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

Thursday, January 11, 2018

Todd Helton

Eligible in 2019

DL from MN Posted: January 11, 2018 at 02:39 PM | 31 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.

   1. DL from MN Posted: January 11, 2018 at 02:49 PM (#5604834)
Player WAA WAR
Helton 32.8 61.2
Beckley 26.4 61.5
Hernandez 31.6 60.0
W Clark 28.8 56.2
B Terry 31.8 54.2
Sisler 22.4 54.5
Palmeiro 30.1 71.6

N Cash 25.9 52.0
Ben Taylor 34.6 69.9 (Chalek MLE)
   2. Carl Goetz Posted: January 11, 2018 at 03:08 PM (#5604853)
I don't think he's a no-brainer type, but he's in for me. I feel he got less MVP votes over his career (particularly the 2000-04 peak period) than he should of. He's my NL MVP in 2000.
   3. Kiko Sakata Posted: January 12, 2018 at 01:09 PM (#5605490)
Reposted from the 2018 Discussion Thread - because this has a pretty big impact on my view of Todd Helton:

I found a mistake in my calculation of (context-neutral) eWins - basically, I wasn't controlling for ballpark effects. I have corrected this on my website. Apologies for the error. Explanation here; examples of the impact here. My apologies for this.
   4. The Honorable Ardo Posted: January 12, 2018 at 08:04 PM (#5605770)
Helton will become the only member of the Hall of Merit to suffer an injury and lose his job to Peyton Manning.
   5. Bleed the Freak Posted: January 17, 2018 at 06:34 PM (#5608417)
Thanks for the update Kiko, losing 7 WORL is ~10 WAR, and makes him a candidate a little shy of HOM level.
An uberstat construction using 1/3 p, 2/3 e wins, zeroing out negative seasons:
http://baseball.tomthress.com/Leaders/UberLeaders.php?y1=1925&y2=2017&p=.33333&e=.666666&w=0.1328&a=2.0059&aa=6.5704&r=1&na=0&naa=0&nr=0&c=1.1926&b1=0.9393&b2=0.9874&b3=0.9393&ss=0.9393&lf=0.9393&cf=0.9393&rf=0.9393&dh=0.9393&ph=0.9393&pr=0.9393&o1=0.9296&sp=0.9296&rp=1.3321&psw=1&psa=1&psr=1>=162&ga=1&n=500

What are your thoughts?
Baseball Reference shows him as a mid-level hall guy, while Baseball Gauge is maybe upper portion of the bottom 1/3 tier.

   6. Kiko Sakata Posted: January 17, 2018 at 07:33 PM (#5608440)
Helton looks like he's probably going to be off my ballot, probably by quite a bit (he's maybe top 75-100; maybe top 50 at an upside). Obviously, the park adjustment hurt him, but I'm pretty sure that I trust the new numbers.

Helton's big problem from my perspective essentially is that he's not really unique enough. There are a ton of good hitting first baseman types sitting out there, both through history - Gil Hodges, Dolph Camilli, Jack Clark (not to mention a bunch of guys who are already in the HOM and/or HOF, of course) - and, especially over the past 20-30 years.

Being fairly generous in the definition of "contemporary", these guys were all essentially contemporaries of Helton that fit the same broad profile:

(**) - already in Hall of Merit
(*) - not eligible yet

Jeff Bagwell (**)
Frank Thomas (**)
Mark McGwire (**)
Jim Thome (**)
Edgar Martinez (**) (if one groups 1B and DH together - ditto for the Big Hurt, of course)
Albert Pujols (*)
Jason Giambi (*)
Lance Berkman
Fred McGriff
Carlos Delgado
John Olerud

Maybe more, depending on how wide a timeframe you're looking to cover (I considered David Ortiz, Miguel Cabrera, and Joey Votto and left them off thinking I might be taking in too wide a time frame).

I think Helton's a better candidate than at least a couple of these guys (Delgado, Olerud), but not by a huge amount. Berkman has the postseason numbers and significant time as an outfielder, which lowers the positional average against which he's being compared. And that is still probably going to leave him (just) off ballot for me this year. I suspect I'll like Giambi when he becomes eligible, and, of course, I would guess that everybody will have Pujols on the top of their ballots five years after he finally retires.

I'm not inclined to argue too strongly against Helton if others put him on their ballots. The uniqueness of Coors Field certainly increases the error bars on how we should best value his career. But, for me, I'm pretty comfortable with him falling into that 50-100 range of "good, solid player, worthy of remembering; but ultimately not a HOF/HOM-er".
   7. Jaack Posted: January 17, 2018 at 09:14 PM (#5608510)
Two things to work through for me with Helton

1. His defensive numbers are across the board pretty good to outstanding, particularly DRA. That being said, I'm juuust a bit sceptical that a First Baseman can be 20 runs above average defensively at Coors Field in the early 2000s.

2. Helton has some astonishing RE24 numbers. For example, his 2000 season is the best offensive, non-Bonds season since 1974. RE24 is naturally adjusted for run environment, but I am weary about it at the extremes - there does seem a bit of a bias towards Coors and towards the 2000 and 2001 seasons.

