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

Wednesday, July 17, 2002

Right Fielders

Here are the right fielders. Jim O’Rourke should be a LF, his stats are now over there.


234 - 32, 29, 28 - 133 - Oyster Burns - 8.8 sea. - 194 batting - 29 fielding - 11 pitching.
RF 66%, SS 17%, LF 9%, 3B 5%, 2B 2%, CF 1%.
notes: 1884-85, 1887-95. 5-year peak from age 22-26. Split career between NL and AA. Played 1884-85, 1887-89 in AA (122 WS), 1890-95 in NL (112 WS).

107 - 35, 16, 15 - 74 - John Cassidy - 7.6 sea. - 85 batting - 21 fielding - 1 pitching.
RF 68%, CF 24%, 1B 4%, 3B 3%.
notes: 1875-85. 5-year peak from age 20-24. Played .6 seasons in NA (not included here). Rest of career in NL, except 1884-85 (AA), 15, 4 WS.

51 - 29, 15, 7 - 51 - Spud Johnson - 2.4 sea. - 45 batting - 6 fielding.
RF 56%, LF 29%, 3B 13%, 1B 1%.
notes: 1889-91. Played from age 29-31. Played 1889 and 1890 in the AA (15, 29 WS), 1891 in NL (7 WS).

421 - 45, 41, 35 - 167 - King Kelly - 13.6 sea. - 341 batting - 77 fielding - 3 pitching.
RF 51%, C 35%, SS 5%, 3B 5%, 2B 3%, 1B 1%.
notes: 1878-93. 5-year peak from age 26-30. Played entire career in NL, except most of 1891 in AA (22 WS) and 1890 (PL) 20 WS.

202 - 30, 28, 26 - 130 - Tommy McCarthy - 9.4 sea. - 156 batting - 44 fielding - 2 pitching.
RF 48%, LF 44%, 2B 3%, 3B 3%, CF 1%, SS 1%.
notes: 1884-96. 5-year peak from age 26-30. Played entire career in NL, except 1884 (UA) 5 WS, 1888-91 (AA), 23, 22, 28, 26 WS respectively (first two years of 5 year peak).

170 - 26, 21, 21 - 95 - Paul Radford - 10.5 sea. - 122 batting - 47 fielding.
RF 60%, SS 25%, CF 8%, 3B 4%, 2B 3%.
notes: 1883-94. 5-year peak from age 25-29. Played entire career in NL, except 1887-88, 91 (AA) 21, 12, 26 WS respectively. Played 1890 in PL (21 WS).

214 - 40, 34, 26 - 118 - Orator Shaffer - 9.8 sea. - 169 batting - 45 fielding.
RF 92%, CF 3%, LF 2%, 3B 1%, 2B 1%, 1B 1%.
notes: 1874-75, 1877-86, 1890. 5-year peak from age 25-29. Played .5 seasons in NA (not included here). Played rest of career in NL, except 1884 (UA) 40 WS, 1886, 1890 (AA) 4, 17 WS.

170 - 32, 32, 27 - 130 - Ed Swartwood - 6.3 sea. - 152 batting - 17 fielding.
RF 71%, CF 14%, 1B 14%, C 1%.
notes: 1881-87, 1890, 1892. 5-year peak from age 23-27. Played entire career in AA, except 1881 and 1892 in NL (0, 2 WS).

289 - 37, 34, 27 - 135 - Sam Thompson - 10.7 sea. - 255 batting - 35 fielding.
RF 100%.
notes: 1885-98, 1906. 5-year peak from age 31-35. Played entire career in NL, except 1906 in AL (8 G).

296 - 35, 32, 31 - 147 - Mike Tiernan - 10.8 sea. - 259 batting - 37 fielding.
RF 79%, LF 11%, CF 10%.
notes: 1887-99. 5-year peak from age 21-25. Played entire career in NL.

201 - 32, 29, 23 - 113 - Chicken Wolf - 9.7 sea. - 158 batting - 43 fielding.
RF 85%, SS 5%, C 4%, 3B 2%, 2B 1%, LF 1%.
notes: 1882-92. 5-year peak from age 20-24. Played entire career in AA, except games in the NL in 1892.

Joey Numbaz (Scruff) Posted: July 17, 2002 at 10:17 PM | 123 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 1 of 2 pages  1 2 > 
   1. John (You Can Call Me Grandma) Murphy Posted: July 18, 2002 at 07:17 AM (#510529)
Oyster Burns: 26.75
John Cassidy: 14.08
Spud Johnson: 21.54
King Kelly: 30.95
Tommy McCarthy: 21.60
Paul Radford: 16.43
Orator Shaffer: 23.09
Ed Swartwood: 26.40
Sam Thompson: 27.17
Mike Tiernan: 27.55
Chicken Wolf: 20.42

Obviously, Kelly is the stud. In this order:

Kelly, Tiernan, Thompson, Burns and Shaffer (the last two are highly questionable for selection).

   2. User unknown in local recipient table (Craig B) Posted: July 18, 2002 at 02:13 PM (#510530)
I would actually have Tommy McCarthy ahead (well ahead) of Shaffer but not quite there with Burns. He should still finish well short of my ballot, and of selection; so at least in one respect the HoM will be of higher quality than the HoF.

I will have Kelly as #1 on my first ballot.
   3. John (You Can Call Me Grandma) Murphy Posted: July 18, 2002 at 02:59 PM (#510531)
I think McCarthy and Shaffer are close. Orator's 1884 season has to be devalued somewhat. I still think he edges McCarthy out (unless you are including Tommy's "pioneer" achievements?)
   4. scruff Posted: July 19, 2002 at 03:10 AM (#510534)
Put me in the camp that thinks Sam Thompson was pretty overrated. I doubt he was any better than Tiernan. He was a truly defensive butcher (these numbers don't say Tiernan was any better out there though, I don't know about his rep.) James give Thompson a C-, Tiernan a C. Modern day defensive comps would be Mike Marshall, Oscar Gamble, Albert Belle and Tony Gwynn for Thompson; Jeromy Burnitz, Barry Bonds (who WS say are overrated w/the leather), Kirk Gibson and Dave Kingman for Tiernan. Pretty amazing that Bonds and Kingman were considered similar defensively as OF's.

Both were pretty good hitter, with fairly short careers, maybe average length for the players on this list. They are both in the grey area for me. Neither will see the light of day on the first ballot, but could get on after we get rid of a few. These two are right on the area of the curve (if you remember your old Abstracts) where it starts to open up and you go from having a pack of 10 or 12 to a pack of 20-30. I doubt either will get in once we start putting our ballots together, there were just too many players that had longer, more productive careers.
   5. John (You Can Call Me Grandma) Murphy Posted: July 19, 2002 at 05:38 AM (#510535)
I agree with Scruff that Thompson was overrated and shouldn't be on the first ballot (same with Tiernan). I do think they should go in, however. They played an above average number of games for the times and had very good WS per 162 games. Maybe they are close to the cutoff, but I think they make it.
   6. John (You Can Call Me Grandma) Murphy Posted: July 20, 2002 at 04:12 PM (#510537)
I'm confused on why Thompson is placed before Tiernan in the NBHA. They played at exactly the same time with virtually the same WS numbers. In fact, Tiernan's numbers are slightly (VERY slightly) better. They look to be at a photo-finish tie. They seem to me to be (quality wise) the same player.

Thompson played in better parks for hitting than Tiernan, so his numbers are superficially better. His OPS+ is a little better, but Tiernan played in more games. I can't say you're wrong about Thompson, but it's extremely close.
   7. scruff Posted: July 22, 2002 at 08:25 PM (#510541)
Chapel -- I've been trying to figure out what those selections are based on myself, and I bet they aren't even park adjusted. My guess is that someone just took a look at the best offensive players in the league for the old-timers. There are some very questionable selections in there.
   8. User unknown in local recipient table (Craig B) Posted: July 23, 2002 at 08:05 PM (#510542)
The BR.com page had a great exemplar of why the 19th century numbers need to be rounded up to account for fewer games played...

In similarity scores, the 4th most-similar player to King Kelly is Chuck Knoblauch. Chuck Knoblauch!
   9. DanG Posted: July 24, 2002 at 08:17 PM (#510543)
I don't see any serious omissions here. A couple more right fielders who might make our first ballot:

Dick Higham 1871-78
Hugh Nicol 1881-90

DG
   10. Marc Posted: July 25, 2002 at 01:59 AM (#510544)
On July 21, ed said:

"I can't see how Thomson would make it and Tiernan would not, maybe the
line between HoMers and non-HoMers is really that thin...

If we pick an arbitrary number of players the line between the worst winner and the best loser will by definition be "that thin." It could be Thompson and Tiernan or better yet, we may find that we can squeeze three or four more players in between them. I think we are going to find this process of picking winners and losers to be painful and not altogether subject to easy and obvious rationalization in hindsight. But however thin, failing to draw a solid line and then sticking to it is the slippery slope to the lack of credibility of the real HOF.

   11. scruff Posted: July 25, 2002 at 02:45 AM (#510545)
Marc, very good point.

I think we can stay ahead of this one though. Over on the distribution list thread I went through the likely election schedule (year by year # of enshrinees). The schedule errs to the side of caution, electing fewer players early. The hope is that there will be plenty of good candidates early, and then the new players on the ballot every year will keep the standards high, and we will be still electing great players. I'm pretty confident this will work, after starting a 'mock' election myself a few months ago.

That's why I don't think Thompson and Tiernan will make it. There aren't too many slots early, and by the time we've caught up the 19th century guys, people like Cy Young and Honus Wagner and all will be coming on the ballot. I really think we are going to be okay here.
   12. John (You Can Call Me Grandma) Murphy Posted: July 27, 2002 at 08:04 AM (#510546)
63 - 32, 30, 0 - 63 - Dick Higham - 6.2 sea. - 55 batting - 7 fielding - 0 pitching.
RF 50%, C 31%, 2B 16%, CF 2%, 1B 1%.
notes: 1871-1876, 1878, 1880. Can not be properly evaluated because of lack of NA prorations. Played during the whole existence of the NA; the rest of his career was in the NL.

Win Shares/162 games: 31.25

126 - 26, 23, 18 - 82 - Hugh Nicol - 7.5 sea. - 78 batting - 48 fielding - 0 pitching.
RF 87%, 2B 6%, CF 3%, SS 3%, 3B 1%.
notes: 1881-1890. 5-year peak from age 25-29. Played in the NL (1881,1882,1890)and in the AA (1883-1889).

Win Shares/162 games: 16.61

   13. Devin has a deep burning passion for fuzzy socks Posted: July 29, 2002 at 01:30 AM (#510547)
A couple of interesting points about Hugh Nicol I came across:

1. He has, by far, the highest Defensive Win Shares/1000 innings among RF in history (4.38 - second is Jim Rivera at 3.66, and he spent a fair amount of time at CF. Larry Walker is at 3.66)

2. In 1887, he stole 138 bases in 125 games. He only got on base 188 times. (By comparison, when Rickey stole 130, he got on base 259 times.)

3. SB were only counted in his last 4 seasons, and he had 345. If you pro-rate that over his career, he'd have had 709 SB in 10 seasons.

I'm not saying he deserves to be voted for, but he does have an interesting record. (Plus, where else am I going to mention this kind of stuff?)
   14. John (You Can Call Me Grandma) Murphy Posted: July 29, 2002 at 02:47 AM (#510548)
While Nicol played in a less competitive time than now, his fielding does seems to stand out (even for his own time). Of course, he played some of his prime seasons in a weaker league (AA), so that has to be taken into account. I would seriously doubt that he would be in the majors today, unless as a defensive replacement

Stolen bases were defined differently back then, so comparisons between Nicol and Henderson won't be accurate. He does have the single season record under the old rules for his 1887 season.

