Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Hall of Merit > Discussion
Hall of Merit
— A Look at Baseball's All-Time Best

Monday, June 18, 2007

Kirby Puckett

Eligible in 2001.

John (You Can Call Me Grandma) Murphy Posted: June 18, 2007 at 03:54 AM | 146 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

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

Page 2 of 2 pages  < 1 2
   101. TomH Posted: September 22, 2007 at 06:40 PM (#2537651)
(bump)

re: post 94 - need to adjust WS for the DH!!
add 5.5% to Puckett, and he beats Murphy every .. single .. matchup .. years 1 to 8
   102. sunnyday2 Posted: September 22, 2007 at 08:45 PM (#2537752)
I am re-eval "bats." Murphy doesn't do so good. Wait and see where Puck ends up ;-)
   103. TomH Posted: September 24, 2007 at 07:44 PM (#2540360)
DanR, your BWAA rate Puckett much lower than BP's for example

player.. DanR BWAA BP adj BRAA
Puckett...... 21.6 ...... 334

Dawson..... 33.8 ...... 328
Butler........ 30.5 ...... 310 (not incl 81,82)
Murphy...... 28.8 ...... 304 (not incl 92,93)

Whazzup? Is BP hosed by about 100 extra-Kirby runs? If so, why?
   104. David Concepcion de la Desviacion Estandar (Dan R) Posted: September 24, 2007 at 07:58 PM (#2540389)
I apply the DH adjustment to the replacement level (where, in my opinion, it rightly belongs), while BP applies it in the WARP1-WARP2 adjustment. BP's 216 unadjusted BRAA for Kirby are a dead ringer for my 21.6 BWAA. Dawson, Butler, and Murphy were all primarily NL players, so that factor doesn't come into play. If you want to put him on a level playing field with NL players, just multiply his career SFrac by 0.6 and add that on. Remember also that BP BRAA include SB/CS, whereas my BWAA do not (SB and CS, along with EqBR, are accounted for separately in the BRWAA column).
   105. Ron Johnson Posted: September 24, 2007 at 08:14 PM (#2540424)
TomH, BWAA has an issue with the runs per win calc. Doesn't adjust for DH. It's been a topic of discussion for over a decade.

I personally adjust players by 5% (best I can tell the adjustment should be somewhere between 5 and 6%, and I'm going to be conservative seeing as how most people don't adjust at all.)

Oddly enough this make very little difference to Puckett's ranking on Dale Stephenson's peak lists.

Moves his peak just past Roy Thomas, Andy Van Slyke and the immortal Ray Lankford. Just behind Earl Averill, Andre Dawson and Fred Lynn.

Good for the 24th best peak among CF as best I can tell (by Dale's method which is best 5 years -- don't have to be consecutive). And since Puckett's case is essentially all peak, ... well a fine player to be sure
   106. David Concepcion de la Desviacion Estandar (Dan R) Posted: September 24, 2007 at 08:22 PM (#2540438)
Ron Johnson, I have to say I don't think it's fair to call it an "issue" that BWAA doesn't adjust for the DH. Wins above league average are wins above league average, period, and it is misleading in my opinion to apply an adjustment to them so that they are no longer actually wins above league average. The correct place to adjust for the DH is in the replacement level, because a replacement level player will be 0.6 wins further below average per 162 games in the AL than in the NL. Also note that the DH adjustment should not be a percentage adjustment, but rather a simple addition of X runs/wins per game/plate appearance. The best players shouldn't be helped by it any more than the worst ones are.
   107. TomH Posted: September 24, 2007 at 08:33 PM (#2540462)
I say all of this not as a Kirby apologist - he was 15th on my ballot last week:

Kirby gains about 10 runs per year going from WARP1 to WARP2. This is apparently a mix of DH adjustment and league qual adjustment. Most would agree the AL was the better league in this period, so I would say 10 runs is not an exaggeration, ergo his wins-above-avg as a CF ought to be closer to what the WARP2 method says than the BWAA+replacement-adjus-of-.6 says.

Ron, Puckett's case is absolutely NOT all peak. It's prime, similar to Reggie Smith & Bob Johnson.
   108. TomH Posted: September 24, 2007 at 08:42 PM (#2540485)
BP translated stats
player.... AB OBP SLG DEF
Puckett 7312 372 524 CF-103
Butler.. 8335 400 418 CF-101
Murphy. 8040 345 517 CF/RF-100
Dawson10228 329 538 CF/RF-100
   109. TomH Posted: September 24, 2007 at 08:51 PM (#2540499)
1987: Twins allow more runs than they score. And win a World Series. Maybe SOMEONE should get a little credit for their team out-pythagging and winning a close division race? Maybe the guy who lead the team in runs scored and was 2nd in RBI should get some of that credit?