I was initially planning of Helton appearing in the mid section of my ballot, but there are enough questions here that I'm not feeling quite as confident in that anymore.
   8. DL from MN Posted: January 11, 2019 at 01:14 PM (#5804687)
Player WAA WAR
Helton 32.8 61.2
Beckley 26.4 61.5
Hernandez 31.6 60.0
W Clark 28.8 56.2
B Terry 31.8 54.2
Sisler 22.4 54.5
Palmeiro 30.1 71.6

N Cash 25.9 52.0
Ben Taylor 34.6 69.9 (Chalek MLE)


Surprisingly Todd Helton has jumped to the front of the frontlog in the latest election. The Ben Taylor MLE has since been adjusted upward slightly

Ben Taylor 35.7 71.6

I agree with Kiko that Helton is quite similar to Jack Clark once the standard deviation adjustments are accounted for. Other similar 1B in my analysis include John Olerud and Jason Giambi.
   9. DL from MN Posted: January 11, 2019 at 01:42 PM (#5804712)
I forgot to note - Dr. C said that Ben Taylor's MLE is not adjusted to a 162 game season but rather is translated to the games played in MLB those seasons. In most cases that is 154 games but in 1918/19 the war shortened the season to 130/140 games. That adds roughly 600 more plate appearances.
   10. Rob_Wood Posted: January 11, 2019 at 03:51 PM (#5804780)
Helton's big problem from my perspective essentially is that he's not really unique enough. There are a ton of good hitting first baseman types sitting out there, both through history - Gil Hodges, Dolph Camilli, Jack Clark (not to mention a bunch of guys who are already in the HOM and/or HOF, of course) - and, especially over the past 20-30 years.


This type of thinking truly baffles me.

Suppose you have 3 players with the following values, with players 1 and 2 being very similar players, and player 3 being unique:

Player 1 = 60 wins
Player 2 = 58 wins
Player 3 = 54 wins

It is inconceivable to me that a voter would vote for Player 3 over Player 1 and Player 2 due to his "uniqueness". A more reasonable approach would acknowledge that the "player type" exemplified by Player 1 and Player 2 is a more valuable prototype than the type of player Player 3 represents.

The Hall of Merit was initially set up as a counter to the plethora of poor Hall of Fame selections, many of which were selected (partially) because they were "unique".



   11. Kiko Sakata Posted: January 11, 2019 at 04:58 PM (#5804801)
This type of thinking truly baffles me.

Suppose you have 3 players with the following values, with players 1 and 2 being very similar players, and player 3 being unique:

Player 1 = 60 wins
Player 2 = 58 wins
Player 3 = 54 wins

It is inconceivable to me that a voter would vote for Player 3 over Player 1 and Player 2 due to his "uniqueness".


Ultimately, this circles back to the positional average discussion we've already had at least twice. That is, having a bunch of guys available who can do the same thing(s) that Todd Helton did - play first base while hitting something like .280/.360/.450 in a neutral offensive context - leads to him being worth fewer wins. So, it's not that I'm picking a 54-win guy (let's call him, I don't know, Mariano?) over a 60-win Todd Helton. It's that I'm saying that I don't think Todd Helton was really worth 60 wins, more like 52 or 53.
   12. Rob_Wood Posted: January 11, 2019 at 08:10 PM (#5804852)
Fair enough.

Thanks for taking the time to reply.
   13. John DiFool2 Posted: January 12, 2019 at 08:22 AM (#5804880)
Your analysis of course ignores Helton's excellent D. And his road stats were better than what you listed: .287/.386/.469 (now, if you also then take the air out of Sillyball on top of that, fine).

If you are arguing that the positional adjustment should vary depending on how a position performed in the league as a whole in a given season, that might be a bit more to the point. No idea why the various WARs don't do that (tho some might).

Checking 1B during Helton's peak, it of course was typically at or near the top (said averages of course include Helton's contributions), but several times was 3rd behind LF & RF.
   14. Kiko Sakata Posted: January 12, 2019 at 12:51 PM (#5804908)
Your analysis of course ignores Helton's excellent D. And his road stats were better than what you listed: .287/.386/.469 (now, if you also then take the air out of Sillyball on top of that, fine).

If you are arguing that the positional adjustment should vary depending on how a position performed in the league as a whole in a given season, that might be a bit more to the point.


Sorry, my response was quick-and-dirty and not meant to be mathematically precise. Rob and I have discussed positional adjustments a couple of times already in the discussion thread and your last sentence here is pretty much entirely my argument "against" Todd Helton. The argument that "there were lots of other first basemen who were similar, if somewhat lesser, in value to Todd Helton during his career" is sort a pre-emptory answer to the question "why should the positional adjustment for Helton be so large". Helton is basically dead even with John Olerud in my system, which doesn't make him a bad HOM candidate - I don't think either of those guys would be a "mistake" were they elected - he's just not, in my opinion, the best returning candidate for the 2020 ballot.
   15. John DiFool2 Posted: January 12, 2019 at 02:21 PM (#5804919)
Note I've personally considered the various positional WAR adjustments to be a bit biased against 1B (and DH's, but that's for when Papi comes up). If 1B was typically even with left and right, wouldn't that necessitate an adjustment in favor of 1B (or at least additional penalties for corner OF)? [note some newer positional estimates do indeed do just that.]

I noticed you had Hodges at 25. What puts him that much over Helton? [he's also a glove man] His peak appears dwarfed by Helton's-playing in Ebbets in the 50's, he never cleared a .400 OBP. I'm not sure why such a heavy weight needs to be placed on what top contemporaries are doing (vs. the league as a whole), especially in a bigger league-Richie Ashburn was the 4th or even 5th best CF in the 50's, but likely would be the best in 70's. If it isn't a player's fault I don't quite get why that needs to be prioritized. Plus Helton appears to have slight edges in several areas over those guys. [McGwire: durability & D. Giambi: D. Clark: more sustained peak. Delgado, D and peak.]