He won't be on my ballot (as I take it he won't be on yours). Interesting stuff Devin!
   15. scruff Posted: October 30, 2002 at 12:55 AM (#510550)
Mark -- I could not disagree with you more. Thompson wasn't in Kelly's league as a player.

First, Kelly played about 3 seasons more than Thompson, his career was significantly longer. Because Thompson's career came later, the seasons were longer, the games played stats are deceptive. The schedules were growing from 80 to 100 to 120 to 140 during that time.

Second, Thompson played his entire career in RF, which in studies done by RobertDudek, has been shown to have a higher replacement rate than DH does today. RF was where they stuck the worst of the worst back then. My guess is that with most hitters being RH, and most of them pull hitters (pitchers weren't throwing as hard), the lack of range was more of a reason to "hide" a player in RF as opposed to putting a player with a good arm out there.

While Thompson played his entire career in RF, Kelly was a catcher for 35% of his career, and spent another 13% of his career at the skill IF positions. Think about it today, if you've got two hitters are similar, and one catches 57 games a year, plays 21 more as SS/3B/2B and plays 84 in RF and the other plays in RF 162 games who is the better player? I don't care how much worse of a defending the first guy is in RF, his ability to catch and play the infield allows you to put another RF in the lineup 48% of the time, as opposed to another C or SS/3B/2B.

OF assists rates are extremely dependant on the number of runners on base. Thompson played a few later, weren't offensive totals going up then as well (more runners to throw out)? Kelly's teams were generally outstanding, Thompson's teams were sometimes outstanding, but sometimes mediocre.

The raw stats you site are meaningless, without adjusting for ballparks, and in that era, ballparks were VERY different.

I know you have problems with the Win Shares methodology, but look at the numbers at the beginning of this thread. I can't possibly think they could be THAT far off, as to show Thompson, in a shorter career, with a much lower peak (no matter how you measure it) being better that King Kelly.

It's like comparing a guy who was 35% Mike Piazza, 13% Jeff Kent and 48% Tim Salmon to Tim Salmon. Tim Salmon playing only 127 games a year.

Kelly was probably a slightly better hitter, in a much longer career. Defense makes it a slam dunk, and is a major point in Kelly's favor, not the reason to move Thompson ahead.
   16. scruff Posted: October 30, 2002 at 03:13 PM (#510552)
Mark, I wanted to add, that using the Offensive W-L records (same as the Trammell/Ozzie argument) I get Kelly at 160-74 (.683) lifetime and Thompson at 129-61 (.680) lifetime.

Kelly's 3-year peak could be any of 1884-86, 85-87 or 86-88 (he's between .776 and .789 during those spans). Thompson's was .753 from 1893-95.

"Given that Mike Piazza is arguably the greatest catcher to ever play the game, and that Jeff Kent is top 10 material for second basemen, and that Tim Salmon is well, Tim Salmon...this is a pretty disingenous argument."

Mike Piazza, Jeff Kent and Tim Salmon are actually pretty similar as hitters. What makes Piazza and Kent great, and Salmon only pretty good is that Piazza and Kent play tough defensive positions and Salmon does not.

The offensive records I listed above take league and park into account. Thompson's leagues saw more runs scored. From the Stats All-Time Handbook, league average for Thompson was 5.87 RC/G, wheras for Kelly it was 5.4 something (can't remember, I looked last night, should've wrote it down).

Even if you give Thompson a nudge for tougher competition, the fact that Kelly played 3 full seasons longer has to offset that.

I think offensively you can call it a wash or very close. Kelly's peak was higher, but against inferior competition (not sure how inferior, but I suppose 1893-95 was a little tougher than 1884-88). Career value, again they are close, Kelly played a lot longer against slightly inferior competition.

But defensively I don't buy it. If Kelly's fielding percentages are low for an outfielder, fine, but are they low for a 2B/SS, if that's where he was really playing? Also, just the fact that he was able to catch 35% of the time makes him better, it's not like Thompson was a Gold Glover at the easiest position of the era.

I still think it's a no brainer for Kelly. Similar offense, only he was able to catch 35% of the time and play infield 13%. That's a monster edge, and it outweighs any quality of defense issues. Derek Jeter or Mike Piazza, even though they are bad defensive players are much more valuable defensively than someone like Gary Sheffield defensively, which is about where I believe Thompson was with the leather, lower middle of the pack, at an easy position.
   17. scruff Posted: October 30, 2002 at 05:06 PM (#510554)
"but how can you give a 60 game season the same weight as a 120 game season?"

My question is how can you not give them the same weight? We've discussed this on other threads, but the common denominator is one pennant. Whether it's a 60 game season, a 120 game season, 154 or 162 game season, one pennant is what is at the end of the rainbow.

If you don't give equal weight to each season, you end up giving another bonus to the players that came later, in addition to the timeline bonus we are giving in our heads already.

Also, who's to say Kelly didn't have a couple of years where he performed poorly in a short season, and his numbers would have gone up had the season been longer. You do what you can with the data you have, but I definitely think you have to adjust for season length. I don't care about ability when talking about the Hall of Merit, we aren't trying to predict how he'll do next year. We are talking about value here. Value is generally measured in terms of pennants, whether it takes 60 or 162 games to win one. Kelly's numbers are consistent from season to season, it's not like he's off the charts one year and poor the next.

I agree a fielding percentage of .820 is low, but we don't know what types of fields Kelly played on. It didn't seem to hurt his teams in the standings, did it? Those low fielding percentages should hurt him a bunch in defensive win shares, yet he has more than twice as many as Thompson.

I'm just saying when it comes to defense, especially in the 19th century, there is a lot more to it than the raw numbers. I still think you are underestimating the impact on his team's offenses by his ability to play catcher and the infield, allowing better bats into the lineup.

Sure the competition got better in a short period of time, but not so much so that you just throw away what everyone before 1893 did.

Does anyone know of a study that quantifies the impact here? I can't imagine 1894 was vastly superior to 1887.

Kelly was a much better player in his time and place than Thompson, they were similar hitters, Kelly had a longer career, played more important defensive positions. The only argument in favor of Thompson is a time line adjustment, and their careers are basically 4 years apart, Thompson was done by 1896, Kelly by 1892. I don't think a four year time-line adjustment is nearly enough to make Kelly's major advantage disappear. I just can't see it.
   18. scruff Posted: October 30, 2002 at 05:17 PM (#510555)
One other thing -- Kelly played one 60 game season, three 85 game seasons and then his seasons started getting into the 100 per game area. It wasn't like these are all 60 game seasons we are talking about. Thompson's teams were generally playing 135 games seasons, 1892, they played 155.
   19. Marc Posted: October 30, 2002 at 08:04 PM (#510556)
Someone posted a numerical rating of the quality of play in the various leagues here about six-eight weeks ago. Does anybody remember where it is, it was great and would help address some of the above debate.

Until I see the numbers again, however, I would have to guess that in fact 1894 WAS vastly superior to 1887, not because of the timeline adjustment but because of the demise of the AA. Of course the number of teams declined from 8 x 2 =16 to 1 x 12=12, not by half. But that has to have some impact.

On Kelly's side, however, is the impact at least on raw numbers of the move to 60 feet 6 inches. Thompson's SA went from the .400s every year from '88 to '92, to .500+ in '93, and .600+ in '94 and '95. Thompson's single season career highs in R, 2B, 3B, BA, OBA and SA all came in his 7th-8th and 9th 100 game seasons at the age of 33-34-35. 60 feet 6 inches had an impact on that. Kelly never had the advantage (OK 20 games and 67 AB) of hitting against pitchers an extra ten feet away.

Re. Thompson I think a relevant comparison, along with the one to Kelly, is to Hamilton and Delahanty. I don't think you can argue Thompson into their class. I know they're not eligible on our first ballot but they are nearer contemporaries (both three years younger) than Kelly was (though I recognize the point of this was to compare RFers). I am also interested in a comparison of Kelly and O'Rourke, both corner outfielders who debuted in the '70s.
   20. MattB Posted: October 30, 2002 at 09:23 PM (#510557)
Re: Thompson v. Kelly

In lieu of original thought, I though maybe we'd like to look at the new Prospectus cards.

I don't pretend to know what WARP-3 means, as it appears to not have anything to do with Star Trek. The glossary says it is Wins Above Replacement Players adjusted both for timeline and to a 162 game schedule, which sounds about what Scruff and Mark want to do. (Personally, I'm not 100% convinced that the "common demoninator" is the pennant and not the win.)

WARP-3 gives the advantage to Thompson, 86.3 Wins to 67.1 Wins.

Adjusted only for Timeline (WARP-2), Thompson wins 75.6 to 57.8.

And Unadjusted (WARP-1) Thompson wins 102.4 to 90.6.

I'd post links to the stats, but their site is down at the moment. When it comes back up, you can confirm my numbers by typing their names in here:

http://www.baseballprospectus.com/cards/
   21. MattB Posted: October 30, 2002 at 09:27 PM (#510558)
I think that I'm leaning toward Mark, although I remain open to convincing. Win Shares does seem to break down with high win team, and its really the only stat that gives Kelly the advantage.
   22. DanG Posted: October 30, 2002 at 10:03 PM (#510559)
If Thompson was better than Kelly, what about Tiernan? Previous discussion established Tiernan as essentially being Thompson's clone, maybe just a notch below Sam. So then, should we now rank Kelly below Tiernan?

On another note, whatever happened to the effort to produce win shares for the NA? Is anyone going to come up with this before our first vote?

DG
   23. Marc Posted: October 30, 2002 at 10:34 PM (#510560)
A question for the other Mark, the one who spells his name wrong.... Who was the best position player prior to 1890? How many players who peaked in the '90s rank ahead of him?
   24. scruff Posted: October 30, 2002 at 10:35 PM (#510561)
Dan -- I'm going to try my damndest to get those NA WS done.

Matt -- I need to look at the WARP-3, how do they adjust for timeline. Can you see numbers adjusted for schedule length, but not timeline? If those come out similar to the WS or my OW-L numbers, then they have a massive timeline adjustment. I'd also like to know how they calculate the defensive portion of their ratings.

I guess what I'm saying is, if we present the WARP-3 offensive numbers, adjusted for season length, but not era, I think they'd come out similar to what I presented. If they don't I'd love to know why so I can choose.

As for defense, I don't see how any rational system can rate Kelly behind Thompson defensively. I just don't see how it's possible to rate, defensively, a guy who caught as much as Kelly behind of a career RF, and one that wasn't too good in RF at that.

I don't have time to look that stuff up now, but maybe a little later, or tomorrow.
   25. MattB Posted: October 30, 2002 at 11:05 PM (#510563)
Perhaps this is too obvious to be worthwhile, but something I learned in Algebra I.

Thompson:

WARP = 102.4
WARP + TIMELINE = 75.6
102.4 + TIMELINE = 75.6
TIMELINE = 75.6 - 102.4 = - 26.8
WARP + TIMELINE + SCHEDULE = 86.3
102.4 - 26.8 + SCHEDULE = 86.3
SCHEDULE = 86.3 +26.8 - 102.4 = 10.7
Therefore, WARP + SCHEDULE = 102.4 + 10.7 = 113.1

Kelly:

WARP = 90.6
WARP + TIMELINE = 57.8
TIMELINE = 57.8 - 90.6 = - 32.8
WARP + TIMELINE + SCHEDULE = 67.1
SCHEDULE = 67.1 + 32.8 - 90.6 = 9.3
Therefore, WARP + SCHEDULE = 90.6 + 9.3 = 99.9

Thompson wins 113.1 to 99.9

Okay, obviously something's wrong with either my methodology or WARP, because it doesn't make sense that Kelly gets less of a schedule adjustment than Thompson. Nonetheless, that's what I come up with.

