for every player so I won’t commit to “here is my baseline for zero-value, and here’s my total of wins above that value.”

I guess what I’m trying to get at is that I don’t view players on a linear curve, not peak or career value. The difference between 600 HR and 500 HR is more significant to me than the difference between 500 HR and 400 HR… it’s not simply 100 more HR, but a premium placed on going to the next tier of exceptional production.

I suppose there’s some way to provide weightings to Tango’s formula that simulate this, but they will be very era-specific and extremely hard to quantify. Since the number of players that are on the fringes is relatively small, it’s not worth the effort. Also, how far should we extrapolate? Should we consider the no-peak, 50-year career player? Should we consider the 3-year career, Bonds-2004 player?

Certainly, whether you agree with my methodology or not, you can understand the process, and have a very good idea of which players are going to be in or out.
   189. Tango Posted: August 06, 2008 at 10:08 PM (#2893258)
Alou: just because a CF is a better hitter doesn't make him a worse fielder. I have this conversation all the time. How about in the late 40s and early 50s where the CF actually out hit the 1B. Your reasoning would suggest that 1B would be better fielders.

How about a 20-yr time period in the Retrosheet era where the SS and 2B hit equally well? Your reasoning would suggest they must be equal defenders as well.

How about in the last several years where the 3B demolished the 2B with the bat? Again, this must mean that the average 2B is a much better fielder than the average 3B.

But all of these things aren't necessarily true. Certainly not acceptable on their face.

***

"SS vs. CF": I'm sure you mean "corner OF vs CF", based on the rest of your post.

I don't even think I can answer your post because I don't know what a "replacement" CF or LF is. Where would someone like Endy Chavez or Michael Bourn or really, almost every non-regular OF be?

The problem I have, especially in the OF, is that the positions are so interchangeable that talking about "replacement LF" as if it's some unique position like catcher doesn't make much sense to me. I'm sure a good portion of the replacement LF played CF at some point in their minor or major league career.

Do we accept that the single best way to compare the positions is to compare players who actually played at both positions (like you would do MLE for minors/majors or Japan/majors or NL/AL), at least for outfielders?
   190. Tango Posted: August 06, 2008 at 10:13 PM (#2893261)
The other methods described, as best as I can tell, are all based on "market efficiency", trying to make sense as to where the players are at each position, based on the talent levels and salaries and other market forces.

For the outfielders, I don't see any need to rely on a market efficiency system when we've got real-life data from real-life people who are actually telling us, with their performance, exactly how they compare to their peers. The average CF in the last 10 years is around +1.0 wins above the corner OF with the glove.

If a market efficiency system tells you otherwise, then there's an arbitrage opportunity, and nothing else.
   191. Blackadder Posted: August 06, 2008 at 10:41 PM (#2893274)
Dan can obviously speak for himself, but it seems to me that the fact that the "worst 3/8ths" regulars approach independently (from the FAT) sets the gap between SS and 2B/3B at a lot more than 0.5 wins is decent evidence that the gap really is bigger (I take this to be an implication of Dan's work; I could be wrong). The selection bias in SS to 2B/3B moves looks pretty bad to me.
   192. Francoeur Sans Gages (AlouGoodbye) Posted: August 06, 2008 at 11:04 PM (#2893287)
Alou: just because a CF is a better hitter doesn't make him a worse fielder.
I never said it did.

To simplify things dramatically, to play CF at a major league level, you have to be a really good fielder. To play SS at a major league level you be a really good fielder, and you also have to be right handed. So CF is drawing from a larger pool than SS. So we should expect CF to outhit SS, and lo and behold that's what happens. This has nothing to do with who is a "better" fielder as between CF and SS. I don't even know what that means.

Do a thought experiment where SS and CF are the only positions on the diamond. Suppose we've got 80 guys, all equally good fielders, all equally capable of playing CF and SS. Hitting is normally distributed. Forget about backups and minor leagues. We now divide them up between CF and SS, and fix the top 31 of each position in place. So we expect replacement level and league average to be equal between the two positions.

Now say that 10% of the players can't play SS, only CF (they're left-handed). So the left-handed SSs get put in the CF pile. What happens? Average and replacement-level SS goes down, because they've lost (approx) 3 players from the top 30 SSs, and are replacing them with worse hitters. Average and replacement-level CF goes up, because similarly they are gaining players and forcing out the worst ones. Note that this is nothing to do with who is a "worse" fielder as we imagined all the guys to be equally good fielders. So the thought experiment confirms what we see evidentially - we should expect CF to outhit SS.