1991: Yes, we shouldn't put a guy in the HoM for one gane (hello, Jack Morris). But Puckett had a great ALCS (MVP), and a great, clutch World Series, with the bat and the glove. Doesn't that count for something?

Again, I don't mean to sound like a Puckett fan boy. Dan's analysis shows his warts; they are real, they are not small. I merely believe they are exaggerated, and we ought to appreciate his good points too.
   110. Howie Menckel Posted: September 25, 2007 at 12:44 AM (#2540799)
Can't a team just be lucky sometimes?
   111. 'zop sympathizes with the wrong ####### people Posted: September 25, 2007 at 12:59 AM (#2540823)
Again, I don't mean to sound like a Puckett fan boy. Dan's analysis shows his warts; they are real, they are not small. I merely believe they are exaggerated, and we ought to appreciate his good points too.

How are Puckett's flaws exagerrated? As far as I can tell, Dan has shown quantitatively that Puckett doesn't meet the HoM standard based upon peak, prime, or career tests. Your response to that has been...that BP's WARP1 to WARP 2 adjustment better captures Puckett's value. Dan correctly shows that the adjustment is inaccurate because it places the DH adjustment in the wrong place. In addition, there's a long line of discussion in the HoM about BP's bizzare, arbitrary, and capricious "league-quality" adjustment---if I'm not mistaken, they haven't published their technique for the adjustments, and I've personally seen some really wacky numbers. Given that Puckett is so far below any number-based standard, you've got to meet a much stronger burden than you've done here.
   112. TomH Posted: September 25, 2007 at 01:06 AM (#2540840)
1 eksyuuze me, but just because DanR and zop SAY the DH adjustment is in the wrong place doesn't mean it IS. Reasonable arguement either way.
2 Puckett is not 'so far below any number-based standard'. He is below the standard put forth in a few posts, whihc do not agree with other standards.
3 BP's league wual adj may not be "right", but let me ask this: Do you think the NL and AL were equal in 1988(ish)? If not, how much difference do you want to grant to the AL? BP has it about .4 wins per full batter-year, as best I can tell. I personally believe that is close to accurate. You can disagree, but if you do you may be in the minority.
   113. 'zop sympathizes with the wrong ####### people Posted: September 25, 2007 at 01:13 AM (#2540858)
And while I'm killing time and avoiding this stupid memo, I'm going to rant for a bit. I think "prime" voting is completely without merit. I understand career voting--vote for the player who accumulated the most value, period. I strongly disagree with it, but I understand the rationale and respect people who vote based on that basis. Its consistent and grounded in something absolute (the value a player provided for his team).

I also understand peak voting. A player should be judged by his established highest level of value. Peak voters don't vote on the best single season to avoid statistical flukes--essentially, they vote based on the highest level of play that was sustained for a sufficiently long period to represent "true-talent value". Again, a consistent approach, grounded in an absolute measure of value.

Then we have prime voting. I think it's meaningless. The fact that a player was good for, say, 7 years is completely arbitrary--why not reward the player who was good for 6 years, but stunk up the joint after that? Is his value any less real? What about the player who was only OK in 7 years, but had, effectively, two primes in a great 14 year career? A player doesn't win extra pennants for being good over a defined, non-career-long interval; its also widely acknowledged that a "prime" generally doesnt capture a players highest true level of play, since a player's peak virtually always is shorter than the length-standard imposed by a prime voter.

The only possible rationale I can see for "prime voting" is that it is a imperfect means for voters to vote for a mix of "peak" and "career" candidates, without having to face up to the abitrary choice that a rigorous voter has to make: "Am I rewarding total value, or the value at a player's best?" Prime is a cop out; a way to create a numerical system that spits out a rating, but it's a rating that has no bearing on the actual merit of the player in terms of how many pennants did he win, which is sort of the point of this exercise.

The following arguments are universally mocked by the analytical fan: "Mark Grace was great because he had the most hits in the 90s!" or "Since Jack Morris had the most wins in the 80's, he belongs in the Hall of Fame". At its core, how does prime voting not fall under the same logical fallacy?
   114. 'zop sympathizes with the wrong ####### people Posted: September 25, 2007 at 01:14 AM (#2540861)
1 eksyuuze me, but just because DanR and zop SAY the DH adjustment is in the wrong place doesn't mean it IS. Reasonable arguement either way.

