Baseball for the Thinking Fan

Login | Register | Feedback

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

Monday, January 10, 2005

Dick Lundy

Since I can’t come up with anything whimsical like my Cary Grant and Felix Unger allusions for Judy Johnson and Oscar Charleston, I’ll just say that Lundy was a hellacious shortstop and hope that we give him the proper credit he deserves.

John (You Can Call Me Grandma) Murphy Posted: January 10, 2005 at 11:57 PM | 190 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

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

Page 2 of 2 pages  < 1 2
   101. rawagman Posted: October 19, 2007 at 11:26 PM (#2584723)
I realize now that Lundy wasn't even in my consideration set before now. I won't change my ballot this tinme, but iit may well have an effect for next year.
   102. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 19, 2007 at 11:55 PM (#2584741)
OK, here is what this translates to, using league average figures for triples:

Dick Lundy MLE's in my WARP

Glossary: All numbers are adjusted for standard deviation. SFrac is the percentage of the season played (compared to a player with league average PA/G in 162 games). BWAA is batting wins above average, BRWA is baserunning wins above average, FWAA is fielding wins above average, Replc is wins above average a replacement player at the same position would have accumulated in the same playing time, and WARP is the first three minus the fourth (wins above replacement). 1919 is straight line adjusted to 162 games.

Year SFrac BWAA BRWA FWAA Replc WARP
1919 00.86 
+1.1 -0.1 +0.6 -2.50 +4.1
1920 00.88 
+0.9 +0.3 +0.8 -2.50 +4.5
1921 00.65 
+1.1 +0.2 +0.6 -1.90 +3.8
1922 00.85 
+1.0 +0.2 +0.9 -2.50 +4.5
1923 00.88 
-0.2 +0.2 +0.9 -2.60 +3.5
1924 00.97 
+2.3 +0.3 +0.9 -2.90 +6.4
1925 00.93 
+0.3 +0.2 +1.0 -2.80 +4.3
1926 00.92 
+1.5 +0.1 +0.7 -2.70 +5.0
1927 00.87 
+0.9 +0.0 +0.8 -2.70 +4.4
1928 00.92 
+1.4 +0.1 +0.8 -2.70 +5.0
1929 00.47 
+0.4 +0.0 +0.5 -1.50 +2.4
1930 00.90 
+1.3 +0.1 +0.7 -2.60 +4.7
1931 00.84 
+0.2 +0.1 +0.0 -2.50 +2.7
1932 00.87 
+2.3 +0.1 +0.0 -2.50 +4.8
1933 00.88 
+0.2 +0.1 -0.1 -2.40 +2.7
1934 00.63 
+1.4 +0.1 +0.0 -1.80 +3.3
1935 00.37 
+0.0 +0.0 +0.0 -1.10 +1.0
1937 00.10 
+0.0 +0.0 +0.0 -0.30 +0.3
TOTL 13.78 16.1 
+2.1 +9.0 -40.5 67.5 


3-year peak: 16.5
7-year prime: 35.1
Career: 67.5

Well, here's how it looks to me. These translations make Lundy look like an easy #1 for career voters--67.5 total WARP2 beats everyone else in the backlog by about 4. For peak and prime voters, he would seem to be rather lacking, sort of like an uber-Maranville. However, I wonder if this is a result of the MLE process flattening his peak. In particular, these numbers do NOT give him ANY seasons that would be good enough to win a Gold Glove in the majors, which does not jive with his reputation as one of the best fielders in NgL history. And they say he never had even one season over 120 OPS+, which we would expect just by random fluctuation if nothing else from a guy who played for nearly 20 years with a career OPS+ of 104. Was Lundy really the Beckley of shortstops?

The closest parallel to him--a defense-first, lowish-peak shortstop with 65 to 70 WARP2--seems to be Bobby Wallace (69.5). Wallace's best seasons by WARP2 were 7.2, 6.1, 5.9, 5.7, 5.6, 5.6, 5.4, while Lundy's are 6.4, 5.0, 5.0, 4.8, 4.7, 4.5, 4.5. So if we give Lundy 67.5/69.5 = 97% of Wallace's peak and prime, we get 6.9, 5.9, 5.7, 5.5, 5.4, 5.4, 5.2, making his 3-year peak 18.5 and his 7-year prime 40.0. That would still make him a low-peak candidate, but would give him a prime comparable to that of Campaneris and Rizzuto.