His D was discussed over there too. Just a thought: an out in Coors is typically more valuable than one in a lower-run environment. Thus an extra out gained thru superior D would be more valuable than it would be in say 1968. And OBP would thus also be more proportionally valuable than SLG.
   16. eric Posted: January 12, 2019 at 02:39 PM (#5804923)
Ultimately, this circles back to the positional average discussion we've already had at least twice. That is, having a bunch of guys available who can do the same thing(s) that Todd Helton did - play first base while hitting something like .280/.360/.450 in a neutral offensive context - leads to him being worth fewer wins. So, it's not that I'm picking a 54-win guy (let's call him, I don't know, Mariano?) over a 60-win Todd Helton. It's that I'm saying that I don't think Todd Helton was really worth 60 wins, more like 52 or 53.


I don't often post in these threads so I apologize if I'm revisiting concepts already discussed, but...

What does having a few other players just as good at the same position have to do with how many wins a player provided? Greatness is not evenly distributed through time. We've had periods of lots of HOF catchers (30's), piles of HOFers (or HOF-quality) at 3B (70's/80's), and of course a run at 1B (90's/00's). At the same time, there's been a dearth at positions at certain times (3B before the 50's, C recently, as examples; and all this is ignoring pitchers who have had their clear peaks and valleys).

To make the argument that Helton has provided less value is to claim that there were more players at his quality overall in MLB, ie, the entire baseline has been raised. And not just at 1B, but anywhere.

Let's look at three simple example distributions:

Year X: 16 six-WAR players in the league. Two at each fielding position--2 at C, 2 at 1B, 2 at 2B, 2 at 3B, 2 at SS, 2 at LF, 2 at CF, 2 at RF
Year Y: 16 six-WAR players in the league. Mixed distribution--1 at C, 5 at 1B, 2 at 2B, 1 at 3B, 1 at SS, 1 at LF, 2 at CF, 3 at RF

There's no overabundance of talent in the league. Just more great players for whatever reason are at 1B. It doesn't mean each of those players provides less value. The runs they create are still on the board, and as above-replacement as anyone else.

Which brings us to the issue that the entire argument that "more great players at a position means each player is worth less" also ignores the remaining players at that position. You can't compare one 1B to four or five others, you have to compare one to the entire pool of dozens of 1B options.

Ultimately, in order for that argument to be persuasive to this observer, it would have to use actual numbers to the rigorous degree of a WAR-like stat. Otherwise it reads a bit like sophistry.
   17. Kiko Sakata Posted: January 12, 2019 at 05:38 PM (#5804944)
[Holy hell this website sucks! I wrote this response three hours ago but couldn't get to this page while being logged in to post it until now.]

What does having a few other players just as good at the same position have to do with how many wins a player provided?


Wins are zero-sum for a league as a whole and all teams are required to field a player at each of the nine positions. Major League Baseball is better than AAA baseball because there are more good players in the former than the latter. Measured in terms of wins, a player who moved from MLB to AAA would be expected to be more valuable in the latter league because of this. Same logic, just extended to major-league first basemen.

Let's look at [two] simple example distributions:

Year X: 16 six-WAR players in the league. Two at each fielding position--2 at C, 2 at 1B, 2 at 2B, 2 at 3B, 2 at SS, 2 at LF, 2 at CF, 2 at RF
Year Y: 16 six-WAR players in the league. Mixed distribution--1 at C, 5 at 1B, 2 at 2B, 1 at 3B, 1 at SS, 1 at LF, 2 at CF, 3 at RF

There's no overabundance of talent in the league. Just more great players for whatever reason are at 1B. It doesn't mean each of those players provides less value.


The example begs the question. It's absolutely possible for the latter distribution to exist and it's absolutely possible for the latter distribution to exist in my system. But to assume the WAR values of the players in the two examples is to assume the positional adjustments in the two examples, which, in the case of Todd Helton, is precisely what we're discussing.

Ultimately, in order for that argument to be persuasive to this observer, it would have to use actual numbers to the rigorous degree of a WAR-like stat. Otherwise it reads a bit like sophistry.


My analysis is based on my own statistic, Baseball Player won-lost records. To calculate the positional average for first basemen in 2000, I add up the total wins earned by first basemen, add up the total losses earned by first basemen, calculate a winning percentage from those totals, and call that the positional average. It's as close to literally the average winning percentage of first basemen in 2000 as possible. That's not to say it's the only way, or even best way, to calculate such a thing, but it's as rigorous a calculation as one can make.

And I'm not really alone on this island with respect to Todd Helton. Baseball-Reference gives him 61.2 WAR but Fangraphs only gives him 55.0 WAR. Part of the difference is that Fangraphs thinks he wasn't quite as good (but still above average) as a fielder (+58 vs. +72) but most of it is in the positional adjustment - -155.5 (runs) at Fangraphs vs. -115 at BB-Ref. That Fangraphs number puts Helton behind, for example, John Olerud (57.3 career WAR), Lance Berkman (56.0), and Fred McGriff (56.9).
   18. GregD Posted: January 12, 2019 at 05:45 PM (#5804946)
My analysis is based on my own statistic, Baseball Player won-lost records. To calculate the positional average for first basemen in 2000, I add up the total wins earned by first basemen, add up the total losses earned by first basemen, calculate a winning percentage from those totals, and call that the positional average. It's as close to literally the average winning percentage of first basemen in 2000 as possible. That's not to say it's the only way, or even best way, to calculate such a thing, but it's as rigorous a calculation as one can make.
I think your system is interesting but have one question: Does that leave the numbers highly vulnerable to outlier seasons--Barry Bonds messes up everyone's positional adjustment, for example. Wouldn't some kind of median be more reliable?