Also, this is from the description of the theory, re: league adjustments:

"The other thing going on in the all-time stats is a correction for league difficulty. I'm going to ask you all to wait for the details on that--I'm planning on using that for a research article in the 2003 Prospectus. Unlike the book DTs, where everything gets translated to a single difficulty level, the historical numbers are adjusted to a sloping difficulty standard. The most extreme cases are visible by looking at players from exceptionally weak leagues, like the 1884 Union Association (Fred Dunlap; note how both the hitting and fielding numbers, after adjustment, no longer stand out from the remainder of his career) or the 1882 American Association (Pete Browning). This is an adjustment that has not been made by other systems, to my knowledge."

The full description of the methodology is here:
http://web.archive.org/web/20040220155001/http://www.baseballprospectus.com/news/20021028davenport.shtml

   26. scruff Posted: October 31, 2002 at 12:00 AM (#510564)
Mark,

First, the catching 23 games would and should give Kelly a boost, I'd say maybe 2 WS. Fielding % is not the be all and end all of defense. Let me look it up. Actually for defense Kelly gets 3.6 WS, Thompson 1.9, which I think is entirely reasonable.

Second, and this is the major one, your RC totals are way off. Offense functioned differently back then. In the Stats All-Time Handbook, Kelly is credited with having created 121 runs, Thompson 109. Kelly created 8.79 RC/27 outs, Thompson 7.56. The seasons really weren't that similar. The W% for Kelly (not adjusted for park) is .704, for Thompson it's .632. When you adjust for park (see below) the gap will widen.

Kelly may get a boost from having been on a team where the sum greater than the total of the parts, after computing the players individually, James normalizes the totals so they add up to the teams runs scored, a practice I wholeheartedly support (obviously there are no clutch adjustments for 1889, there is no data). I don't have the time to see if Kelly gets a boost right now, and I don't have the formula for 1889 handy, I had to call home and get the raw numbers from my girlfriend, she'd kill me if asked her to go to the glossary and find the entire formula :-)

Bottom line, modern RC formulas, XR formulas, Linear Weights, etc. simply do not work in this time frame. (NOTE THE FOLLOWING EXPLANATION OF THE OUTS ADJUSTMENT IS WRONG, SEE POST at 10:02 PM 10/30) James also normalizes the outs a team made to total to 27 per game, when figuring RC/game. This I agree with as well, as it takes into account things like baserunning (adjusted at a team level obviously) outs, extra inning games, etc. If a team totals to 26.5 outs per game, when figuring individual RC/27 outs, Bill takes the player's outs, divides by 26.5 and multiplies by 27 before running the rest of the formula. Again, I agree with this concept completely.

Third, what were the parks like? Philly had a 109 park factor, Boston 106. This is a final factor, meaning Philly inflated offense something like 17%, Boston just 11%. Another edge for Kelly, since his park was less offense friendly.

Finally, Philly was right on their pythagorean record, Boston exceeded their's by 3 wins. Their players should and do get a slight boost for this, because their R/RA produced more victories than expected. This has nothing to do with quality of team, if their pythag was 3 wins over their actual total, Boston players would get a slight demerit, relative to their raw stats. Again, I wholeheartedly agree with this concept.

Taking all of those factors into account, it's pretty easy to see how Kelly can come out with 20.2 offensive win shares, Thompson 14.7 and I agree with this assessment of their seasons.
   27. scruff Posted: October 31, 2002 at 12:05 AM (#510565)
Thanks Matt. I think your adjustment have to be off somewhere, because like you said, there's no way Thompson should get a bigger schedule adjustment than Kelly. I'll try to take a look at it as soon as I get a chance.
   28. scruff Posted: October 31, 2002 at 04:02 AM (#510567)
"If a team totals to 26.5 outs per game, when figuring individual RC/27 outs, Bill takes the player's outs, divides by 26.5 and multiplies by 27 before running the rest of the formula. Again, I agree with this concept completely."

My bad -- I explained this wrong. James doesn't adjust based on the team, that wouldn't make sense. I meant to say that James adjusts to 27 outs based on the league. If the league averages 26.5 out/game, he'll adjust the players to the league averaging 27. This isn't really for things like baserunning, etc. as I thought, it's to account for outs missing in the record, for things like DP's, CS, etc. Sorry for the confusion.
   29. MattB Posted: October 31, 2002 at 05:50 AM (#510569)
Damn. I knew I should have paid more attention in Algebra in sixth grade. Now, I'm a successful adult, with a good job, beautiful wife, and healthy daughter, but what is it worth it I can't figure out simple algebraic equations?

Thompson:

WARP = 102.4
WARP X TIMELINE = 75.6
TIMELINE = 75.6/102.4 = 0.738
WARP X TIMELINE X SCHEDULE = 86.3
102.4 X 0.738 X SCHEDULE = 86.3
SCHEDULE = 86.3/102.4/0.738 = 1.106
Therefore, WARP X SCHEDULE = 102.4 X 1.106 = 113.25

Kelly:

WARP = 90.6
WARP X TIMELINE = 57.8
TIMELINE = 57.8/90.6 = 0.638
WARP X TIMELINE X SCHEDULE = 67.1
SCHEDULE = 67.1/ 90.6/ 0.638 = 1.160
Therefore, WARP X SCHEDULE = 90.6 X 1.160 = 105.10

Thompson wins 113.25 to 105.10

Well, at least this time Kelly gets the bigger timeline adjustment, 16% to 10.6% That looks reasonable to me, although Thompson remains ahead, even leaving aside the timeline adjustment, which is also to Thompson's advantage, but now looks fairly severe. (Kelly is discounted to 63% of his base value and Thompson to 74%. Pretty severe for overlapping careers.)

   30. MattB Posted: October 31, 2002 at 03:11 PM (#510571)
Also, under WARP, Thompson beats Kelly in 1889, to 5.7 to 5.3 Wins.

Scruff writes:

"Kelly may get a boost from having been on a team where the sum greater than the total of the parts, after computing the players individually, James normalizes the totals so they add up to the teams runs scored, a practice I wholeheartedly support."

I'm on the fence on this. On the one hand, it's a fudge factor that makes some sense. Sort of like saying, "If we did this perfectly, the numbers would add up. So, since we know what the result should be, we'll just make the numbers add up."

On the other hand, if the whole really was greater than the sum of the parts, then it is wrong, by definition, to give more value to the parts. If some of the credit goes to luck, why should that credit be given to a player. If some of the credit goes to a manager who figures out how to get more with less, why should that credit go to the player?

Also, did Sam Thompson's teammate Ed "59 OPS+ with below average fielding" Mayer contribute negative amounts of Win Shares over his 800 plate appearances that were zeroed out, taking away some of Thompson's?

http://www.baseball-reference.com/m/mayered01.shtml
   31. MattB Posted: October 31, 2002 at 03:48 PM (#510575)
"By the way, your algebra was fine...it was your data interpretation that was fine. Any good 6th grade math teacher would give you partial credit, no doubt. :-)"

Hm. You obviously never had Mrs. Molinaro for Algebra.
   32. scruff Posted: October 31, 2002 at 04:10 PM (#510576)
2 things:

Mark, the park factors on baseball reference already take into account the fact that only half the games are in the park. The factors there are directly applicable. I've asked Sean about this in the past. For example, Colorado's 2001 PF is 122, and we know Colorado inflates offense about 45-50%, not just 22%.

Second, my explanation of the outs adjustment at 10:02 10/30 was almost right. Third times a charm:

James actually figures the league outs per game and uses that for the "27". So if the league that year accounted for 26.5 outs, then RC27 is really RC26.5 for that year. Sorry for my continued butchering of this explanation, rest assured, my spreadsheets and OW-L records are calculating this stuff correctly. Sorry again for any confusion
   33. scruff Posted: October 31, 2002 at 04:10 PM (#510577)
Mark, I'll get you that formula when I get home tonight. I may have it here at work, let me take a look.
   34. scruff Posted: October 31, 2002 at 04:27 PM (#510578)
Okay Mark:

A = H + BB + HBP

B = ((BB+HBP)*.34) + ((AB-SO)*.04) + (TB*1.1) + (SB*.7)

C = AB + BB + HBP

Remember the new combining formula for individuals, NOT (A*B)/C, but:

((((2.4*C)+A)*((3*C)+B))/(9*C))-(C*0.9)

Let me know if you have any questions.

I plugged in the numbers for each (I happened to have WS calc spreadsheet here with all the formulas!):

Kelly 119.1
Thompson 112.9

So Kelly gets an extra 1.9 RC because his team scored more runs than expected, Thompson loses 3.9 RC because his team was inefficient. Again, I have no problem with this, we know the formulas aren't 100% correct, and we know the true "answer" so we should use it. A few RC either way isn't going to make a massive difference anyway.

Thompson's triple crown numbers were a little better, but Kelly walked almost twice as much (65-36) and stole almost 3 times as many bases (68-24), which were big advantages back in the day. In addition, Thompson made 17 more outs, and played in much friendlier hitter's park.

   35. scruff Posted: October 31, 2002 at 09:51 PM (#510581)
Mark, I don't think saying his value primarily comes from walks and steals is accurate. One part of his specific advantage over Thompson comes from walks and steals. But you get more steals if your park allows more hits, right? And if it's a hitters park, pitchers will probably tire quicker and walk you more.

Kelly isn't benefiting from park factors really, he played in a hitter's park. He's just being docked less, because the runs scored in his park are more valuable.

Again, when you throw in the defensive advantage, the advantage in that he had a better year offensively and the fact that he did it in a less hitter friendly park I think calling one season a 24 and the other a 17 is entirely appropriate. I don't have any problems with those numbers and I think I've done a good job of showing how those numbers were arrived at.

I think 109-106 is a reasonably big difference. It's not massive, but it's still playing half your games in a park that inflates by 17% vs. 11%. Also, those are the bb-ref factors. I don't know how they compare to the factors Stats uses, so there might be a wider discrepancy.

I think the numbers 24 and 17 are fine. I really don't see any issues. I think it's entirely reasonable that a player who created 12 more runs, using 17 fewer outs, in a worse hitter's park, while catching 18% of his teams games would be worth 2 1/3 more wins over the course of a 131 game season. I don't have any problem with it at all.
   36. scruff Posted: November 01, 2002 at 04:48 AM (#510583)
Mark, I checked the park factors in Stats, more explanation for the gap. The All-Time Sourcebook has Boston 1889 at 97, Philadelphia 108. This helps explain the 24-17 gap as well.

In both parks, the hitters were better at home, the pitchers were not. Without game by game data, to know who the opponents were, etc. we just have to take the raw numbers. I know that Stats only uses 1-year factors. Win Shares uses 1-year factors until 1909 I think.

When figuring players for their career (not peak value), it really doesn't matter if you use 1-year or 3-year factors, because it evens out over time anyway. If you use 15 1-year factors, or 15 3-year factors, the CAREER numbers will be the same. Also, with 'parks' changing all the time in the 19th Century, you pretty much have to use 1-year factors. But this explains more of the difference.

Good job Tom. I'll try to work on Kelly, but he had a lot more power than Ashburn, relative to his time. He was a much better offensive player, twice led his league in RC and RC/27, 3 times he led in 2B. And you also have the catching as well.