My understanding is that Lundy is very much considered in the pantheon of Negro League players--there's no "But why haven't we heard of him??" issue here as there is with Clarkson. Given that reputation, finding out that he was not even quite the black Bobby Wallace actually seems like something of a disappointment. That said, 97% of Bobby Wallace is clearly a deserving HoM'er. Unless there are further changes to these MLE's, making the above adjustments to his peak puts him 2nd to McGraw on my peak-heavy, rate-based salary estimator and an easy first on my consensus estimator. If there are no further changes to these MLE's, I will have him #2 on my 2006 ballot.

I know this is coming late, but check his thread--I have been asking for this information about Lundy for an extremely long time, and it never surfaced until now.
   103. Brent Posted: October 20, 2007 at 12:00 AM (#2584745)
Here's a link to the NeLg data compiled by the HoF study:

NLBatters.pdf.
   104. sunnyday2 Posted: October 20, 2007 at 01:20 AM (#2584784)
>For peak and prime voters, he would seem to be rather lacking, sort of like an uber-Maranville. However, I wonder if this is a result of the MLE process flattening his peak.

I think that is exactly right. It has always been true that NeL MLEs show odd, flat career curves considering the other data--career totals, OPS, etc. Look at Alejandro Oms, e.g. Same pattern. So, I'm a peak/prime voter, but I agree with Dan that the likelihood is that as a 104 OPS+ player, Lundy would clearly have had a somewhat higher peak than is shown here. Of course, I did vote for Bobby Wallace once upon a time because as a WS voter I recognize that WS does not do quite enough to reward defense (except at CF). So whatever weaknesses Lundy's peak seems to have, there is plenty of analysis to suggest those weaknesses are an artifact of our methods more than they are of his actual play.

I don't support Oms, however. Not because of his flat peak but because, contrary to Lundy's case, there are plenty of other guys with similar records to Oms.' In Lundy's case, the new MLEs would pretty clearly raise him from around 25th all-time to maybe 15th or better. 347 WS as a career total appears to be about 11th, assuming that Lloyd and Wells have more (and depending on how you treat Ward). Larkin I think is about 350. If you adjust the rest from 154 games to 162 then Wallace (345) clearly beats 347, Cronin and maybe Banks are right around 347, and if you give WWII credit to Pee Wee Reese, then I think he exceeds 347. Worst case, Lundy would be about 15th all-time, it seems to me.

With less than 347, no matter the adjustments, you've got Trammell, Boudreau, Jennings and Sewell, all HoMers. Sewell and Boudreau each have 277 WS (not adjusted). The only SS with more who are not HoMers are Aparicio, Fernandez, Campy and Maranville, none of whom had an OPS+ anywhere near 104. OK, not true--Fernandez at 101.
   105. Dr. Chaleeko Posted: October 20, 2007 at 03:14 AM (#2584843)
Sunny2,

let me take this one at a time:
So, Doc, from .299/.333/391/92 to .303/.360/.406/104...

There's a number of differences between Chris' method and mine. A biggie is that he uses a conversion on rates, I do it on runs. Chris discounts average at x% and slg at the square of x%. I use a single converter and run it through the Willie Davis comment method. I convert all NgL years through the mid 1940s at .90 on runs (actually RC). Chris' method probably reduces RC by more than I have chosen to. My SLG comes out 4% ahead of his, and my AVG 1% ahead in any event. The difference in OBP is a more robust 8%, most of which is explained by the 1% difference in average plus the new walk information from the HOF study. The ISOOBP for my MLE is .057, for his .034, so that's the biggest difference there.

In terms of looking at that difference, we previously were looking only at Lundy's 1928 walk rate. According to the HOF study he walked in 7.5% of PAs that year (actually those are PAs I have calculated with some other assumptions about HPB, but close enough). The HOF study shows him walking in about 8.3% of PAs during his career. Gary's data for 1928, which I think Chris used for his MLEs, shows him walking rather less often. And Chris' MLE has him walking in about 4.9% of career PAs. In addition, my own studies suggest to me that NgL players walked about 8-10% less often than MLBs of comparable ability. I do, therefore, build some walks back in when I convert them to MLB in an attempt to avoid situations where context illusions are causing us to think that NgL players in general just didn't walk much.