I do think your analysis does some interesting things to the concept of replacement level. There are some times when the replacement level for shortstops is different than at other times
   19. Bleed the Freak Posted: January 12, 2019 at 05:47 PM (#5804947)
Some comments from Helton I had compiled from post 187 of the 2019 ballot discussion thread:

4. bachslunch Posted: January 24, 2018 at 11:55 AM (#5612400)
7. Todd Helton. Excellent WAR and easily the best qualified 1B.



12. The Honorable Ardo Posted: January 24, 2018 at 08:31 PM (#5612946)
5) Todd Helton - Little things add up. Helton is akin to Will Clark with a shallower peak/longer prime: a very good on-base offensive threat, defender and base runner.




16. DL from MN Posted: January 25, 2018 at 12:59 PM (#5613248)
Todd Helton = Norm Cash but with less bat than Cash after the standard deviation adjustments. All the modern players had high standard deviations in run scoring.
66) Todd Helton



19. Jaack Posted: January 25, 2018 at 02:23 PM (#5613326)

Todd Helton - This year's biggest problem for me. I think he's worthy. It's possible I could change my mind in the future, but that's where I lean now. The problem is how high up the ballot should he go. Before any adjustments, I have him about tied with Jeff Kent for 7-8 range. But two things give me pause. First are his excellent RE24 numbers that are pumping up his batting record. I'm not sure RE24 adjusts correctly for extreme run enivornments, which would explain Helton's very high scores (as well as Giambi's as well, who also sees a huge boost based on the metric). The second thing is his defensive numbers. I believe Helton was a fine defender. - both the numbers and the eye test agree he was quite good. My main concern is how valuable an elite defensive first baseman in 2000-2002 Coors Field was. Until I figure those two out, he'll remain just off ballot.

16. Todd Helton



93. cookiedabookie Posted: February 05, 2018 at 04:05 PM (#5619914)
25 Todd Helton



102. Michael J. Binkley's anxiety closet Posted: February 06, 2018 at 02:00 PM (#5620446)
I use a JAWS-like system as one of the inputs for my player calculations. The one difference is that unlike JAWS, I don't divide the value by 2. So in my system, using Dr.Chaleeko's MLE's, I have Taylor at a JAWS value of 114.12 (I do include a slight defensive regression to the mean for the NeL players). For comparison (and these are the value using my WAR calculations, not bWAR):

Palmeiro - 119.96
Murray - 119.86
Sisler - 113.79
Easter - 113.74
Hernandez - 113.46
McCovey - 111.81
McGwire - 111.52
Killebrew - 109.30
Terry - 101.92
Beckley - 101.36
Giambi - 100.75 (will make my PHoM soon after eligible)
Helton - 100.11




186. Dr. Chaleeko Posted: November 04, 2018 at 01:30 PM (#5781891)
After all the Negro Leagues data got sifted through, I'm thinking about my 2019 ballot. I think it's looking this way on a by-position basis. We're electing to roughly the 23rd best at each batting position. My own methods of assessment strongly suggest catchers are our one area of keen need. Avoiding 1Bs might be a good idea and to a lesser degree at LF and CF. I think we could use a few more pitchers too. We're at about 26%, which might be a tad lower than I'd like. Other figures suggest that we could use help in the 1910s, 1970s and 1980s. And that we ought to avoid the entire 19th Century the 1920s, the 1930s, and the 1950s.

1B
T Helton: 115 CHEWS+, 13TH
B Taylor: 113 CHEWS+, 17TH

4) Todd Helton (despite having a lot of 1Bs, being 13th at a position is a very strong argument for a vote)



I appreciate Jaack's comment about RE24 potentially having issues with extreme environments, where do you base this on, anyone else have thoughts on this?
456.9 base runs to 635.1 RE24, 42.4 wins to 54.7, or +10.3 wins, offset partially by a clutch score of -2.4.

Some impressive work has been done at the Baseball-Fever to quantify the impact a home ballpark factors a player WAR, Helton is at -5.4, downloadable spreadsheet found in this post:
https://www.baseball-fever.com/forum/general-baseball/statistics-analysis-sabermetrics/83320-progressing-toward-better-stats-thread#post83320

Those who haven't weighed in, are we comfortable with Helton as a HOMer either this year or soon?
   20. Kiko Sakata Posted: January 12, 2019 at 06:09 PM (#5804949)
Does that leave the numbers highly vulnerable to outlier seasons--Barry Bonds messes up everyone's positional adjustment, for example. Wouldn't some kind of median be more reliable?


Yes. Doing this empirically every season leads to occasional spikes - although, for full careers as long as we tend to discuss here, I'm not sure that single-year spikes are a big deal. But you can definitely see positional "gluts" in the numbers over time - CF and C in the 1950s, 1B in the 1930s are examples that I know show up really obviously - 1B in the 1950s is the opposite (to answer the earlier question, this benefits Gil Hodges who I rate as being the "best first baseman in the 1950s" (*) with the caveat that he beats Stan Musial only because Musial played so much outfield). And I do think it's absolutely a fair question to ask whether one should smooth out the numbers to eliminate these things - i.e., acknowledge that the "average" first baseman in the 1950s was below average.