Maybe he was like an Enos Slaughter, only one that stole 50-60 bases a year and caught about 1/3 of the time? That'd be my most solid guess.

I looked through the NHBA for someone around 420 WS, with 2 years in the 40's and a 5 year peak around 167. I came up with:

Jimmie Foxx (435, 41, 40, 34, 173)
Eddie Mathews (447, 38, 38, 36, 167)
George Brett (432, 37, 36, 33, 154)
Robin Yount (423, 39, 34, 33, 144)
Arky Vaughn (356, 39, 36, 36, 169)
Duke Snider (352, 39, 37, 36, 171)
Reggie Jackson (444, 41, 32, 32, 148)

I realize you need to adjust for timelines, etc. But in his day, I think Kelly had the same kind of impact as these players, especially when you consider he only played 13.6 seasons.

I'll put out a similar list for Thompson, I'm looking for around 300 WS, peaks in the mid-high 30's and 5-year around 135:

Keith Hernandez (311, 33, 29, 29, 136)
Bobby Grich (329, 32, 31, 29, 143)
Stan Hack (318, 34, 33, 31, 140)
Sal Bando (283, 36, 31, 29, 143)
George Foster (269, 32, 30, 25, 132)
Jimmy Wynn (305, 36, 32, 32, 141)
Dave Parker (327, 37, 33, 31, 150)

It's tough to find similar players, because most players that peaked as high as these guys had much longer careers. But I think the two lists above reflect my feelings about these two. Kelly is a slam dunk, one of the best handful from his era, and Thompson was a very good player, but is on the fence, he may or may not make it, depending on who he's up against.

   37. scruff Posted: November 01, 2002 at 03:24 PM (#510585)
Absolutely I can see that kind of gap.

I never said he was similar to Jimmie Foxx. I said that in his time and place, he was just as valuable. Value is the key measurement here. I also don't think it's fair to take the best player from one list, and compare him to the second worst from another list.

Kelly only gets 77 of his 421 adjWS from defense anyway, he's still got an 85 WS edge based on his offense. That's a massive difference.

You just leave out the fact that this guy was a catcher 35% of his career Mark. It's like you just toss it aside as a footnote. This a massive positional differential you've got two equal hitters, and one of them catches 35% of the time and plays skill infield positions (not 1B) 13% of the time. It's a no brainer as to who is better.

Second his career was 27% longer. His best years were with shorter schedules, so of course his counting stats won't look as good in comparison. But you need to adjust for that. If you don't, you are going to say that all of the 1890's stars were more valuable than the 1880's stars, because they played longer schedules, and that makes no sense and it's simply not accurate. You have to make the majority of your evaluation based on how the player dominated his time. A timeline adjustment should strive to have equal representation from each era, decade, whatever; it should not make ALL stars from one generation rank above ALL stars from another.

If they were similar in terms of value, I'd take the guy that played later, absolutely. But these two aren't close. Kelly just has a lot more to offer. More defensive value (because he played tougher positions), much longer career, and just as good offensively. It's a no brainer.

If Bud were to expand the schedules to 200 games, will all of the best players in history be the ones who play in the next 20 years? Their counting stats are going to be obviously better. The sample size will be bigger, so we'll just say we don't know what today's players would have done with those extra games, so they get NO credit for what would have happened. That's what you are saying if you compare Kelly and Thompson's counting stats without adjusting. Not to mention the ballpark differences.

Also, the park factors weren't chosen to make Kelly look good and dock Thompson. They are calced the same for all 132 years of baseball, back to 1871. If you have a problem with them fine, but I haven't seen what the 'perfect' park factors for the 1889 NL are yet, please let me know how you'd calc them (I'm being serious, not sarcastic).

It's like you are starting with a theory (that Thompson was better) and citing everything to prove your point. I'm just looking at everything and finding an answer, I have no emotional tie to this one.

If you don't have roughly the same number of players from each era, there's a big problem in my opinion. I can't see how you can say no star players before 1890 should make it in ahead of the stars that played after. I'm not saying each decade should have exactly the same number of players, I can see a slight timeline adjustment, but one of the reasons we are going back to 1906 is to insure that all eras are represented. I'm getting the impression that if we voted for 215 guys starting today, with everyone available, you'd take Brett Butler over Kelly, because if you transported Butler back, he'd be the best player in the game. I just don't agree with that thinking. Please correct me if I'm interpreting you incorrectly.
   38. John (You Can Call Me Grandma) Murphy Posted: November 01, 2002 at 05:17 PM (#510586)
Let me jump in on Scruff's side. I have a Kelly as the better player than Thompson (and it's not even close). I took into account their respective eras and can't see how Thompson (or his WS twin Tiernan) can best Kelly.

I want to say a few words about the competition in 1893. I think this is being overstated a wee bit. If we use Dick Cramer's timeline adjustment method, slugging percentage after 1892 actually went down in 1893 (indicating the overall quality of hitting went down) and returned to 1892 levels in 1894. In fact, there is no noticeable improvement until 1896. While there were problems with the methodology when spanning eras, it makes sense when the years in question are close together.
   39. scruff Posted: November 01, 2002 at 05:36 PM (#510588)
Thanks John. Can you expand on your 2nd paragraph? I'm not sure I understand how SLG going down means the quality of hitting went down. Could just mean something in the game changed to give pitchers an advantage, parks, etc. That's the year the mound moved back (although you'd think that'd have the opposite effect), etc. I think there's more to what you said then you explained, could go a little deeper? Thanks!
   40. John (You Can Call Me Grandma) Murphy Posted: November 01, 2002 at 05:54 PM (#510589)
Let me jump in on Scruff's side. I have a Kelly as the better player than Thompson (and it's not even close). I took into account their respective eras and can't see how Thompson (or his WS twin Tiernan) can best Kelly.

I want to say a few words about the competition in 1893. I think this is being overstated a wee bit. If we use Dick Cramer's timeline adjustment method, slugging percentage after 1892 actually went down in 1893 (indicating the overall quality of hitting went down) and returned to 1892 levels in 1894. In fact, there is no noticeable improvement until 1896. While there were problems with the methodology when spanning eras, it makes sense when the years in question are close together.
   41. scruff Posted: November 01, 2002 at 06:22 PM (#510590)
"Kelly only gets 77 of his 421 adjWS from defense anyway, he's still got an 85 WS edge based on his offense. That's a massive difference.

You just leave out the fact that this guy was a catcher 35% of his career Mark. It's like you just toss it aside as a footnote. This a massive positional differential you've got two equal hitters, and one of them catches 35% of the time and plays skill infield positions (not 1B) 13% of the time. It's a no brainer as to who is better.


There's a MAJOR problem in these statements. In one breath you say that there's a large offensive WS gap. In the next, you say they were equal hitters. Win Shares does not adjust offensive numbers for position."

I see no problem. They were equal in term of quality, but Kelly beats him on quantity, because he played 27% longer. What I'm saying is that even if you completely disregard the fact that Kelly played a longer career, he was the same as Thompson in terms of quality. At WORST you can call them equal, if you give any weight to career value, you have to conclude Kelly had more offensive value in his career. I'm sorry if I didn't explain it well.

I know oWS does not adjust for position. I was trying to interpret the data in a way most favorable to Thompson. I was saying that EVEN IF YOU REMOVE DEFENSE completely, Kelly is way ahead. This is because they were pretty equal in quality, but Kelly had a much longer career. Throw in any advantage for Kelly for catching (more on that in a minute) and it becomes a no brainer, even if you go strictly on peak value.

I guess I feel you aren't looking at everything here because you haven't responded to any of my specific criticisms of your point. You brought up 1889, I think I explained that pretty well. I mention catching, and you just keep saying Kelly made a lot of errors. You even said they were similar as hitters, but you refuse to adjust for season length, and falsely conclude their careers were the same length. You keep saying that Thompson's numbers are better, but you ignore the fact that Kelly walked 20% more, stole many more bases (the first 8 years of Kelly's career we don't have steals, Thompson is only missing the first year of his career, yet Kelly stole 368 to 229). Thompson's career AVG shoots up like a thunderbolt because of 1893-95, years where the mound moved back. These 3 years are 26% of Thompson's career. That doesn't even count 1896 when his skills diminished, that year looks worse in context because of the offensive levels. You make no accounting for this, you just say look at the numbers and how can you tell me he's better? I know you know these things are important, that's why I feel you aren't looking at it with an open mind.

You can't compare their raw numbers. Thompson's combined leagues were in an environment where half a run a game more was being scored. Both WS, and the OW-L record I came up with give Kelly an edge on peak and career value offensively.

Thompson's career skims into the time where the mound moved back, coincidentally, these are his 3 best years. Yet you compare his and Kelly's raw numbers and conclude they are similar. That's why I feel like you aren't looking into everything.

In your last post, you say that I mentioned catching shortening careers, I never once said that. I said catching has value because it allows another good hitter into the lineup, and in the 19th century, catchers as a group were just awful hitters. I don't care that catching shortens your career. If you aren't playing you don't have value. I don't give any extra credit for it in that respect, I do realize many others do.

I mentioned the timeline thing (about you not having any players in prior to 1890) because of your post at 2:34 p.m. on 10/31. You said there are very few players who played their entire career before 1890, I guess I interpreted that as you saying there were very few players prior to 1890 that you would put in. In re-reading, I realize I probably didn't interpret this accurately, I apologize for that.

I still think you are making way to big of a timeline adjustment for players whose careers ended 4 years apart. The adjustment for moving the mound back in 1893 more than outweighs any timeline adjustment.
   42. John (You Can Call Me Grandma) Murphy Posted: November 01, 2002 at 06:40 PM (#510591)
Scruff:

Let me see if I can explain myself better.

The average player's slugging percentage (using 1976 numbers from the NL) went from .267 in 1891 to .285, so there was a huge leap in quality that year because of the AA folding up (which is unprecedented for an expansion year).

However, it fell to .270 in 1893 and 1894, plus .271 in 1895. 1896 shows the number up to .272 (which is still not a huge difference).

I honestly don't know what the hell happened in 1893. Maybe some historian out there can make a surmise.

I do think the quality of play improved during that time, but we're talking about a difference of maybe 4% at the most.
   43. John (You Can Call Me Grandma) Murphy Posted: November 01, 2002 at 07:22 PM (#510592)
Scruff:

Let me see if I can explain myself better.

The average player's slugging percentage (using 1976 numbers from the NL) went from .267 in 1891 to .285, so there was a huge leap in quality that year because of the AA folding up (which is unprecedented for an expansion year).

However, it fell to .270 in 1893 and 1894, plus .271 in 1895. 1896 shows the number up to .272 (which is still not a huge difference).

I honestly don't know what the hell happened in 1893. Maybe some historian out there can make a surmise.

I do think the quality of play improved during that time, but we're talking about a difference of maybe 4% at the most.
   44. John (You Can Call Me Grandma) Murphy Posted: November 01, 2002 at 07:23 PM (#510593)
I don't know why I got the duplicate post.
   45. DanG Posted: November 01, 2002 at 07:37 PM (#510594)
TomH wrote:
"I dare ya to match Monte Ward with anybody."

Actually, I did in the Shortstops thread October 24. Tom is right, there is no modern match. MLB has evolved to the point that such a performance is not possible today.

To get an idea of Ward's quality, I looked for a comparable modern position player and a comparable modern pitcher. I concluded that if you took Ewell Blackwell and moved his career six years forward and tacked it on to Maury Wills' career you'd have a pretty good match. A six-time all-star pitcher and a five-time all-star shortstop. Total win shares are over 350, an easy Hall pick.