Yes, but an appeal to authority (That's how BP does it! It must be right!) does not in and of itself constitute reasonable argument.
   115. Dr. Chaleeko Posted: September 25, 2007 at 01:23 AM (#2540873)
Well, I'm on the record a couple times that I don't give a hoot about postseason except MAYBE as a tie breaker. It's an opportunity-driven thing and either it's such a small sample as to be nearly meaningless when taken in the larger context of what a player's career looks like or it's a Yankee thing, which makes it increasingly less meaningful.

Post-season play is a real thing, but so is clutch hitting. It's the ultimate in clutch play. But like clutch hitting, I don't personally believe that post-season play tends to be indicitive of any special ability that's worth crediting a player for. They don't play 162-game post seasons, and offering up Puckett's post-season heroics (or Fingers') only magnifies the fact that he probably NEEDs such credit (and such credit with a multiplier) to advance up the rankings. The other 1800 games he played are so much more informative about the player's merit (IMO).

There are many, many reasons why why teams reach the post season, being good is a big one. Sometimes being pretty good but being lucky is a good one. But the success of ones teammates is the biggest reason any one player reaches the post season. It's axiomatic, but no one player puts a team into October. But once there even Scott Brosius or Dusty Rhodes or Billy Hatcher can be a great performer. That Puckett played well over four to eight weeks in October is just a thing, it's not his case, and, I believe, shouldn't have much of anything to do with it. Because if so, then Don Mattingly had better get heaps of credit for murdering the Mariners...in a series his team lost. I may be overly indulgent about MLE'ed guys, but this is one slippery slope that I just can't see the point in climbing.

(BTW, no offense, Tom. We've had this fight before, and we just disagree, but I just couldn't resist chiming in again....)
   116. Dr. Chaleeko Posted: September 25, 2007 at 01:34 AM (#2540881)
Prime is a cop out; a way to create a numerical system that spits out a rating, but it's a rating that has no bearing on the actual merit of the player in terms of how many pennants did he win, which is sort of the point of this exercise.

Nah. It's about individual cases. I trend toward prime because I have a bunch of different measures that are sometimes geared toward career, sometimes to peak, but that are sometimes geared toward points that are suggestive of neither. You may not like my criteria or my system, and that's fine, but my goal is to look at a player through a variety of perspectives. I use Keltner for that, others do different things.

But in another sense, I think you can credibly suggest that prime is simply vague language. I mean peak voters are looking at arbitrary lengths of time, and so are career voters, but what I see you saying is that they have a more clearly defined concept of what peak and career means. A prime voter who chooses ten years or 12 or 8 is just picking another interval than the peak/career guy, but the word appears to seem too much like a catchall to you, I guess. But if a prime voter's system says instead something like "Years the player was X% above league average" and measured out primes in terms of their varying lengths, would that be more "rigorous"?
   117. TomH Posted: September 25, 2007 at 01:41 AM (#2540892)
Eric, no offense, we merely disagree. I do not place even 5% of a player's HoM-iness on post-season. But it's enough to move Kirby up a few places, since I multiply his October antics by about 5, compared to regular season.

zop, I didn't mean to appeal to BP's authority as absolute; sorry if it came off that way, but honestly in re-reading, I don't see how I did. BP is one measure, just as Win Shares is. And DanR's WAR.
re: prime voting; would it be different if it was stated as 'value above average', or 'established value'? Those two are proxies for prime to some people.
   118. 'zop sympathizes with the wrong ####### people Posted: September 25, 2007 at 02:01 AM (#2540936)
But in another sense, I think you can credibly suggest that prime is simply vague language. I mean peak voters are looking at arbitrary lengths of time, and so are career voters, but what I see you saying is that they have a more clearly defined concept of what peak and career means. A prime voter who chooses ten years or 12 or 8 is just picking another interval than the peak/career guy, but the word appears to seem too much like a catchall to you, I guess. But if a prime voter's system says instead something like "Years the player was X% above league average" and measured out primes in terms of their varying lengths, would that be more "rigorous"?

I'll start by answering your question in the last line: I'd prefer it, but I don't know if I'd call it more rigorous. I don't know if I'd agree with it, since I don't think that the number of years a player is above average matters much to his total value. (Shouldn't 3 years of 8 wins above average be worth more than,, say 5 years of 1 win above average? Is there any substantive basis for rewarding years above average?) I'll grant that "years above average" avoids the arbitrary-endpoint fallacy critique of prime.

I understand the merits of "multiple perspectives", but shouldn't we restrict ourselves to perspectives that purport to measure the actual value of the player?

I don't think career voting uses an arbitrary length of time--in fact, I think that's a great strength of career voting---it gives the player as long to accrue value as he plays. I agree that career uses -a length of time-, but I don't think its arbitrary; its simply the length that allows us to best determine the player's value at his best. The length chosen flows from the value we want to measure.