To answer your specific example, here are positional averages for LF from 2000 - 2005. These numbers are offense-only (fielding averages are 0.500 for all positions for all seasons, by construction) relative to non-pitchers: 2000 - 0.527, 2001 - 0.531, 2002 - 0.530, 2003 - 0.529, 2004 - 0.530, 2005 - 0.521. The average across all seasons for which I've calculated Player won-lost records (1921 - 2018) is 0.526. So, yeah, Bonds is single-handedly bumping up the positional average for LF by perhaps 0.005 - 0.010 during his crazy seasons. I agree that this is not ideal.

I did a little experimenting with medians instead of means with first basemen in the 1950s and it basically made no difference. I can't swear that would be as true for other anomalous periods (I would guess it would help in cases of a single superstar player - Ruth in the 1920s, Bonds in the 2000s).

It's a shame that Dan Rosenheck disappeared from here. He had a system that did a lot of the same things that I do (not coincidentally, I followed his example on some things that I particularly liked about his system). Apropos to this discussion, he didn't use positional "averages"; he used positional replacement levels, which he calculated empirically and allowed to vary by season (I can't recall if he did anything to smooth the year-to-year results). I do think it's a fair point to note that the existence of Barry Bonds doesn't really affect the "value" of a league's left fielders - there's only one Barry Bonds and he's not available as an alternative for other teams. It's a lot of work, but I may play around with what would happen if I lowered my baseline positional comparatives to "replacement level" instead of "average" - one obvious issue with such a thing is that average is an objective standard (even if there's not necessarily a single correct way to calculate it, at least everybody understands what we mean by the term), whereas "replacement level" is not (I believe Rosenheck used the average of the three worst starters at a position each season).

(*) - also, just to clarify, I don't award bonus points or anything for being "the best X" and players' cardinal rankings by position don't enter my rankings. Hodges' ranking is based on him being 13.0 eWins and 18.9 pWins better than positional average and Helton's ranking is based on him being 13.3 eWins and 6.8 pWins over positional average over his career. (As you can see, Hodges also benefits relative to Helton in that he looks much better in pWins - which tie to team wins - while Helton looks much worse; in effect, Hodges is getting credit for playing for better teams than Helton did. I completely understand that many people will disagree with holding this against Todd Helton and I would encourage those people to focus exclusively on my eWins - by which Helton would show up higher in my rankings).

EDIT: Actually, I think the numbers I quoted for 2000-05 LF are for a non-DH league (which is what Bonds played in, of course; I calculate my averages using data from both leagues, but with the DH league numbers adjusted to reflect the fact that pitchers don't hit. Anyway, that doesn't affect the bottom line: Bonds is definitely goosing up the average a little bit).
   21. John DiFool2 Posted: January 12, 2019 at 06:50 PM (#5804953)
Some impressive work has been done at the Baseball-Fever to quantify the impact a home ballpark factors a player WAR, Helton is at -5.4.


1B in the 1930s are examples that I know show up really obviously - 1B in the 1950s is the opposite (to answer the earlier question, this benefits Gil Hodges who I rate as being the "best first baseman in the 1950s.


I'm still confused-do you think Hodges would help win fewer games in the Sillyball era than he did (or more?), and/or that his OBP/overall stats would proportionately improve? Can these both be true? On what basis can you perform these time machine exercises, what methodology would you use? [I'll also note he was playing in a league which was only starting to integrate] That's a strangely paradoxical outcome of the assumptions that you are using. Reductio ad absurdum.

A hitter isn't hitting against other hitters. He's hitting against pitchers. What Jim Thome or Jason Giambi are doing in some other game that day has zero bearing on what value Todd Helton is bringing to his. I recall Rosenhack's system, and didn't think much of it, for these reasons. [and see 16]
   22. Kiko Sakata Posted: January 12, 2019 at 07:23 PM (#5804958)
I'm still confused-do you think Hodges would help win fewer games in the Sillyball era than he did (or more?), and/or that his OBP/overall stats would proportionately improve?


I don't know and don't really care.

What Jim Thome or Jason Giambi are doing in some other game that day has zero bearing on what value Todd Helton is bringing to his.


But what Jim Thome or Jason Giambi are doing for the other team in the same game absolutely affects the value Todd Helton is bringing to his team assuming you're measuring that value in terms of wins - how many wins a run of offense is worth very much depends on how many runs the other team is likely to score.

Todd Helton batted .316/.414/.539 in 9,453 career plate appearances. Willie Stargell batted .282/.360/.529 in 9,027 career plate appearances. Surely your analysis of their relative offensive performances doesn't stop there and lead you to argue that Helton was a better hitter. You have to adjust for the run-scoring context. And the run-scoring context for a season or a career is absolutely calculated taking into account games in which neither Helton nor Stargell participated.

Barry Larkin had a career OPS+ of 116 in 9,057 plate appearances (+200 Rbat per BB-Ref). Mark Grace had a career OPS+ of 119 in 9,290 plate appearances (+223 Rbat per BB-Ref). Both were above-average fielders at their positions (BB-Ref thinks Grace was moreso, +76 vs. +18). Do you really believe that Mark Grace is more HOM-worthy than Larkin? No? Then you're acknowledging the need for positional adjustments. We're just debating the specific values of those adjustments.
   23. Michael J. Binkley's anxiety closet Posted: January 12, 2019 at 07:25 PM (#5804959)
Addressing a number of issues from the past few posts:

If you are arguing that the positional adjustment should vary depending on how a position performed in the league as a whole in a given season, that might be a bit more to the point. No idea why the various WARs don't do that (tho some might).