   46. scruff Posted: November 01, 2002 at 07:49 PM (#510595)
John, thanks for the explanation, I guess that what I'm asking is -- Why is SLG a barometer? In 1987 SLG went up and in 1988 it went down, does that mean the talent level dropped? SLG went up in 1993, but we had expansion, so we know the talent level didn't go up. I guess I need more explanation of the method itself. Thanks!
   47. John (You Can Call Me Grandma) Murphy Posted: November 01, 2002 at 08:05 PM (#510596)
Scruff:
I'm not using the actual SLG numbers for the years 1891-1896. These are what the average hitter would have hit in a 1976 environment. If you look at the numbers I had, they don't remotely match the SLGs for the respective years.

Dick Cramer used a stat called Batters Win Average, which he then compiled for every player for every season from 1876 to 1980 (except for players with less than 22 plate appearances). He then compared each player's season with every other season that player had. In other words, we would compare Ted Williams in 1939 to 1940 to 1947 to 1960. Every conceivable comparison is then done.

Cramer's essay can be found in "The Hidden Game of Baseball" and "The National Pastime" (both the actual SABR magazine and the book that came out in the 80s).

   48. John (You Can Call Me Grandma) Murphy Posted: November 01, 2002 at 08:06 PM (#510597)
That should have been 20 plate appearances.
   49. Marc Posted: November 01, 2002 at 10:53 PM (#510598)
Actually, Ward is better than that, Dan...409 WS. I've already said he looks better to me than Kelly or Thompson, even Brouthers and Connor, an easy top 5, maybe top 2-3.
   50. DanG Posted: November 02, 2002 at 03:30 AM (#510601)
Marc wrote:

"Actually, Ward is better than that, Dan...409 WS."

When I said over 350 I was referencing the analysis in my October 24 posting. Basically, I noted that his adjusted WS for his years as a position player (249) plus half his WS during the years he pitched (102) was almost exactly equal to the WS sum of Blackwell (103) plus Wills (253).

Bill James admitted that the WS for pre-1893 pitchers were the biggest flaw in the system. This is why he halves tham in the NHBA for rating purposes.

Taken literally, the best player not in the HOF is Tony Mullane with 399 WS. If we halve his pre-1893 pitching WS his career total is approximately 242. If we take that number literally he ranks outside the top 300 players in history, which I think is about right. Modern pitchers in that range include Dolf Luque, Frank Tanana, Dazzy Vance, Jerry Koosman, Herb Pennock and Rick Reuschel.
   51. scruff Posted: November 02, 2002 at 05:50 AM (#510602)
Well said Mark, we can agree to disagree, no problem there.

One thing though, I'm not basing this solely, or even too much on WS. My OW-L (largely drawn from James' pre-WS methods though) come out w/a similar conclusion.

I just think that WS, while flawed, aren't THAT flawed. It's just a nice number to use to facilitate the discussion that's all, I don't mean to make them out to be more than they are. I do think they are a solid starting point though, because it all adds up, it's hard to go too far wrong.

As for the short seasons, really I'm saying give Kelly full credit for the short seasons, and once you've put everyone on a common ground, then make the timeline adjustment.

I didn't mean to imply that you were basing it on AVG and HR, I know you know better than that. But Thompson's raw numbers benefit a good deal (w/respect to Kelly) based on the moving back of the pitcher's mound. Over 30% of his career came after that.

As you said, we can agree to disagree, just wanted respond to a few points from your last post. This was a fun debate.

I'm sure there will be many more :-)
   52. MattB Posted: November 04, 2002 at 08:08 PM (#510603)
Not to dredge the debate back up, but I was away for the weekend.

Question for Scruff:

You wrote:

"Thompson's career AVG shoots up like a thunderbolt because of 1893-95, years where the mound moved back. These 3 years are 26% of Thompson's career."

Thompson's average does shoot up during that period, but these three years are not just the years that his average are highest, but the years his OPS+ is highest as well (by a significant margin).

By moving the mound back, Thompson did not merely improve, he improved relative to all the other hitters in the league.

Is a possible explanation that the shorter distance to the mound pre-1893 was handicapping Thompson in a way different than others?

I mean, the OPS+ leaders in 1892 were Dan Brouthers (1), Roger Connor (2), Oyster Burns(3) and Buck Ewing (5). These four showed declines in 1893 and beyond. Oyster Burns was young (28). The others were older, but comparable to Thompson in age.

Ed Delahanty (#4) improved, but did not peak for a few years, and did not show the spike that Thompson did.

In fact, beside Ed Delahanty, very few players finished in the Top 5 in OPS+ on both sides of the 1893 divide (Billy Hamilton did, there may be a few others.)

Could it be that moving the mound changed to way the game was played so much that only a few hitters had the skills to be among the best on both sides of the divide?

I guess my point is, isn't it a little unfair to penalize Thompson for playing better after 1893 when so many other great pre-1893 hitters failed to even match their own pace, let alone Thompson?
   53. scruff Posted: November 04, 2002 at 08:30 PM (#510604)
Very good point Matt.

I wasn't saying penalize Thompson by completely disregarding those seasons. I'm just saying that when you compare Thompson to Kelly using raw numbers, you need to water down the numbers for those years to allow for the league offense going up. That is, if Thompson was able to adjust better, great, more power to him and he gets appropriate credit for that, as long as you compare his numbers to the league averages (adjusting for park, etc.).

But you have knock those numbers down a little to account for the league. That's all I meant. Both WS and the OW-L records I calculated do that. Those methods still consider those years as his peak and he gets appropriate credit for his outstanding adjustment.

But since Kelly's career peaked earlier and he was out of the league by 1893, his raw numbers never get that boost. You have to compare players to what they did relative to the leagues they played in, that's my only real point here.
   54. MattB Posted: November 04, 2002 at 11:59 PM (#510605)
Assembling a ballot is hard. I made a list of "sure-fire top 10" players, and got about 20.

Absent a more appropriate thread, I'm putting this here, specifically to be torn apart by those who can point out what I'm missing. I am assuming my final ballot will look very different.

Personal preferences: Consistency, National Association numbers, and peak performances.

Provisional ballot:

1. Jim O'Rourke (for over 20 years, he did not have an off year, I'm pretty sure that he's the best 19th century player)
2. Roger Connor (as we discussed earlier, I prefer peak to career in certain circumstances, and have made my first base preferences known)
3. Jim Clarkson (the best 19th century pitcher in a time when pitchers pitched much more often. Should he be first?)
4. Deacon White (the strongest offenseman in traditionally defense-centered positions)
5. Dan Brouthers (very similar to Connor)
6. Cap Anson (racked up too many points coasting in his fourties and having fluke seasons in the NA. Lower peaks than the five above.)
7. Joe Start (great pre-NA player who carried through for an above average major league career)
8. Sam Thompson (nearly identical to Kelly, I think)
9. King Kelly (see above)
10. Paul Hines (yet another dominating outfielder)
11. Tim Keefe (the second best pitcher on the ballot)
12. George Stovey (the pre-Negro league black pitcher, not the fourth best first baseman)
13. Jack Glasscock (apparently, he was pebbly. The best shortstop on the ballot. George Wright maybe should be here instead for NA dominance.)
14. Hardy Richardson (the best second baseman)
15. Ezra Sutton (the best third baseman)

Ten more who are just off the ballot (in no particular order): Buck Ewing, Bid McPhee, Ross Barnes, George Wright, Tip O'Neill, Levi Meyerle, Hoss Radbourn, Monty Ward (who should probably be on, but where?), Bobby Mathews, and Dickey Pearce

   55. Marc Posted: November 05, 2002 at 12:59 AM (#510607)
Wow! I guess this is going to be really interesting! Buck Ewing, George Wright, Hoss Radbourn and Monte Ward not in the top 15!? Oh, and Albert Spalding! Wow! I've got those five, all of them, in my top 7!

And you say you want to reward high peaks: O'Rourke certainly didn't have the kind of peak that many others had, and you have him #1!

And I'm sure you mean John Clarkson.

My point is not to disagree, though I do, nor to ridicule, but just to note that the 19th century evidence is so tricky that massive differences of opinion among a pretty darn knowledgeable crew seem likely--heck, are already happening. And my point further is to ask on this basis that we please not arbitrarily throw out any ballots that we deem to be eccentric. By today's standard it is the era, the statistics, the evidence, the rules that are eccentric, not Matt's choices (or mine).
   56. MattB Posted: November 05, 2002 at 04:06 AM (#510609)
Re: O'Rourke v. Anson

That is weird, and it is one of the problems I was having in constructing a ballot. I had already gone through each position and ranked the players. That is somewhat easier because, even though you're going across eras, it's still apples to apples. When you start comparing positions, it throws everything up in the air. I had already decided that Anson was my third best first baseman, and O'Rourke was my best outfielder. I started by comparing my best at each position (plus a wild card category for multi-position players, early black players, and players with significant pre-1871 experience), and then I just picked who I saw as the best of that group of 10, replacing the chosen player with the second best on my list for that position.

O'Rourke seemed better than Connor on career value, but I had already picked Connor over Anson based on their peak, so Anson wasn't even up for comparison until Connor was gone. I've got to reassess my strategy and probably either move Anson up or O'Rourke down.

Also, I can't find an O'Rourke playing card on Prospectus, so I don't have that data to balance against the basic stats, Win Shares, and non-statistical historical info.

I also probably have an irredeemable bias against Anson that just "happened" to drop him out of the Top Five.

Re: Ewing v. White

That's a lot more straight forward. They were both catcher/third basemen, although Ewing played about 10 years later. They are fairly even on adjusted Win Shares, which has a lot to do with White's best seasons coming earlier in his career with shorter seasons.

On career length alone, White played five more seasons that Ewing (20 to 15) with at least 100 plate appearances. It's not all about peak over career, just an overweighing in that area. Ewing's higher peak doesn't make up for White's career being a third longer.

In general, numbers 13, 14, and 15 were sort of thrown on at the end, with the idea that the best player at a position over a 30 year period should at least make the ballot. Based on the current HoF, I should probably have at least one more pitcher on the ballot, but where? I started out thinking Barnes would be on the ballot as my number 2 or 3 at his position, but I never quite got around to him before I ran out of numbers. Then there's Monte Ward, who looks to be high based on his combined stats, but never quite seemed to measure up against any actual individual player, since there was always someone better than him at everything. (Better pitchers and better shortstops).

Spalding and Wright (along with Barnes) are good people to reconsider too. The 1972-5 Boston team was the first pro dynasty, and at least one of their top players should probably have made the list.

But then, who do you remove? It's a conundrum.
   57. jimd Posted: November 05, 2002 at 05:09 AM (#510610)
To get the O'Rourke player card, use 'o rourke' (blank instead of apostrophe).

I'm not sure why you say Ewing has a higher peak than White, unless you're not adjusting for season length (which is your right :-).

327 - 32, 30, 28 - 132 - Buck Ewing
332 - 42, 34, 32 - 145 - Deacon White

   58. MattB Posted: November 05, 2002 at 02:50 PM (#510611)
Thanks for the "o rourke" tip.

Maybe "peak" isn't the best way to describe the area where I saw Ewing above White. What I was noting was Ewing's 11 consecutive years with an OPS+ above 130 (although one was abbreviated), followed by a relatively short 3 year slide to retirement, compared to 7 consecutive 130+ years for White, followed by a prolonged nine year slide.