My criticism of prime is that its exactly the other way around; the length of time is chosen, and THEN the value flows from that. What, exactly, is prime actually measure, other than best-in-seven years?
   119. sunnyday2 Posted: September 25, 2007 at 02:20 AM (#2540966)
>I understand the merits of "multiple perspectives", but shouldn't we restrict ourselves to perspectives that purport to measure the actual value of the player?

This is in re. #111 and #118: Maybe I'm wrong but I think the translation is: Everybody has to scrap your methods NOW and adopt DanR's because his are correct.

Did I miss something?








You know, this project has gotten along fine for over 100 years, most of them w/o said "correct" methods. Most of us always thought there was more than one way to skin a cat. (Apologies to cat fanciers.) Again, I guess I missed something. Should we ####-can this entire project and start over now that we have a "correct" method that we can/should all adopt?

And while I'm at it, I can't quite fathom what in the world TomH has to apologize for.... It didn't come off that way.

My apologies for using DanR's name in vain here because I am not responding to his words.
   120. Paul Wendt Posted: September 25, 2007 at 02:25 AM (#2540977)
Joba
I also understand peak voting. A player should be judged by his established highest level of value. Peak voters don't vote on the best single season to avoid statistical flukes--essentially, they vote based on the highest level of play that was sustained for a sufficiently long period to represent "true-talent value". Again, a consistent approach, grounded in an absolute measure of value.

That explains peak in a consecutive sense ("sufficiently long period").
What follows, quoted below, explains prime in a fixed-period sense ("length standard imposed").

There is a strong similarity here and the two supposedly different approaches may be equivalent, hence equally arbitrary, among contemporary regular position players or among contemporary starting pitchers. For example consider quantification of this peak in terms of the degree of confidence that distinguishes "established" and "fluke" levels of play. Then (increasing) degrees of confidence probably correspond closely to (increasing) numbers of seasons; imposing a degree of confidence in a peak may be equivalent to imposing a length standard for a prime.

Then we have prime voting. I think it's meaningless. The fact that a player was good for, say, 7 years is completely arbitrary--why not reward the player who was good for 6 years, but stunk up the joint after that? Is his value any less real? What about the player who was only OK in 7 years, but had, effectively, two primes in a great 14 year career? A player doesn't win extra pennants for being good over a defined, non-career-long interval; its also widely acknowledged that a "prime" generally doesnt capture a players highest true level of play, since a player's peak virtually always is shorter than the length-standard imposed by a prime voter.
   121. rawagman Posted: September 25, 2007 at 02:53 AM (#2541036)
My criticism of prime is that its exactly the other way around; the length of time is chosen, and THEN the value flows from that. What, exactly, is prime actually measure, other than best-in-seven years?

Pretty strong statement. Can you qualify it?
   122. rawagman Posted: September 25, 2007 at 02:54 AM (#2541038)
My criticism of prime is that its exactly the other way around; the length of time is chosen, and THEN the value flows from that.

Sorry - took too much last time.
   123. Howie Menckel Posted: September 25, 2007 at 12:32 PM (#2541270)
I generally lean to "prime" candidates, but it seems bizarre to put that in such a strawman box like above.
8 years of 145 OPS+ in the OF speaks to me more than 5 years of 150 and 3 years of 120. Is that odd? No.
12 years of 110 OPS+ at 2B speaks to me more than 5 years of 120 at 2B and 7 years of 90. Is that odd? No.

I'm going to assume that the tone given off by the posts is not what was actually meant. Otherwise, as noted, I'd wind up thinking that either a god has just joined the voting - or someone who thinks he's one.

I prefer not to believe that.
I think we all would.
   124. 'zop sympathizes with the wrong ####### people Posted: September 25, 2007 at 12:38 PM (#2541275)

I'm going to assume that the tone given off by the posts is not what was actually meant. Otherwise, as noted, I'd wind up thinking that either a god has just joined the voting - or someone who thinks he's one.


Oh, tone was absolutely intended. I think-and have thought for many "years" that the collegiality of the electorate was a weakness, not a strength, because it effectively legitimizes weak methodologies. For example:
<i>"8 years of 145 OPS+ in the OF speaks to me more than 5 years of 150 and 3 years of 120. Is that odd? No.
12 years of 110 OPS+ at 2B speaks to me more than 5 years of 120 at 2B and 7 years of 90. Is that odd? No."(/i>
Not only is that odd, I've yet to see an argument for this that goes beyond, "It just feels right to me!" Why does the longer career strength "speak to [you]"? Do you believe it helped the team win more ballgames? Does it reflect more value?