Does that leave the numbers highly vulnerable to outlier seasons--Barry Bonds messes up everyone's positional adjustment, for example. Wouldn't some kind of median be more reliable?


Kiko has already stated that he adjusts by positional average to some extent. And like he said, DanR also adjusted his replacement level by using a 9-year rolling average of how the worst players at a position performed vs. overall average in baseball. My own twisted version of WAR that I am now using in my HoM evaluations starts with the replacement level agreed upon by the major WAR systems (BB-Ref, FG, BBGauge). But I adjust that at each position every year by a rolling 9-year average of how the median player at each position performs vs average. I end up with a lot of the same conclusions that DanR did, particularly that shortstops are criminally undervalued throughout most of ML history, particularly 1968-87. And yes, I used median instead of positional average to prevent star gluts or extreme outliers to skew the numbers. Unlike DanR, I do treat DH as a separate position, and they have been criminally underrated throughout their history (not once in the 46 year history of the DH has the rolling-average median DH been an above average player

102. Michael J. Binkley's anxiety closet Posted: February 06, 2018 at 02:00 PM (#5620446)
I use a JAWS-like system as one of the inputs for my player calculations. The one difference is that unlike JAWS, I don't divide the value by 2. So in my system, using Dr.Chaleeko's MLE's, I have Taylor at a JAWS value of 114.12 (I do include a slight defensive regression to the mean for the NeL players). For comparison (and these are the value using my WAR calculations, not bWAR):

Palmeiro - 119.96
Murray - 119.86
Sisler - 113.79
Easter - 113.74
Hernandez - 113.46
McCovey - 111.81
McGwire - 111.52
Killebrew - 109.30
Terry - 101.92
Beckley - 101.36
Giambi - 100.75 (will make my PHoM soon after eligible)
Helton - 100.11


Bleed -
I have since revised my system since this post. I've accepted that a lot of Easter's history is speculative, so he's dropped down to below my current in/out line, but he's above the line below which I would not want to elect any players.
Killebrew, like with Dr. Chaleeko's system, now is also outside my PHoM, but he's within 5% of my current in/out line, so I have no problem if he eventually makes it to my PHoM (all the little things hurt him - poor defense and baserunning, playing in a hitter's park, in a high-standard deviation era due to expansion, and in the clearly weaker of the two leagues).
Giambi because he has a very good peak I expect to make my PHoM next year.
Beckley, because of utter lack of a peak, is now nowhere near my PHoM and never will be.
And Helton is like Easter and Killebrew, below my in out/line, but fine if he eventually makes it (Although he is above Easter and Killebrew, I have him slightly below Olerud and almost even with Berkman - Berkman just inches ahead due to post-season credit.) So I have no problem with his likely eventual election, he's just nowhere near the top of the frontlog for me.
And Taylor, I probably punished him too much for uncertainty in my most recent ballot, because Eric is fairly confident in his numbers, at least compared to some of the other Negro League MLEs. By the raw numbers, he is right out Palmeiro territory (about a 111 PEACE+).
   24. eric Posted: January 13, 2019 at 12:33 AM (#5805004)
The example begs the question. It's absolutely possible for the latter distribution to exist and it's absolutely possible for the latter distribution to exist in my system. But to assume the WAR values of the players in the two examples is to assume the positional adjustments in the two examples, which, in the case of Todd Helton, is precisely what we're discussing.


And why do the first basemen get penalized if more great players happen to be first basemen? The number of great players in the league overall hasn't changed in the two examples. The entire AAA - MLB comparison is a feint that actually supports my skepticism--at some point you have to compare players at the league level not the positional level, even across longer time spans. Unless you're on board with Jack Morris, the pitcher of the 80's? Ok, low blow, but you bring up Hodges as another example.

My analysis is based on my own statistic, Baseball Player won-lost records. To calculate the positional average for first basemen in 2000, I add up the total wins earned by first basemen, add up the total losses earned by first basemen, calculate a winning percentage from those totals, and call that the positional average. It's as close to literally the average winning percentage of first basemen in 2000 as possible. That's not to say it's the only way, or even best way, to calculate such a thing, but it's as rigorous a calculation as one can make.


And as I alluded in my first post, and others have mentioned, players can bring up the average. If you have two equivalent leagues except one has Babe Ruth, and the other has, I dunno, Jeff Francoeur, then by definition the players in the first league are "worse" than the players in the second, even if they're the exact same players. It's tough to gauge the accurate ability of a player which is why we use comparisons, but that doesn't mean the comparison method is perfect, as we've discussed with these gluts of great players at positions at times.

Ultimately, it seems you only compare position to position with no context of the entire league. I'm fully on board with not holding a SS to a hitting standard set by 1B/DH types. But being blind to the larger league doesn't seem to be a benefit to any system, as it limits the data making the results more susceptible to noise. Such limited viewpoints are a handicap to any system.
   25. Kiko Sakata Posted: January 13, 2019 at 01:03 AM (#5805007)
If you have two equivalent leagues except one has Babe Ruth, and the other has, I dunno, Jeff Francoeur, then by definition the players in the first league are "worse" than the players in the second, even if they're the exact same players.