What I was referring to was something between "peak" and "career" to describe the heart of their careers. Maybe "heart" is the best term. Ewing had an 11-12 year heart where he was a sure-fire (anachronistic) All-Star each year. White's "heart" was 7 years, plus the first three non-contiguous NA years, where numbers could be exaggerated by the low level of some of the "co-op" competition.

I am certainly not arguing that Ewing was better, just where he got a "consistency bump" that narrowed the gap between him and White.
   59. MattB Posted: November 05, 2002 at 04:05 PM (#510612)
Thanks for the "o rourke" tip.

Maybe "peak" isn't the best way to describe the area where I saw Ewing above White. What I was noting was Ewing's 11 consecutive years with an OPS+ above 130 (although one was abbreviated), followed by a relatively short 3 year slide to retirement, compared to 7 consecutive 130+ years for White, followed by a prolonged nine year slide.

What I was referring to was something between "peak" and "career" to describe the heart of their careers. Maybe "heart" is the best term. Ewing had an 11-12 year heart where he was a sure-fire (anachronistic) All-Star each year. White's "heart" was 7 years, plus the first three non-contiguous NA years, where numbers could be exaggerated by the low level of some of the "co-op" competition.

I am certainly not arguing that Ewing was better, just where he got a "consistency bump" that narrowed the gap between him and White.
   60. Marc Posted: November 05, 2002 at 06:54 PM (#510614)
Good points, Andrew. It is difficult to go by position considering the multi-positions played by people like Ewing, Kelly, Ward, White et al, though you can cover all the positions I think. I'll follow MattB with a ballot for comment. I am perhaps valuing "performance above the norm" a bit too highly, which works to the favor of NA stars. And '80s pitchers maybe too highly because of the number of innings pitched; the concept of halving of their WS is actually reflected, however.

1. Clarkson
2. Ewing
3. G. Wright
4. Spalding
5. Radbourn
6. Ward
7. Brouthers
8. Kelly
9. Connor
10. Rusie
11. Barnes
12. Keefe
13. D. White
14. O'Rourke
15. Hines

It hurts not to have room for Glasscock, Caruthers, G. (not H.) Stovey, McPhee and Thompson on the ballot. D. White would be the 3B, BTW. I have a hard time justifying Ewing so far ahead of Kelly other than the obvious respect he had of old-timers circa 1936. And I'm the guy who is going to keep certain players off my ballot for one year each for alleged misdeeds (Anson, Jackson, Rose). Pls don't disqualify my ballot!

   61. Rob Wood Posted: November 05, 2002 at 08:15 PM (#510616)
Since we are floating our tentative ballots, here's mine. I have chosen, thus far, to go exclusively by the numbers. Of course, I will feel free to sprinkle in a fair amount of subjectivity later on. I have attempted to adjust for the quality of the leagues (both contemporaneous as well as timelines), to balance peak vs career value considerations, and estimate NA values. I'll have to admit that I boosted a few positions that may have been undervalued in the win share calculations for the 19th century. Anyway, here goes:

1. Roger Connor
2. King Kelly
3. Dan Brouthers
4. Paul Hines
5. Cap Anson
6. Jim O'Rourke
7. George Gore
8. Deacon White
9. Buck Ewing
10. Jack Glasscock
11. John Clarkson
12. Ezra Sutton
13. Hardy Richardson
14. Fred Dunlap
15. Harry Stovey

When the NA win shares are available, players may move around. Also, I expect to eventually find room on my ballot for George Wright, Joe Start, and probably Monte Ward.

By the way, is Amos Rusie on our first ballot? He is the only other pitcher that I would consider voting for. And does anyone know where he was in 1896?

   62. scruff Posted: November 05, 2002 at 08:33 PM (#510617)
I'm not at the point of ranking them, but these are guys I would give a YES to in a Y/N vote, going by position (they are ranked by position:

Deacon White
Buck Ewing
Charlie Bennett

Cap Anson
Dan Brouthers
Roger Connor
Joe Start

Bid McPhee
Hardy Richardson

Ezra Sutton

Jack Glasscock
George Wright

Harry Stovey

Paul Hines
George Gore

Jim O'Rouke
King Kelly

Monte Ward
Bob Caruthers

John Clarkson
Tim Keefe
Old Hoss Radbourn
Amos Russie
Al Spalding

That's 24 players already!!! I think every one of them deserves election. I've also got 4 OF's and two P's on the fence, Pete Browning, Charley Jones, Sam Thompson and Mike Tiernan also Tony Mullane and Pud Galvin too. But I couldn't vote any of them before these 24 guys though, and I can't see any of those 24 not in.

I'm wondering if we are cutting this too tight on the 19th Century guys and maybe we should start the elections earlier. I guess some of them will have to wait, there will be years where the crop at the time is thinned out some, and maybe a few of the ones missed in the opening elections will get in. I'm starting to worry that we are going to be too tight early though . . .

   63. scruff Posted: November 05, 2002 at 08:35 PM (#510618)
one mistake above, rank Connor above Brouthers, although they will be next to each other on my ballot wherever they, they are as close as could be (I might even rank them both T3 or T6 or something).
   64. scruff Posted: November 05, 2002 at 08:40 PM (#510619)
Ed McKean has a decent case too, he's clearly the 2nd best to Glasscock (if you disregard Wright for now). So many players, so few spots . . .
   65. MattB Posted: November 05, 2002 at 08:44 PM (#510620)
Scruff,

How would you score a ballot that had a T3 on it?

And would it star Arnold Schwartzenegger?
   66. John (You Can Call Me Grandma) Murphy Posted: November 05, 2002 at 08:50 PM (#510621)
I have to admit that I like the different ballots that I am seeing. I rather have it that way than to have a "rubber stamped" ballot. There is more suspense this way.

Of course, they are all wrong. :-)

Seriously though, I'm having a hard time trying to create positional adjustments based on durability for each position. If not for that, I would have my ballot posted here also.
   67. scruff Posted: November 05, 2002 at 08:51 PM (#510622)
I would split the points for 3rd and 4th between the two players tied for 3rd. Just like if you are tied in a fantasy league stat category, for example.

Liked the Aaanuld comment.
   68. John (You Can Call Me Grandma) Murphy Posted: November 05, 2002 at 08:52 PM (#510623)
I would rather have it that way than to have a "rubber stamped" ballot.
   69. MattB Posted: November 05, 2002 at 09:39 PM (#510624)
Interesting note:

Of the ballots posted so far, there are no high ranking players who could be considered "Stars of the American Association" (Latham, O'Neill, Silver King, Foutz, Carruthers, and Charlie Comiskey, to name just the St. Louis players). Bid McPhee was listed as possible on Scruff's list. No mention of Tony Mullane, John Reilly, or Oyster Burns.

Will the Best Players of the 19th Century be coextensive with the Best NL Player of the 19th Century?
   70. John (You Can Call Me Grandma) Murphy Posted: November 05, 2002 at 10:11 PM (#510625)
I would rather have it that way than to have a "rubber stamped" ballot.
   71. John (You Can Call Me Grandma) Murphy Posted: November 05, 2002 at 10:17 PM (#510626)
Don't ask me about that last post. You got me.
   72. scruff Posted: November 05, 2002 at 10:47 PM (#510627)
I definitely think McPhee should be in, he's the Brooks Robinson of his time.

I think Caruthers should be in too.

I do think the AA was the inferior league though. Also, the league wasn't around that long, so anyone who was exclusive to the AA, would have a short career, or have played in both leagues (save the pitchers, whose careers were generally a lot shorter).

I realize you were just shooting off names, so don't take this the wrong way Matt, but Charlie Comiskey shouldn't be anywhere near the Hall of Merit. As a player it would be much worse than electing Tino Martinez. As a human being, he's up there with Bud Selig.
   73. scruff Posted: November 05, 2002 at 10:50 PM (#510628)
Just to clarify, in case I gave the wrong impression at 4:47, Comiskey's being the king of all ######## shouldn't have anything to do with his evaluation, we are talking about players, not personalities (unless the personality affected his teammates in a directly negative to wins and losses manner).
   74. MattB Posted: November 05, 2002 at 10:52 PM (#510629)
I agree. I included him because, whatever else, he is a Hall of Famer and therefore has earned to right to be consciously excluded unlike, say, Tommy Tucker, who, although a better player, could have just gone by unmentioned.
   75. MattB Posted: November 05, 2002 at 11:41 PM (#510631)
A little comparison, because I'm having second thoughts on my pitcher rankings.

Tell me why Pud Galvin (1879-1892) is not a better pitcher than his contemporary John Clarkson (1884-1894):

Points for Clarkson:

ERA+ of 134, compared to 108 for Galvin.
k/9IP: Clarkson 3.92 vs. 2.70
Nonhomer hits per 9IP: Clarkson 8.20, Galvin, 9.42 (but how much of this should go to the defense? I don't know, but certainly some. Clarkson played for better teams).

Points for Galvin:

Adjusted Win Shares (as posted by jimd in the pitcher thread):

Galvin: 621 (most for a pitcher); Clarkson 491

IP: Galvin pitched 33% more (6003 to 4536 IP)
BB/9IP: Galvin 1.12 to Clarkson 2.36
HR: 161 for Clarkson (0.32/9IP), 122 in 3/4 the innings for Galvin (0.18/9IP)
Wins: 364 for Galvin, 328 for Clarkson

Basically, Clarkson struck out one more per game, and gave up one less hit on balls in play per game, while Galvin walked one less per game, gave up homers at about half the rate, played a lot longer, and won more games for worse teams.

Why is the consensus unanimous that one of these contemporaries is on the top of the ballot, while the other doesn't crack it? Does ERA+ trump?

   76. jimd Posted: November 06, 2002 at 01:36 AM (#510633)
(Posting this in the right discussion...)

I might as well post my preliminary ballot also. I drew up 4 different ballots:
1) Career Double-Defense Adjusted Win Shares
2) Estimated Peak Value from Double-Defense Adjusted Win Shares
3) Career BP Player Cards
4) Peak Value from BP Player Cards,
then awarded points using our ballot system and totaled them up to form one combined ballot.

.1) 64 (16-12-23-13) Connor
.2) 54 (24-xx-24-06) Anson
.3) 53 (12-09-21-11) Glasscock
.4) 52 (23-07-22-xx) O'Rourke
.5) 51 (xx-21-08-22) Rusie
.6) 49 (15-06-16-12) Brouthers
.7) 44 (22-08-14-xx) Ward
.8) 44 (21-11-12-xx) Deacon White
.9) 40 (xx-16-xx-24) Barnes
10) 39 (13-13-13-xx) Hines
11) 38 (xx-22-xx-16) Radbourn
12) 33 (09-24-xx-xx) Spalding
13) 26 (11-xx-15-xx) McPhee
14) 24 (xx-15-xx-09) George Wright
15T 24 (14-10-xx-xx) Kelly
15T 24 (10-14-xx-xx) Gore

And I've got 12 more names that made it onto at least one of the ballots.

Now I'll decide whether I want to tinker with the results. Glasscock and Rusie's rankings sure surprised me. Real NA numbers may change this also.

(I don't understand the infatuation with Clarkson. I'll have to study him some more...)

   77. Marc Posted: November 06, 2002 at 01:52 AM (#510634)
Some responses:

>I definitely think McPhee should be in, he's the Brooks Robinson of his time.

I think McPhee should NOT go in, he's the Brooks Robinson of his time.

>Tell me why Pud Galvin (1879-1892) is not a better pitcher than his contemporary John Clarkson (1884-1894):

>Points for Clarkson: ERA+ of 134, compared to 108 for Galvin.

End of story. WAAAAAYYYYY to big of a difference there. Nothing else could possibly matter after that.