I'm posting what I think is a completely legitmate question to "prime" voters; why do you think your system is a better identifier of the most Meritorious players than peak or career?

(The key word above is "why", and, "because I said so" is not an answer.)
   125. TomH Posted: September 25, 2007 at 12:49 PM (#2541286)
Well yes, 8 years of 145 OPS+ DOES reflect more value than 5 years of 150 + 3 years of 120. It should be obvious.

Now, that is not in and of itself an argument for "prime" ; because career voters would agree with the conclusion as well. So would 'Pennants added'.

An argument for prime might be that 12 seasons of WARP=8 are better than 9 seasons of WARP=9 and 9 seasons of WARP=2, because altho the latter's total WARP is higher, as is the latter's peak, the seasons of WARP=2 are essentially replacement years that they can be safely deemed as virtually valueless.
   126. Howie Menckel Posted: September 25, 2007 at 01:00 PM (#2541296)
Guh, I'm off to work.

In a nutshell, for now:
I believe the advantage of prime (especially long prime) over career is merely that some players either are mostly irrelevant for extra years (a little below average at their positions) or that they are terrible.
I see it as neither rewarding mediocrity nor overly penalizing an over-the-hill season. The latter, to me, in many cases reflects that the team didn't have anything better. Some voters "reward" that because counting stats go up. Some penalize it because it lowers career rates.

I prefer neither option.



Disagree if you like, but acting obnoxious about it doesn't help.
   127. Dr. Chaleeko Posted: September 25, 2007 at 01:21 PM (#2541313)
I think Paul's reasoning is very interesting. Prime includes peak, and it lends more weight to the conclusions drawn from peak by virtue of the presence of other seasons of high-level value. The ability to sustain high-level play is an important differentiator among candidates.

But the problem with all this is that while career is the most highly and defensibly defined concept among the three (and is still squishy yet, particularly for MLE-based candidates), to some degree or another the ideas of peak, prime, and career are all up for interpretation on both a voter-by-voter basis and a candidate-by-candidate basis.

For instance, let's say there's a candidate with this shape in WS:

35 35 35 20 20 | 20 20 20 20 20 | 5 5 5 5 5 | 5 5 5 5 5

Is he the player with 3 35 WS years? Is he a guy with 145 WS in five years? Is he really a 20 WS guy who happened to have three outstanding years? Or is he a ten-year prime guy with nothing else to offer? Well, he's all of them, of course, the only question is what kaleidescope you look at him through.

The ideal system identifies not peak or prime or career, but finds some way of viewing the totality of the player. No one's found that perfect system, but most have one we think approximates our own concept of merit.

But another truth is that the wisdom of crowds makes no differentiation of ideal or perfect systems from flawed ones. Indeed it embraces all systems to arrive at consensus. The collegiality of the HOM has several main advantages:
1) It keeps the process enjoyable and keeps conversation from breaking down into flame wars
2) It attracts and retains electoral membership
3) It conveys the seriousness and dedication of the electorate to the mission
4) It embraces a variety of perspectives and encourages all comers so that we can attain the delphi/wisdom-of-crowds, rather than prescribe what voters should do and be forced to punish idiosyncracies on that prescription
5) It allows interesting and potentially important insight to arise from people whose perspectives are further from the mainstream.

That said, even this highly tolerant electorate has limits to its tolerance, and we do challenge ballots and vet and challenge new voters to ensure that their ballots meet certain minimum credibilities.

Looking at our results, most of us think there are two to five clinkers in the HOM. I think plural perspectives (peak, prime, career, or idiosyncratic) are less likely to blame for that than either the size of the HOM or the election system (as DanG has complained). Or simply the fact that in any group of elected greats, there are always borderline members that no one will quite agree on. And you might well disagree, 'Zop, on many HOM selections, but the results speak well to our process, including its collegiality.
   128. John (You Can Call Me Grandma) Murphy Posted: September 25, 2007 at 01:35 PM (#2541327)
Oh, tone was absolutely intended. I think-and have thought for many "years" that the collegiality of the electorate was a weakness, not a strength, because it effectively legitimizes weak methodologies.


From personal experience here and elsewhere, tone doesn't do squat. Only reasoned, logical arguments do the trick. Attitude, if anything, entrenches arguments.

As for prime voting, of course it's arbitrary. So is peak and career voting.
   129. DL from MN Posted: September 25, 2007 at 01:51 PM (#2541351)
I've never understood how a "prime" could be anything less than 10 years. 10 years is the requirement for career length to be HoF eligible.
   130. 'zop sympathizes with the wrong ####### people Posted: September 25, 2007 at 05:00 PM (#2541660)
I think Paul's reasoning is very interesting. Prime includes peak, and it lends more weight to the conclusions drawn from peak by virtue of the presence of other seasons of high-level value. The ability to sustain high-level play is an important differentiator among candidates.