Yes, this is true. It's true whether Ruth and Francoeur play the same position or different positions. I don't understand the point you're trying to make. The 1947 NL was better than the 1946 NL because Jackie Robinson joined the league. Barry Bonds going insane on the NL from 2001 - 2004 affected the quality of the NL. His performance those seasons led to actual wins by the San Francisco Giants which left fewer wins available for the other teams in the NL and, therefore, fewer wins available for the players on those other teams. This is just a tautology.

I'm fully on board with not holding a SS to a hitting standard set by 1B/DH types.


And how do you propose that we do so? If you want to rank players - and, to vote in the Hall of Merit, you need to rank players - you have to come up with a specific number. What is yours?
   26. Kiko Sakata Posted: January 13, 2019 at 02:04 AM (#5805011)
Sorry for the tone of the last comment. I understand and appreciate the concern that single-year positional averages are volatile. But I think if you just do something like a 5-year moving average, that's really not going to make very much difference when evaluating 10-15-20 year careers. And if you do something like an 80- or 100-year average, I think you run the risk of missing actual changes in the relative value across positions. I would also note that traditional sabermetric measures like OPS+ and ERA+ rely on single-year averages to set their baselines. We don't adjust 1968 batting and pitching stats for the average batting stats over a 5- or 10-year period; we adjust 1968 based purely on what happened in 1968 - even though 1968 was an outlier season. In fact, we do it this way precisely BECAUSE 1968 was an outlier.
   27. eric Posted: January 13, 2019 at 04:30 PM (#5805101)
Sorry for the tone of the last comment.


You have nothing to apologize for. I actually feel like I do. You've obviously put a lot of thought and effort into your system, and here I am questioning and to some degree even denigrating it. I feel a bit like someone who not only invites himself over for dinner, but then criticizes the meal.

Yes, this is true. It's true whether Ruth and Francoeur play the same position or different positions. I don't understand the point you're trying to make. The 1947 NL was better than the 1946 NL because Jackie Robinson joined the league. Barry Bonds going insane on the NL from 2001 - 2004 affected the quality of the NL. His performance those seasons led to actual wins by the San Francisco Giants which left fewer wins available for the other teams in the NL and, therefore, fewer wins available for the players on those other teams. This is just a tautology.


I agree with all that. I'm saying the presence of those players didn't make the other players any worse. They're still the same. I agree 100% that total wins is zero-sum and if a player takes more of them that means others must lose. But the addition of a great player doesn't change the ability of the other players, thus introducing (more) error into player evaluation based on comparison.

The point I was trying to make with my initial post is that you can have two leagues with the equivalent number of great players, even the exact same players. But if a few more of them are at a particular position for whatever reason (randomness, customary to the times, etc) then somehow that makes the other players at that position worse. So paradoxically the more good players are at a position, the "worse" they all are. That doesn't seem right given the uneven distribution of talent through time and across positions. I wouldn't want to dock Willie Mays because of the presence of Mantle, Snider, and Ashburn. Nor do I think, say, Bobby Murcer or Fred Lynn should get into the hall of fame due to the dearth of CF talent in the 70's.

Ultimately, I think Helton should be compared and ranked as well as possible against the all-time 1B list, but making that evaluation should involve as much data as possible. Whether or not Jim Thome switched from 3B to 1B, or Jason Giambi stuck at 3B or LF instead of moving to 1B, or Carlos Delgado stayed at C, doesn't actually effect the value Helton provided.

Now, how to evaluate players correctly? Well I'll have to get back to you on that. :) That is the "ultimate" question, after all.

I understand and appreciate the concern that single-year positional averages are volatile.


I would also note that traditional sabermetric measures like OPS+ and ERA+ rely on single-year averages to set their baselines. We don't adjust 1968 batting and pitching stats for the average batting stats over a 5- or 10-year period; we adjust 1968 based purely on what happened in 1968 - even though 1968 was an outlier season. In fact, we do it this way precisely BECAUSE 1968 was an outlier.


I think it's correct to evaluate season by season across leagues. 1968 and 1969 aren't all that comparable. However, there is still noise at the season level. And the narrower the sample size goes, the more noise there is--see park effects, or team pythag vs actual record, or defensive stats. I just think every comparison statistic has to be taken with a grain of salt, and the smaller the comparison set is, the larger that grain gets. How to resolve all those issues? Well...lots of people smarter than me haven't done it yet. But maybe I'll come up with something.
   28. Rob_Wood Posted: January 13, 2019 at 05:37 PM (#5805111)
HOM discussions have uniformly been civil and productive over the years, especially as compared to other places on BBTF and elsewhere on the internet. Occasionally posts (even by me!) stray too close to the incivility boundary but I hope that is attributable more to passion rather than malice.

I go back to early discussions of WAR and similar systems held on BBTF and elsewhere. These discussions were quite informative and worthwhile to the "consumers" of these new evaluations systems, and, I hope, to the "producers" as well. Sean Smith and other creators of these systems provided a great deal of background and motivation into these systems as well as a great deal of explanation.

IIRC the positional adjustment was a way to "close" the system. Without that construct it is difficult for offensive and defensive values to be "combined". That is, absolute defensive value is an animal difficult to cage. But if things are done in relative terms, everything is manageable.

Although the two concepts are not identical, the idea of a positional adjustment based upon the offensive performance of the players at a position (shortstops hit worse than first baseman so we will use this difference as a measure of "how difficult" it is to play shortstop relative to first base) is closely related to the idea that all positions have equal value (offense + defense).

As I have said before, I am not entirely sold on this premise and think that it leads to many difficulties, some of which have been discussed in this thread and the recent thread where we talked about catcher values (catcher boosts).