If in addition (as some would suggest) SS is a more demanding defensive position than CF, then obviously that tilts the balance further.
How about in the late 40s and early 50s where the CF actually out hit the 1B. Your reasoning would suggest that 1B would be better fielders.
My reasoning suggests nothing of the sort. If you can play a decent CF you can (almost certainly) play a decent 1B. If you play a decent 1B you cannot (necessarily) play a decent CF. Replacement level for 1B can never be lower than replacement level for CF for this reason - the replacement level CFs can also fill in at 1B. But it is simply not the case that the replacement-level CFs can necessarily fill in at SS.

If CF is consistently outhitting 1B then it suggests that the league quality is uneven i.e. deviation in quality between players is "high." An extreme example would be in high-school baseball where the best hitter is generally the shortstop (if he's right handed) or the CF (if he's left handed) - or the pitcher.
   193. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 06, 2008 at 11:09 PM (#2893290)
Tango, you only run into the trap you are describing if you look at positional averages. Going back in time, my WARP system uses the average rate production of the worst 3/8 of MLB regulars (defined as leading team in PA compiled while playing the position) to track the evolution of the defensive spectrum, thus avoiding the effect of "star gluts" like 1980s AL SS that make a position appear artificially easy. The worst-regulars averages are much more stable and less susceptible to outliers, since if you are significantly below replacement level you won't last long.

Using this data to address the CF vs. 1B question, the late 40s/early 50s were a transition era. As far as I'm concerned, baseball's modern era began in 1947, and not just because of Jackie Robinson--because that's when you started getting leaguewide HR levels that look reasonable to a modern eye (Ralph Kiner led the NL with 23 HR in 1946, and 51 HR in 1947), and when familiar linear run estimators start to work well (the RMSE of eXtrapolated Runs or a standard version of Linear Weights shoots up pre-1947, forcing you to use something more flexible like BaseRuns or EQR to retain accuracy). In a related phenomenon, this is when the modern defensive spectrum develops: you have the first true slugging 3B in Eddie Mathews (Home Run Baker aside), cementing the 2B/3B switch, and the previous clump of 1B and all three OF positions really shakes itself out. In the 20s and 30s, those four positions were basically interchangeable. 1B was much scarcer then than it is now, I imagine because they had to field all those bunts, and because the lack of Three True Outcomes meant that immobile 1B couldn't produce as much offensive value in those days as they do now (you had to run out your doubles and triples). CF, by contrast, was deeper, I suspect because all of the asymmetric ballparks meant the defensive demands on corner OF could be quite large. As a result, you had guys like Hack Wilson in CF, and guys like Charlie Grimm at 1B. That would never happen today. This shakes itself out by the mid-50s. As a result, I have CF and 1B just about equal in 1944, but 0.9 wins apart by 1953, and 1.3 wins apart by 1959.

Yes, I meant LF/RF vs. CF. Yes, the conflict in the results given by the two methods suggests that there is indeed an arbitrage opportunity, as I suggested above. The question then arises, if the market is inefficient, should we use the empirical (market) or the theoretically optimal positional weights when assessing value?

Blackadder, I'd just like to add that you've shown yourself to be a real connoisseur of my work, and I really appreciate it. It's very encouraging to know that a thoughtful poster like yourself has taken the time to really familiarize himself with my research and apply it!
   194. Tango Posted: August 07, 2008 at 12:04 AM (#2893365)
The question then arises, if the market is inefficient, should we use the empirical (market) or the theoretically optimal positional weights when assessing value?


To me, the value is based on what's really available, and not what is currently being used as available.

If CF is consistently outhitting 1B then it suggests that the league quality is uneven i.e. deviation in quality between players is "high." An extreme example would be in high-school baseball where the best hitter is generally the shortstop (if he's right handed) or the CF (if he's left handed) - or the pitcher.


Bingo. You've echoed my many writings on this very topic.

I also don't have any issue with the rest of your post as we've talked about that issue as well on my blog.

I just don't see the point of your previous post comparing the hitting of the two positions.

But, how about we just stick to OF (LF/RF/CF) and IF (2B/SS/3B), rather than OF/IF, since we'll obviously have issues there. Let's talk about the easy ones first.