Sure, it's a differentiator among candidates. But so is height, weight, or race. Just because some method of measurement provides a distinction amonng players doesn't mean that it measures which players are more Meritorious.

Paul's argument is -not- legitimate. He argues:
"Then (increasing) degrees of confidence probably correspond closely to (increasing) numbers of seasons; imposing a degree of confidence in a peak may be equivalent to imposing a length standard for a prime."

This is untrue. Just for the sake of argument, lets say that variations in hitter BABIP above and below "true talent" is a substantial cause of inter-season variability (I'm fairly certain it is). Lets imagine that we divide the player population into 3 group, based upon their BABIP compared to what it should have been; the bottom third are the ones with an actual BABIP farthest below true talent (unlucky), the top third highest above true talent (lucky) and the middle third players have BABIPs that accurately reflect their true talent.

Now, for a player to be overrated, over a 3-year peak, lets say that a player would have to be in the top 3rd 3 times, or in the top 3rd twice and the middle third once. The odds of that occurring, after 3 years, is only 3%. After 4 years, that drops to 1%. After 5, to 0.3%. And so on...

Meanwhile, the odds of a player maintaining true talent over an interval decreae as the years go on, because of aging + injury. What are the odds of a reduction in true talent in any given season: 1%? 5%?

As long as the risk of a dimunition in true-value is greater than the increased precision we gain with a larger sample size, extending the interval over which you evaluate a player reduces, not increases, your accuracy in assessin their highest level of true-value.

Does anyone really doubt that Al Rosen was as good as he was over his peak? Seriously? You really think that you need 7 years to tell the flukes from the true stars? A one season standard is absolutely too short to judge a player; Jorge Posada is not a .335 hitter. But 2 seasons is probably adequate, and if a guy reached a level for 3 seasons, consecutive or otherwise, you can bet the farm that he was "that" good.

Lastly, in response to Eric Chalek's argument in favor of collegiality; wisdom of crowds requires vigorous argument. Not to get all "Holmesian" here, but the HoM should act as a marketplace of ideas--all ideas are welcome, but inferior ideas should be identified as such, critiqued, and people should be compelled to defend the rigor of their method.
   131. 'zop sympathizes with the wrong ####### people Posted: September 25, 2007 at 05:01 PM (#2541664)
Now, for a player to be overrated, over a 3-year peak, lets say that a player would have to be in the top 3rd 3 times, or in the top 3rd twice and the middle third once. The odds of that occurring, after 3 years, is only 3%. After 4 years, that drops to 1%. After 5, to 0.3%. And so on...

This makes no sense, because i was originally going to do a much more compelx argument but changed it for simplicties sake. I mean, 3 years in the top third. It would be great if we had an "edit" function in the HoM threads...
   132. John (You Can Call Me Grandma) Murphy Posted: September 25, 2007 at 05:09 PM (#2541681)
Not to get all "Holmesian" here, but the HoM should act as a marketplace of ideas--all ideas are welcome, but inferior ideas should be identified as such, critiqued, and people should be compelled to defend the rigor of their method.


But that can still be done in a civil and friendly matter, just like we have been doing for over 100 elections now. We have had hundreds upon hundreds of debates concerning everything related to this project. The idea that we just accept anything posted here is simply not correct. Read the old threads and you will see that this is so.
   133. OCF Posted: September 25, 2007 at 09:20 PM (#2542019)
One possible way of viewing the peak/prime/career continuum would be to consider only value above some threshold, and to ignore all performances that lie below that threshold. This does ignore the issue of whether the peak/prime years are consecutive - personally, I don't put much stock in that.

If you set the threshold at replacement level (or some arguable estimate of replacement level) and count everything above that, then you're being a career voter.

If you set the threshold so high that you're only looking at all-star caliber seasons, then you've being a fairly extreme peak voter.

But you can set the threshold at any number of places in between. "Prime" would be the name for some intermediate range of settings.
   134. David Concepcion de la Desviacion Estandar (Dan R) Posted: September 25, 2007 at 11:57 PM (#2542316)
The ideal system identifies not peak or prime or career, but finds some way of viewing the totality of the player


I would humbly tout the salary estimator as a simple, quantitative way to view the totality of a player's work. The best thing about the method is that you can tweak it to taste--simply change the exponent to reflect your preferences (an exponent of 2 will give very peak-heavy results, while 1.25 would probably be similar to something like Joe Dimino's Pennants Added). According to Nate Silver the empirical market result for 2007 is an exponent of 1.5.