Anyway, I'd like to thank Kiko (and everybody else) for engaging in these discussions. Kiko, like Sean and others previously, has put a lot of thought, time, and effort, into developing a comprehensive evaluation system and we should not fail to acknowledge that.




   29. John DiFool2 Posted: January 13, 2019 at 06:27 PM (#5805115)
But I think if you just do something like a 5-year moving average, that's really not going to make very much difference when evaluating 10-15-20 year careers.


But by contrast you're introducting very small sample sizes into the mix, where the value of a player will jump up or down with a lot more volatility than if you used a larger sample. Classic analytical dilemma. I think the way you are doing it you are simply chasing artifacts, and not authentic player value, and certainly not from a more timeless perspective.

I think you have to, as Bill James has typically done, introduce a subjective aspect to your evaluations AND not become so deeply wedded to just one system.

[I wish the various WAR sites would use error bars or something, even tho they would look a bit goofy and might dissuade people from putting much weight on them. But maybe that would be for the best...]

I'm simply skeptical that Helton is 100 (75 or even 50) player-slots worse than Hodges, when even a basic comparison reveals, even with the Coors air adjusted for, that Helton was a pretty significantly superior player, pretty much across the board. I just checked several NL seasons in the 50's, and, while there may not have been many all-time greats at first then, 1B typically rated 2nd in the league (yep to CF usually, or left). I don't spot any major differences between the two eras in that respect. Extraordinary clains require extraordinary evidence, and what I'm seeing simply doesn't warrant such a massive flip-flop in value.

Now. If you wish to argue that batting stats "decentralized" in the Roid Era (standard deviations increased substantially), that might be a bit more to the point. I'm not sure that's what your method is establishing, tho.

[Coke to eric]
   30. Kiko Sakata Posted: January 13, 2019 at 06:57 PM (#5805123)
I'm simply skeptical that Helton is 100 (75 or even 50) player-slots worse than Hodges, when even a basic comparison reveals, even with the Coors air adjusted for, that Helton was a pretty significantly superior player, pretty much across the board. I just checked several NL seasons in the 50's, and, while there may not have been many all-time greats at first then, 1B typically rated 2nd in the league (yep to CF usually, or left). I don't spot any major differences between the two eras in that respect. Extraordinary clains require extraordinary evidence, and what I'm seeing simply doesn't warrant such a massive flip-flop in value.


This conversation has given me a great deal of food for thought. One thing that I would like to do is allow people to set their own positional and replacement levels. I think I might be able to do that for a finite set of options: single-year, some kind of smoothed series that still varies over time, and constant long-run levels (and maybe just straight 0.500 for everybody, although I tend to think that this rises to the level of "wrong" - the rest are all, I think, defensible and just a matter of personal preference). Ideally, the only "hard" numbers in my system are the wins and losses and I'd like people to have as much freedom as possible to use those numbers as best they see fit in their analyses.

Eric also raised a point that I hadn't really considered that if my comparison point for 1B in 2000 is an empirical calculation of the average of first basemen, then my evaluation of 1B in 2000 isn't taking any account of LF in 2000 or 3B in 2000 (or 1B in 1999, but I already knew that). Anyway, this isn't something that I'm going to cobble together in the next day or two. But hopefully I can figure some things out and have something to share in the next month or two.

As a first-order "sanity check" on Todd Helton vs. Gil Hodges, you can just look at their raw Player won-lost records. They were 99% first basemen who played in non-DH leagues. I mentioned in a parenthetical above that part of (actually, it turns out, most of) the difference is that Hodges looks much better in pWins and Helton much worse in pWins relative to eWins, in both cases basically because of their "choice" of teammates. Anyway, comparing their eWins, Hodges' eWins work out to a record of 219.1 - 179.1, Helton is 252.0 - 201.4. So, the difference is Helton has an extra record of 32.9 - 22.3, which is a winning percentage of 0.596, which is far greater than any positional average for any position I've calculated for any season. So, (a) the underlying numbers are, I think, telling the story that you would expect (i.e., I don't think there's a calculation error in either set of numbers) and (b) I think John is right in #29 that a claim that Hodges is better than Helton is definitely an "extraordinary claim". I think the use of pWins that tie to team wins is doing more of the work in building that claim than the differences in the positional averages for the two players (Hodges's positional average for his career is 0.518; for Helton, it's 0.526) and I tend to go back and forth on how much to care about pWins - on the one hand, they're the centerpiece of the system, and I can see the argument for "for history, we care about what actually happened; not what should have happened". On the other hand, I don't feel right holding it against Todd Helton that Rockies' management couldn't figure out how to find, develop, or sign pitchers that could handle the thin air.

Anyway, thank you all. And again, I do apologize if anybody took offense at any of my comments. I definitely have a tendency to love my own work and sometimes get overly defensive about people criticizing it.
   31. progrockfan Posted: January 15, 2019 at 10:48 AM (#5805627)
@John DiFool2:
I think you have to, as Bill James has typically done, introduce a subjective aspect to your evaluations AND not become so deeply wedded to just one system.
If I have just one mantra here, that's the one.


@Kiko Sakata:
I do apologize if anybody took offense at any of my comments. I definitely have a tendency to love my own work and sometimes get overly defensive about people criticizing it.
That's because you care. I for one appreciate your contributions, on the 'Evaluating Relievers' thread and elsewhere.

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Dynasty League Baseball

Support BBTF

donate

Thanks to
Brian
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 0.5558 seconds
59 querie(s) executed