And yes, Rusie is on the first ballot. He held out in 1896, all year, because of a difference of opinion, shall we say, with the criminally insane Giants owner Andrew Freedman. He hurt his arm in '98, sat out two full years, came back in '01, but aborted the comeback after just three games, so his "official" retirement was year-end '98.

Finally, when I think that we will elect about 30 players by 1920, and when I see lamentations that we might "only" elect 24-25 from the 19th century, and when I see (with all due respect to scruff) Charlie Bennett as a possible HOFer, I worry that we are going to elect too many. I can't get much past 15 without dragging Young and Nichols and Delahanty to the party.
   78. Marc Posted: November 06, 2002 at 01:58 AM (#510635)
Also, scruff, was it you who suggested that about 30 percent of our HOFers should be pitchers? The half-WS formula is going to put us in a deep hole re. 19th century pitchers, yes? No Clarkson, no Keefe, no Galvin, no Rusie? And Rob, one pitcher among the first 15? I need some help understanding that.
   79. dan b Posted: November 06, 2002 at 03:33 AM (#510636)
"24 players already!"

Not a prayer that we put that many players in who are eligible on the first ballot! I haven't filled out my top 15 yet, but when I do, it will include at least 5 guys who will never climb high enough on my ballot to get the bonus points. After Anson, Brouthers, Connor, Ewing, White, Clarkson, Rusie, Kelly and maybe O'Rourke and Radbourne are in, I don't see anybody I will be listing above Childs, Delahanty, Collins, Nichols, Duffy, Ryan, Davis, Young etc. - I believe the 19th century will be fairly represented, but it will be mostly with players whose career extended into the 20th century.
   80. MattB Posted: November 06, 2002 at 04:38 AM (#510637)
">Points for Clarkson: ERA+ of 134, compared to 108 for Galvin.

End of story. WAAAAAYYYYY to big of a difference there. Nothing else could possibly matter after that."

Somehow, I put my response in the wrong thread (must be contagious today.) Here it is:

Clarkson leads on ERA+, but his ERA was almost identical to Galvin's (2.81 to 2.86). The difference is therefore almost entirely due to "park factors", but how reliable are park factors when one pitcher is pitching over half your innings. Let's look.

Clarkson was Chicago's number one pitcher from 1885 to 1887. Chicago used West Side Park for those three years and three years after that (1888-1890).

Pitching Park Factors (provided by baseball-reference) for the Clarkson years: 108, 110, 110. Pitching Park Factor the next three years: 106, 104, 103. Did the park change? Not that I'm aware of. This result could entirely be caused by Clarkson himself pitching better on the road.

Then Clarkson goes to Boston: Park Factor over his four year tenure: 100, 104, 105, 109. Did Boston's park (it was the same all four years) go from neutral to increasing offense by 18%, or is it more likely that Clarkson (or one of his colleagues) skewed the numbers by a couple of bad outings at home and good outings on the road?

It seems a lot more reasonable that one or two pitcher can skew the numbers when you see that they was pitching such a large percentage of their teams innnings.

Meanwhile, I showed that Galvin was better on walks and homers allowed. Walks is not park-dependent and while homers can be, the gap between them was huge. Galvin also has more Wins, more Wins Shares, and leads on Baseball Prospectus's WARP3 by a fair amount (80.7 to 66.9).

Clarkson struck out more and allowed fewer hits on balls in play, but strikeouts were a smaller part of the game then, and hits on balls in play are necessarily more defense dependent, irrespective of how closely the correlations match those seen today.

Meawhile, I believe that Galvin spent most of the late 1870s playing in the California League at a time when it was not obvious going in that the California League would turn out to not be a "major league" with stats that would "count". That will pad at least his counting stats (not that he needs more wins).

Just a few reasons to look past ERA+...........
   81. DanG Posted: November 06, 2002 at 03:17 PM (#510638)
Marc asked:

"Also, scruff, was it you who suggested that about 30 percent of our HOFers should be pitchers? The half-WS formula is going to put us in a deep hole re. 19th century pitchers, yes? No Clarkson, no Keefe, no Galvin, no Rusie? And Rob, one pitcher among the first 15? I need some help understanding that."

Actually, I suggested that. I'm not the big number cruncher, and there are many contradictions and difficulties, especially when analyzing the 19th century. So I like quotas.

From what I've studied, 30% pitchers is a well-established percentage. Win shares disses pitchers compared to other methods, more like 25% pitchers. F'rinstance, I did a WS study of HOF candidates for the upcoming BBWAA ballot (which I'm hoping Sean will put up soon, hint, hint) and it's surprising how poorly Blyleven compares to the position players.

So I don't entirely buy the WS numbers for post-1920 pitchers; they seem about right for 1893-1920 pitchers; it seems WS should maybe be halved for pre-1893 pitchers.

I haven't looked at it in detail, but maybe this is the solution: rather than halve Clarkson's total WS, what if we just halve the pitching WS, leaving alone the batting and fielding WS? Does this yield reasonable totals? Maybe a factor for shorter seasons needs to be worked in? Or maybe the factor shouldn't be as low as .50 (James seems to have come up with that off the top of his head)? Try using WS*.60, or something else.
   82. Rob Wood Posted: November 06, 2002 at 08:47 PM (#510640)
I think one reason why people started floating their preliminary ballots was to see how much "consensus" we had for the 19th century players. Just based upon the early ballots and comments, there does not seem to be much. In fact, there seems to be quite a bit of disagreement, even at the very top of the ballot. I guess this is reflecting all the particular issues associated with evaluating 19th century players (quite a challenge to say the least).

Another reason to post preliminary ballots is to get reactions from people. My list, based largely upon the adjusted win share figures posted previously, had George Gore as the 7th best 19th century player. I know little about Gore, so I take the figures on face value only.

However, just because someone is ranked higher than expected does not mean that the rankings are dubious. Indeed, one of the benefits of systematic evaluations is ensuring all players are treated fairly. If there is some aspects of the win shares system that doesn't work so well for 19th century baseball, I am all ears and would not hesitate to adjust the figures in light of that information.

Finally, I just do not see much "value" arising from 19th century pitching, especially the early years. This is why I'll only have two pitchers (Clarkson and Rusie) on my first ballot. Nichols and Young will undoubtedly be the next two pitchers to find a place somewhere on my ballot.
   83. scruff Posted: November 06, 2002 at 08:55 PM (#510641)
Marc, if you don't think Brooks Robinson should be a Hall of Famer, your standards are way too high in my opinion, or you don't think defense is worth anything at all.

Brooks had a pretty high (though short) peak, and a very long career. He's a no brainer in my opinion, one of the 7 or 8 greatest 3B of all time.

Charlie Bennett has a pretty decent case after you get past the big two, he's the best catcher (except for maybe John Clapp). He was 24th of the 24 and I could see why you probably wouldn't think he should be in. I'd rate him about even with Sam Thompson, Mike Teirnan, etc. I probably should have put him in with them and called it a 23 man list.

I think 30% pitchers is pushing it a little too high actually, but I really don't have any set quota. I think individual pitchers are generally overrated as to their impact, especially today, when only a handful of guys throw more than 225 innings in a season.

As for the quota, I think something like what James said in the first HBA is appropriate, I'll paraphrase, since I don't have the book handy:

Basically you want a team, one at each regular position, four starting pitchers (preferably a lefty and a righty among the 4) and sprinkle in relievers and multipositional guys where appropriate.

The second rule was don't pay too much attention to the first rule. If there are 3 1b that are better than everyone else (pretty close to true here) you don't have to put a C in.

I pretty much agree with that, although I don't see the need to distinguish between lefty and righty pitchers. Outs is outs. But everything else is pretty right if you ask me, and that does work out to about 30% pitchers.

Like I've said elsewhere, I don't see a problem with 20 OF's and 6 or 7 3B, but I do see a problem with 25 OF's and 2 3B (or 18 OF's and 13 3B). Somewhere inbetween is a reasonable balance.
   84. scruff Posted: November 06, 2002 at 09:00 PM (#510642)
I dunno Mark, I think Gore was a helluva player, maybe not 7th, but he should definitely be enshrined, he was one of the 4 or 5 best OF's of the 19th Century. He and Hines are far and away the best CF's of the 19th Century.

Maybe that's a reasonable standard. Find the best guy at a position, and elect those who are pretty close to him as well. This would work pretty much everywhere, and is actually basically what my list is, except for Bennett (who I probably overrated) who is clearly behind White and Ewing.
   85. scruff Posted: November 06, 2002 at 10:54 PM (#510644)
Mark, I think we disagree on the degree of inflation when you get away from the mean too far. I don't think that difference is enough to make players like Gore and Kelly worse than Sam Thompson.

It depends on what you consider reality Mark. If you don't adjust for season length you are shortchanging the players that played during the shorter season, or overrating the players that played during the season. I think it makes a lot of sense to put everyone on an equal playing field with regards to season length and then worry about adjusting for a timeline bias. Unless you think schedule length is directly correlated with the strength of the league, you must do this, or your numbers will be flawed.
   86. Rob Wood Posted: November 06, 2002 at 10:55 PM (#510645)
Mark, I appreciate your comments and I'll have to think more about how to incorporate those issues into my evaluations. Here are my top of mind thoughts:

1. I don't really share your concern about using win shares even for teams with very high win pcts. I would be more concerned with the "linearity" assumptions inside the claim points formulas for 19th century ball. If one is willing to swallow those assumptions, one should be able to swallow the scaling up to team victories. To be perfectly honest, I have more concerns about how the win shares are allocated between pitcher and fielders in the 19th century.

2. I fully subscribe to the view that scaling up each MLB season to the same basis (i.e., number of games) is appropriate. A pennant in 1885 is just as valuable, in some sense, as a pennant in 1985. If your concern is that the scaling up process will exaggerate the player's win shares, I looked into that issue awhile ago. Using some hocus-pocus sampling theory, I estimated that the scheduling uplift may exaggerate the best players' win shares by about 3-5%. The 1885 NL schedule was 112 games, so I recommend a 3% discount (I use a 5% discount for pre-1883 seasons).

Finally, Gore's 1885 season is not all that great of a season for him. If it were, then I would share your concerns even more. As it was just another great season by Gore, I think the adjusted win share figures pretty fairly represent his contributions.
   87. scruff Posted: November 06, 2002 at 10:56 PM (#510646)
"overrating the players that played during the season"

should be "overrating the players that played during the LONGER season" apologies . . .
   88. scruff Posted: November 06, 2002 at 11:06 PM (#510647)
Rob there are some definite issues w/WS in the 19th Century (or any year with a short schedule).

For example . . . several of the claim point formulas have straight line adjustment, like the pitcher HR component on the team level (CL-4). I noticed this when trying to test WS on a 7-game Diamond Mind series I played (to see if it concurred with who I though should be MVP, it did by the way :-))

Anyway, what would be great would be for me to rework the spreadsheet to adjust these things out to 162-game schedule and then refigure WS for a season and see how much difference it makes.

These 'issues' would trickle through, basically it has the effect of not adjust as strongly as it should in certain areas. If a team gave up very few HR, the team's pitchers would be a little underrated and the fielders a little overrated, etc.

I don't think it's a MAJOR flaw, just one to be aware of.
   89. jimd Posted: November 06, 2002 at 11:07 PM (#510648)

1. Win Shares was not designed to handle players on teams that won 78% of their games (see Gore's best season--1885)


Win Shares understates the contributions of players on great teams. This is because the linear approximation used to convert marginal runs to wins assumes these teams will get more wins than they actually do. In reality, too much of the marginal production goes into bigger and better blowouts instead of actually winning more games. If you want to get the most Win Shares for a given set of stats, get on a bad team.