But obviously you have to set your replacement level before you get there. Clearly if you base your votes on BP WARP or WS you have to subtract some constant (generally thought to be around 2.5-3 WARP or like 10 WS per 162 games) before applying the estimator.
   135. David Concepcion de la Desviacion Estandar (Dan R) Posted: September 26, 2007 at 12:30 AM (#2542404)
Also, I wasn't aware that the AL was meaningfully superior to the NL in the late 80s/early 90s...my stdev research doesn't show that at least (although it wouldn't necessarily catch it by definition). Can anyone show me the evidence?
   136. TomH Posted: September 26, 2007 at 12:42 AM (#2542448)
mmmm, don't have a link to any study, let's see, where would I have seen it...
I'll have to find it later. Obvious small-sample evidence would be all-star games (NL has won, what, 6 of the last 24 games?) and World Series data - in Kirby's career, the AL won more than it's share.
   137. David Concepcion de la Desviacion Estandar (Dan R) Posted: September 26, 2007 at 12:44 AM (#2542458)
Yeah, I'm afraid that's not enough to really grab my attention. I'd need to see a league-switcher study or something...
   138. sunnyday2 Posted: September 26, 2007 at 03:15 PM (#2543407)
>Not to get all "Holmesian" here, but the HoM should act as a marketplace of ideas--all ideas are welcome, but inferior ideas should be identified as such

This is exactly who we are and what we do. Inferior ideas are identified as such by virtue of the majority/consensus voting otherwise. I don't see a better way than that, certainly not...well. never mind.
   139. Paul Wendt Posted: September 27, 2007 at 02:23 AM (#2544729)
Time is scarce. FWIW

(a) I wrote #120 before reading anything beyond Joba #113, which I quoted as point of entry. This is important

(b) Joba #130 "refutes" my point inadequately. If quantified, the opinions that one year is "absolutely too short" and "you can bet the farm" on a three-year average rate will provide instances of my point.

(c) A few days ago I quoted KJOK and called it a concept peak that should be in the casebook of every serious student of HOM thought :-)
I wonder what KJ thinks of that interpretation and of my point in #120. (When I learned the term "peak voter" here, and for a while thereafter, Kevin was my paradigm case. As Joe D and Dan G were paradigm "career voters". Today I still think a newcomer to the jargon and the scope of thought can do well by looking at Kevin and Dan as examples. JoeD's official role has required more diplomacy than DanG's; maybe he has become more eclectic introspectively too.)

(d) I agree with Joba that my statistical interpretation of peak, in terms of distinguishing established from fluke, cannot plausibly support anyone's focus on seven seasons. Five is a stretch. With time I could do some simulations that would be illuminating for me, if no one else. Lacking that I am willing to confide my guess that the practical scope of that kind of statistically justified peak voting system is two to four seasons for regular batter-fielders. Only within that range does a peak voter of the statistical motivation (again, see KJOK's expression) plausibly have scope for a "reasonable" "arbitrary" choice. Eg if willing to play the peak voter role, Joba might choose two full seasons for regular players, Paul three, Kevin four --each derived from essentially arbitrary level of confidence that he requires (and the latter is the main point of #120).
   140. KJOK Posted: September 29, 2007 at 05:13 AM (#2549593)
c) A few days ago I quoted KJOK and called it a concept peak that should be in the casebook of every serious student of HOM thought :-)
I wonder what KJ thinks of that interpretation and of my point in #120. (When I learned the term "peak voter" here, and for a while thereafter, Kevin was my paradigm case. As Joe D and Dan G were paradigm "career voters". Today I still think a newcomer to the jargon and the scope of thought can do well by looking at Kevin and Dan as examples. JoeD's official role has required more diplomacy than DanG's; maybe he has become more eclectic introspectively too.)


I've always rejected in my mind the label as 'peak' voter, as I look at the player's entire career. My ballot ends up looking like a 'peak voter' ballot because I don't give much/any credit for 'non-HOM calibre' seasons (at least above average) so players that pile up lots of 'average to below average' seasons in their 11th - 20th best years don't get any more credit than a similar player who retired after 10 years (note: players who play less than 10 years do get a 'deduction' for those 'missing' years)
   141. sunnyday2 Posted: October 11, 2007 at 05:56 PM (#2571924)
Continuing to Puck away--er, I mean, pick away--at some of the candidate players:

Puckett played against what I would characterize as a reasonably strong but not overwhelming set of CFers. You had some pretty good ballplayers.