Do you think that the best way to establish the relative value of the outfielders is to compare how guys who played both positions do? It's not like it's just Endy Chavez that's doing this. There are several hundred outfielders in this position doing this. Even if you cap the weight of the career switchers so that they don't overly weight things, we're going to find that the average CF is +1.0 win over the average corner OF with the glove.
   195. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 07, 2008 at 01:17 AM (#2893560)
Well, no I don't think that in principle, although it might work that way in practice. I think that the appropriate baseline for establishing player value is replacement level, and I think that any effort to establish replacement level needs to look at real replacement players. Let's leave the technical/logistical issues (defining our sample of players, weighting their performance, etc.) aside, and just say that we have the Definitively Correct pool of replacement players. In the abstract, I'd want to just look at their performances and use that as the answer. However, I would be willing to look at position-switcher data as a check on market efficiency. If the data really show that LF/RF lose 10 runs a year with the glove moving to CF and vice versa, then the we want to just consider all replacement OF available for all three positions, adding 10 runs to the CF when we consider them in LF/RF and vice versa. E.g.:

The "market data" show CF -1.5, LF/RF -.9. If we try to play a replacement LF/RF in CF, he'd be -1.9, which is not as good as -1.5, so therefore a replacement LF/RF would be below replacement in CF. However, if we try to play a replacement CF in LF/RF, he'd be -.5, which is better than -.9. Thus, assuming there's an unlimited pool of replacement CF who can be played at the corners, our proper replacement level for LF/RF should be -.5 rather than -.9. However, it would seem weird to me that playing the OF would be worth only 0.2 runs a year more than playing 1B.
   196. Tango Posted: August 07, 2008 at 01:32 AM (#2893608)
Well, it could be then that the CF is -1.9. No reason that you'd necessarily alter the corner OF.
   197. David Concepcion de la Desviacion Estandar (Dan R) Posted: August 07, 2008 at 02:09 AM (#2893665)
No, because we know empirically that there are freely available players who can hit 1.5 wins below league average and play a league average CF. If you're right that all of those guys would be +10 LF or RF, then by definition their rep level has to be -0.5.
   198. Francoeur Sans Gages (AlouGoodbye) Posted: August 07, 2008 at 02:22 AM (#2893674)
Do you think that the best way to establish the relative value of the outfielders is to compare how guys who played both positions do? ...the average CF is +1.0 win over the average corner OF with the glove.
No, I don't. Look, take an extreme example - suppose the only positions were SS and DH. The average DH might be -10 wins at SS, but it doesn't follow that an average-hitting SS is worth 10 wins more than an average-hitting DH. And a SS would be precisely average in the field at DH (by definition) but it doesn't follow that an average-hitting SS and an average-hitting DH are equal in value.

Or, at length:

There is only so much damage you can do in the corners (that's why you hide a bad glove in the corners). But at the same time there are some players in the corners who have good gloves, but play for a team with an even better glove in CF. So, let's model it like this for a thought experiment: your fielding "rate" is doubled in CF. Let's say the average LF is +0.5 with the bat, -0.5 with the glove. The average CF is -1.0 with the bat, +0.5 with the glove. So the average LF and the average CF both work out to league average players. And if the average CF moves to LF he will be +1.0 win over the average LF with the glove (to fit in with your claim).

But this tells us nothing about the replacement level, which could look like anything. E.g. replacement level LF is 0 with the bat, -1.5 with the glove. The replacement level CF is -4.0 with the bat, +1.0 with the glove. So in this case a league-average LF is worth 1.5 wins above replacement, whereas a league average CF is worth 2 wins above replacement. Alternatively, the replacement level LF is 0 with the bat, -2 with the glove. The replacement level CF is -4 with the bat, +1.5 with the glove. So league average LF is 2 wins above replacement, league average CF is worth 1 win above replacement. So you're method tells us nothing about replacement level.

Ah, but maybe you're concerned about league average, because of situations where the league quality is uneven, so the CFs are the best hitters (or whatever). OK, suppose the league average CF is +1.0 with the bat, +0.5 with the glove. The league average LF is +0.5 with the bat, -0.5 with the glove. So if the average CF moves to LF he will be +1 win over the average LF with the glove (to fit in with your claim). So your method would give +1.0 win to the league average CF's batting line, as compared to LF - which is not enough, you need to give him +1.5 wins. Note that your method also fails to properly account for positional differences in the original example (again giving too low an adjustment for CF as compared to LF).