This effect is counteracted by a quality-of-opposition effect. In an 8 team league, a .500 team is playing a .500 schedule. A .780 team is playing a schedule whose average strength is only .460, so maybe they deserve to be docked a little. Because you can't play against yourself, your pitchers don't face all the best hitters, and your hitters don't face all the best pitchers. However, I've never looked at the relative magnitudes of the two effects to see which is the larger.

There may be other internal distortions I don't know about; imbalances between offense and defense or whatever, but I don't think this particular one is a big problem. Even at .780, each regular is only getting screwed out of about one Win Share apiece. At .830 it's about two each (1884 St. Louis UA and 1872 Boston NA); and at .900 it's about four each (1875 Boston NA). Bad teams experience similar distortions, but receive extra Win Shares instead of losing them.

If there's something that I'm overlooking here, I'm sure you guys will let me know ;-)

   90. scruff Posted: November 06, 2002 at 11:54 PM (#510650)
Mark, I think 'out of whack' is a bit of an exaggeration. In need of slight adjustment, sure, but out of whack, no way.

I also think the extremes are further out than .750. Where they break down is with .900 PLAYERS, guys like Bonds, Williams and Ruth. There are quite a few .750 players every year, guys like Eddie Murray, etc. and most methods work fine for those players. I think win shares are fine between .250 and .750 on the team level.

The issue is with teams like the 1899 Spiders, who played .130 baseball. Tommy Dowd and Joe Quinn were decent players but there just aren't enough WS to go around, so they might get screwed, but like Jim said above, it's also possible that since everyone else was terrible, these two would get almost all of the WS when they shouldn't deserve them. It's hard to tell which factor is stronger.

But on a good team there are plenty of WS to go around, but there are also plenty of good players to gobble them up. I don't really think it's an issue either way. We're talking a WS here and there, nothing that would give major distortions either way.
   91. scruff Posted: November 07, 2002 at 12:07 AM (#510651)
And Jim is right about the great players getting hurt on extreme teams, at least in theory. Think about an all-star team of Ruth, Mantle, Bonds, Schmidt, Honus, Morgan, Gehrig, and Berra. With a rotation of Koufax, Maddux, Johnson (pick Wally or Randy), and Mathewson, with Wilhelm closing and Gossage setting him up, all players in their best years of course

That team is going to go something 155-7 or close to it. But there aren't enough WS to go around. If you figure the bench will grab 15 WS (1 or 2 each), that leave 450 WS for the 14 above. That's 32 apiece. Everyone one of them, except maybe the closers had a year better than that though. The OF's all had years in the 50's. So at the far extremes, jim is correct.

Now put Bonds on the 1899 Spiders. Maybe they win 14 more games. So now we are talking about a .260 team (40-114). That's only 120 WS to go around, figure Quinn and Dowd will get 15 each, the fielders have to get 25 by definition (I think). The pitchers will get a few by definition also.

Barry isn't going to 55 of them (which is what his 2001 season was IIRC). There wouldn't be enough WS to go around so a great player on an extremely bad team would be underrated. Again, there are rarely great players on extremely bad teams, so it's not an issue, especially from an HoM perspective.

Again, I don't think there is much distortion on a .750 team, those teams are great because they have great players, who deserve a lot of WS.
   92. jimd Posted: November 07, 2002 at 12:47 AM (#510652)
Here's an example showing where I'm coming from.

Assume I have a .500 team (81 wins) with a 20 WS 3B-man. Sign Cap Anson 1876 (32 AWS) to that team. The team's improved by 12 WS so we'd expect the team to win 85 the next year, AOTBE. Anson adds 12 WS of marginal production and the team produces 12 extra WS.

Now do the same thing to the 1875 Red Stockings. Their (adjusted) record is 146-16. Assume Harry Schafer is a 20 AWS 3B-man (he's got to have a great glove to get any more :-). Would you expect this team to now go 150-12? No, because they're way past the point of diminishing returns, where the exponent on Pythagoras has kicked in big-time. Only 10% of the marginal runs added by Anson can improve the record, instead of 50% when he is added to a .500 team; the other 90% goes to running up the score. Anson adds 12 WS worth of marginal production, but the team does not produce 12 extra WS, so instead the available WS get spread thinner when the accounting is done.
   93. jimd Posted: November 07, 2002 at 01:22 AM (#510653)
(i.e. Kelly and Gore seem to have unrealistically high WS totals),

I say to this -- "Somebody has to get credit for winning those games". If it wasn't Kelly and Gore, then it was Goldsmith or Quest, or somebody.

That's what I like most about Win Shares, whatever it's flaws in allocating the credit. The individual totals can't be off by 10 wins from the team totals and then the author says, "They got (un)lucky". If the little-things/intangibles/missing-links aren't being measured or weighted or allocated properly, then the whole team shares the credit/blame until they are.

   94. jimd Posted: November 07, 2002 at 03:12 AM (#510655)
That would be difficult, particularly since it figures into the WS formulas for catcher defense.
   95. scruff Posted: November 07, 2002 at 03:25 AM (#510656)
Mark, I'll look into them specifically later, but my first thought would be that Catcher A was much better defensively, and fielding percentage, especially for catchers does not tell you who the better defensive catcher is.

I'll let you know more later, but that's my initial guess and I have no idea who is who yet.
   96. scruff Posted: November 07, 2002 at 04:00 AM (#510658)
Catcher A - Barney Gilligan 114 G for the Providence Grays. They went 84-28.

Catcher B - Charlie Bennett (my man!) of the 28-84 Detroit Wolverines, who also played 114 G.

Let's look closer. First we'll look at the offense. Per the Stats All-Time Handbook, Bennett created 63 runs, 6.77/27 outs. Gilligan created 53 runs, 6.44/27 outs. Bennett made 251 batting outs, Gilligan 222 outs, so Bennett was a a little bit better offensively, but not much. This doesn't account for park.

The raw numbers show Bennett was better, let's put them in the formula and see how much each was adjusted, based on the team over or underperfprming their RC numbers:

Gilligan: 47.8 runs. He was given 5.2 extra runs created, because Providence outperformed their runs created estimate. This has nothing to do with them being a great team. It's because they were more efficient than their offensive stats would have otherwise indicated.

Bennett: 64.8 runs. He was docked 1.8 runs because Detroit underperformed their RC estimate. I have no problem with that. This has nothing to do with them being a terrible team. It's because they were less efficient than their offensive stats would have otherwise indicated.

According to Stats, the park factor favors Gilligan. Providence's park factor was 88, Detroit's was 95. Bennett is a little bit better on the raw numbers, Gilligan has a park advantage. I'd say Gilligan probably ends up with a slight edge per out, Bennett a slight edge because of the extra PT. Not knowing anything about any evaluation system, and know these facts, I'd say the players should rate about even on any given offensive metric.

What does offensive WS say?

Gilligan: 8.2
Bennett: 7.7

I really don't have any problem with this. You've got two players that after adjusting for park and team efficiency, we see as having similar raw stats. They end up with similar WS numbers as well.

Where Gilligan gets his big edge is on defense. He gets 5.2 defensive WS and Bennett gets just 3.2. I don't really have a problem with that. I'd say there's a good chance the catcher on a .750 team is better defensively than the catcher on a .250 team, and that's where the difference comes from.

Again, your problem may be with the way runs created were calculated, or the way the park factors were calculated, but you've got a case here of two guys with similar RC numbers (based on the park factors fed into WS) ending up with similar WS. I think this proves that .250-.750 is a reasonable boundry (probably further boundry is appropriate even). I am absolutely convinced WS does not inherently distort player values based on whether or not the player played on a good or bad team, or even an extremely good or bad team. .130 teams probably have a distortion. .870 teams probably do also. But there aren't too many of those to worry about.


   97. scruff Posted: November 07, 2002 at 04:12 AM (#510659)
Mark, team wins are a part of the CS factor for catchers before there is actual CS data. It actually takes into account assists and team victories. I don't have any problem with that assumption. Without any data, it's reasonable to assume that teams that won more games had catchers that were better at throwing runners out than those that didn't. Teams that win more games, generally have fewer baserunners trying to steal on them as well. It's only one part of the catcher evalution formula and I don't have a problem with it.

I know you know this, but there's a lot more to evaluating defense than fielding percentage, especially at catcher. If you take away team strikeouts, Gilligan and Bennett had the same number of putouts not by strikeout (each had about 40 more PO than the team had K's, don't forget, backups had a bunch of those K PO as well). But Gilligan had less opportunities to make plays, because his pitchers struck 150 more batters, and he presumably had better fielders playing with him, so they made more plays, leaving fewer for him.
   98. User unknown in local recipient table (Craig B) Posted: November 07, 2002 at 05:55 PM (#510661)
So many comments...

If your concern is that the scaling up process will exaggerate the player's win shares, I looked into that issue awhile ago... I estimated that the scheduling uplift may exaggerate the best players' win shares by about 3-5%. The 1885 NL schedule was 112 games, so I recommend a 3% discount (I use a 5% discount for pre-1883 seasons).

No way, not for pitchers. The presence of so many off days within the schedule gave pitchers a much larger advantage than 3-5%. In the NA, it's extreme... in 1872 or '73 with teams playing one game every three days, pitchers pitch every game. By 1875, with the schedule expanding to over 70 games, most teams carry two pitchers, with the top starter working about as much as he did in 1872. But the percentage of team games pitched has declined (a few guys, Jim Devlin in 1876 for instance, are still ding the ironman thing). As 1883 comes around and schedules are 100 games, a third pitcher starts making an appearance, splitting team WS even further. There are still lots and lots of off days in the schedule, teams are only playing four or five times a week.

In short, I don't think WS are scalable for pitchers until playing every day (well, six days a week) becomes the standard around 1890.

we find teams who win fewer games than a team full of "replacement" players might be expected to

I'm not sure we do. I mean, we find bad teams who are also unlucky and don't meet their Pythagorean (1890 Pittsburgh Alleghenys, for example) and teams who manage to underperform the marginal rate on either offense or defense (ditto, on defense, and you have to see it to believe it), but you don't really find anyone with the winning percentage below .136 or so that James's all-marginal team would manage. (The all-marginal team allows three times as many runs as it scores, and 0.33^1.81 is about .136)

Next, I want to deal with the criticism of WS that it gives a bonus to players on good teams.

This is absolutely true, it does. It does this for an excellent reason, as well.

Good teams (teams with good win%) are marginally more likely to have outperformed their Pythagorean ratio than bad teams (with a low win%) are. If two players have similar ability to create and prevent runs, and one plays for a team that exceeds its Pythagorean ratio and the second plays for a team that doesn't meet its Pythagorean ratio, the player on the more efficient team will have more WS.

I think this is entirely appropriate.

Trust me, I'm very sceptical of WS as well, but not nearly as much of the methodology as of the farcically ad hoc way of giving defensive credit.
   99. Rob Wood Posted: November 07, 2002 at 06:43 PM (#510662)
Just to be clear. My comment about discounting schedule scaling that Craig replied to was only intended to apply to position players. See my original post if you want to see what issue I was concerned about. I have absolutely nothing to say about 19th century pitcher win shares, except that I don't put much faith in them for all the reasons that people have raised.
   100. User unknown in local recipient table (Craig B) Posted: November 07, 2002 at 09:26 PM (#510663)
Thanks, Rob. I did misunderstand you... I think you're right for position players.
Page 1 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
rr
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 1.4092 seconds
49 querie(s) executed