Yount overlapped 9 years with a high of 34 and a median of 20 WS.

Moseby 7 years at high of 26 median of 17.
Devon White 7 years at 24 and 14.
Willie Wilson 7 years at high of 23 and median of 17.
Ken Griffey 6 years 30 and 24.5.
Ellis Burks 6 years at high 24 and median 15.5.
Pettis 6 years at 18 and 11.
Lance Johnson 5 years at 21 and 13.
Butler 4 years at high of 23 and median of 20.
Chet Lemon 4 years at high of 24 and median of 18.5.
Dwayne Murphy 4 years 22 and 16.5.
Fred Lynn 4 years of serious decline.
R. Henderson only 3 years but at 38-28-26.
Brady Anderson 3 years at 29 and 18.
Lofton just 3 years but 24-25-21.

Across 11 years Kirby beat the median by 93 WS or let's round it off to 8.5 WS per year. Cut out his rookie year (-3) and you get 96 for 10 or 9.5 (rounding to nearest half).

For comparison, in Yount's CF career you get +75 WS over 9 years or 8.3 WS per year. Ken Griffey in his 5 overlapping seasons was +48 or +9.6 per year. Pretty much all of Lloyd Moseby's prime is in there, though, and he was +16 for 7 years (+2.3). Ditto Devon White who was +9 for 7 years (+1.3). Obviously this is applying Puckett's endpoints, but just for a frame of reference.

This compares to Dan's numbers of 8 WS each for Concepcion and Doyle, 4 for Nettles and 3 for Rizutto, among those players that I've looked at. My numbers pretty much agree with Dan's except for Concepcion's which I have at half as many as Dan does. No idea where the discrepancy comes from at this point.

All the usual caveats of WS of course apply. But these are all CF so there's no advantage to Kirby on that account.

I hope to to Dale Murphy next and in doing so will fill out the rest of Brett Butler's career as well.
   142. jimd Posted: October 18, 2007 at 02:23 AM (#2581943)
I apply the DH adjustment to the replacement level (where, in my opinion, it rightly belongs),

As well you should. That's where it belongs in your system.

WARP1 chooses to assert that the DH has no practical effect on replacement level (as Davenport has implemented it in WARP1). Win Shares asserts that it has no replacement level to be affected.

But this isn't the DH adjustment that most people are referring to. The DH changes the offensive context of the game, adding another professional hitter, resulting in more runs per game. It changes the conversion factor from runs to wins because it now takes more runs to get a win. In Win Shares, the BWS are now split more ways. In WARP1 and DR-WAR, the same number of batting runs produce less wins-above-replacement. It makes it more difficult to achieve the same offensive value when comparing AL players post-DH to pre-DH, or AL players post-DH with NL players.

Its net effect is to disadvantage offensive players who play in the AL post-DH. Should it be adjusted for? I think that's a matter of opinion. If the NL had adopted it at the same time, we probably would have ignored it as just another rules change. But the NL didn't, so there are two parallel sets of rules in play simultaneously, one of which values offensive players more highly than the other. The same player that evaluates to be a marginal HOMer in the NL would be a marginal miss in the AL, and vice-versa.

Which doesn't quite seem fair, and is the main argument in favor of adjusting.
(And a similar argument applies to adjusting for relative league quality, both for and against.)
   143. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 18, 2007 at 02:36 AM (#2581956)
Jimd, great post. THAT's why the DH has such a strong negative correlation to standard deviation, even if you only take the stdev of left-spectrum positions! I could never figure out exactly what was causing it before.

I'll have to think this through thoroughly before figuring out how to address this issue in a future version of my WARP. For now, I'm focusing on trying to get pitcher numbers, though.
   144. Paul Wendt Posted: October 19, 2007 at 01:33 AM (#2583277)
jimd's point evidently pertains to wins and it implies a relation of the DH and the distribution of wins, such as a negative correlation between the DH indicator (1=DH, 0=none) and the standard deviation of batting wins.
Which other distributions and standard deviations? I'm not sure. The DH is correlated with matters such as intentional bases on balls (NL batting position 8), batting with runners on base (AL leadoff vs NL leadoff), and so on.
   145. Dr. Vaux Posted: October 19, 2007 at 04:25 AM (#2583970)
This ad is awesome:

Kirby Parts, All models
Cords, hoses, wands, rollers etc Fast secure shipping.
kirbypartstore.com
Page 2 of 2 pages  < 1 2

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
JE (Jason Epstein)
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

Syndicate

Demarini, Easton and TPX Baseball Bats

 

 

 

 

Page rendered in 0.7888 seconds
49 querie(s) executed