So I don't see how your method is useful, even on a theoretical level.
   199. Tango Posted: August 07, 2008 at 03:04 AM (#2893696)
But this tells us nothing about the replacement level, which could look like anything.


This is the point. The replacement pool for CF, LF, RF is identical... it's one pool. Managers may inefficiently allocate it (or the non-fluidity of player movement between teams may not make it very stable year-to-year), but it's the same pool.

You can put Swisher or Junior or Matsui or really any OF at all in CF, and we'll pretty much know how he'll field. And we'll know because of how similar players do when they move around.

We know the profile of players that move from LF to CF or CF to RF.

I think we are definitely not going to agree here.
   200. Langer Monk Posted: August 07, 2008 at 03:57 AM (#2893716)
I feel compelled to add a little something about Bobby Orr, since it was mentioned up-thread. Feel free to ignore this, as it would mostly be to give a little context into why he's considered one of the best ever, despite a little more than 7 and a half seasons, and finishing really shortly after turning 27.

As a third year player (turning 22 near the end of the season), he became the first defenseman ever to score 100 points in a season (1969-1970 season). In fact, the prior season had been the first in which any player ever had scored over 100 in a season. He followed that up with 5 further consecutive 100 point seasons, holding in total 6 of the 14 seasons by a defenseman with over 100 (Coffey had 5 (3 with those Gretzky Oilers), Potvin, MacInnis and Leetch each had one).

In the 1970-1971 season, he tallied 102 assists, the first ever player to do so. Only that Gretzky kid (11 times) and Lemieux did that as well. His 139 points that season were second only to teammate Phil Esposito (yeah, he did score 76 goals that year).

So, what would baseball's equivalent to Orr be? Maybe a catcher or shortstop putting up 6 consecutive MVP-type seasons in his first 8 years. Baseball age equivalent to the hockey age of 21-27 would be like 24-30.

Anyway, something to think on.
Page 2 of 3 pages  < 1 2 3 > 

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
Martin Hemner
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogMLB creates pace of game committee
(101 - 6:29pm, Sep 23)
Last: Bhaakon

NewsblogOMNICHATTER 9-23-2014
(25 - 6:28pm, Sep 23)
Last: BDC

NewsblogGene Collier: No shame in staying home for Pirates
(14 - 6:26pm, Sep 23)
Last: JE (Jason)

NewsblogChad | Have a heart! Root for the Royals
(33 - 6:23pm, Sep 23)
Last: cardsfanboy

NewsblogBrisbee: Adam Wainwright and the Jan Bradys of years past
(8 - 6:10pm, Sep 23)
Last: Matt Welch

Newsblog10 Degrees: Why WAR doesn’t always add up
(342 - 6:02pm, Sep 23)
Last: villageidiom

NewsblogOT: Politics, September, 2014: ESPN honors Daily Worker sports editor Lester Rodney
(3557 - 5:56pm, Sep 23)
Last: Joe Kehoskie

NewsblogMets near extension with GM Alderson
(38 - 5:55pm, Sep 23)
Last: ReggieThomasLives

NewsblogBaseball Caught Looking as Fouls Injure 1,750 Fans a Year
(163 - 5:47pm, Sep 23)
Last: CrosbyBird

NewsblogOT: The Soccer Thread, September 2014
(389 - 5:40pm, Sep 23)
Last: Sean Forman

NewsblogHBT: Talking head says Jeter is “a fraud” and “you are all suckers”
(265 - 5:27pm, Sep 23)
Last: Srul Itza

NewsblogOT August 2014:  Wrassle Mania I
(221 - 5:07pm, Sep 23)
Last: SouthSideRyan

NewsblogPrimer Dugout (and link of the day) 9-23-2014
(8 - 4:59pm, Sep 23)
Last: Eric J can SABER all he wants to

NewsblogCalcaterra: If Bobby Cox has anything to say about it, Fredi Gonzalez will manage the Braves for a long time
(20 - 4:59pm, Sep 23)
Last: TerpNats

NewsblogJeter Denies Gift Basket Rumors
(33 - 3:46pm, Sep 23)
Last: Barry`s_Lazy_Boy

Page rendered in 1.6069 seconds
52 querie(s) executed