Because the HOF study contains much more walk data than just 1928, I feel fairly comfortable in suggesting that my walk rate projections are likely to be closer to the mark than Chris' because they are made with the scarcest of data---no one should take that as a criticism of Chris' work because that was all the data we had at that juncture. On the other hand, maybe I'm wrongheaded in my attempt to account for the BB-rate context differences between MLB and the NgLs.

And anyway, I'm known for being wrong about stuff.

But the walk rate explains the bulk of the difference. One other place where some difference can lie, and that could be puffing my numbers a bit, is that for NgL years, in absence of any leaguewide R/G or RC/G data, I made an estimate of the NgL's RC/G based on the HOF study's reported AVG and SLG numbers and a few studies I've run about BB and SH and SB rates as compared to MLB. I then compare this estimate to a neutralized RC/G average of 4.5 per game. Chris' technique takes the known NgL averages and discounts them without any league context (correct me if I'm wrong!) for the simple reason that at the time his method was developed we didn't have any reported leeague averages to work with. Where the puff may come into my figures is that I don't account for leaguewide K and GIDP rates which would tend to lower the RC/G estimate a little bit. I haven't estimated the size of this effect yet.

So there's some give and take in both Lundy's numbers and the methods used to interpret them.

And from 186 + 114 = 300 WS to 221 + 110 = 331. Or 347 adj to 162 games.

Again, my disclaimer note above in post #95 is important. The possibility exists that I'm getting slightly puffy BWS results, which could be pushing him upward. I think it's safer to express WS as a range: 320-350 (in 162 notation). And then there's the possible puff in the paragraph above this one. When you deal with known unknowns you have to make some assumptions to get from A to B to C, and the assumptions here could be inflating things. I don't really know because I don't have much to control against.

This is 67 percent bat and 33 percent glove.
Given Lundy's defensive and offensive reputations, i think this feels OK.

Those 347 WS compare to Vaughan 356, Cronin 333, Banks 332, Ozzie 325. I don't know exactly where Larkin ended up but I'm guessing about 350. All of these were < 30 percent glove. Oddly Ripken is a better parallel with 32 percent glove, but of course 427 WS.

Let's put this in 162 notation for all the mentioned players.
Vaughan 374
Larkin 359
Glasscock 350 (with UA deduction)
Cronin 350
Lundy 347
Banks 343
Wallace 342
Smith 331

And again, Lundy could be 320 or 347 in WS.

Anyway, Doc. Which is the real Dick Lundy? And was Lundy better than Ozzie Smith? Was he better than Tony Fernandez? Was he Barry Larkin?

Lundy was a guy who walked a bit above the league average (actually 10% more), hit for an average that was a little above the league (5% more), and hit for average power. The pitcher-removed NL batted .289/.347/.406 during 1919-1937. It walked in 7.4% of known PAs (per SBE), Lunday walked in 8.2% per my MLEs. On the whole, a little over average as a hitter, well above position. Defensively he was a stud for ten years and very good after. Here's some thoughts about specific analogs:

-Ray Durham: I think he MLEs as very slightly more effective at bat than Ray Durham but with a similar relative mix of skills. Durham has a little less on the AVG side and a little more ISO than Lundy. Lundy is a much better defensive player.
-Joe Sewell: A long-career Sewell with a little less average, better defender.
-Harvey Kuenn: with less AVG, much better glove.
-Jay Bell: with better AVG but slightly less ISO, better glove.
-Egardo Alfonzo: with a little more emphasis on SLG a little less on OBP, and a SS with a great glove.

The glove is clearly a big distinguisher here. But this is just based on my poking around bb-ref.
   106. Dr. Chaleeko Posted: October 20, 2007 at 03:36 AM (#2584859)
OK, wait. People are like changing ballots and now stumping for Lundy and stuff like no tomorrow. I just want to say that I'm not 100000% sure my system is exactly 100% right. As I've mentioned above, there are places where the numbers are probably puffy. Please don't take these as gospel, they are a good-faith attempt but nothing like law.

KJOK should probably make an appearance now. He's the one other person in the world who has actually seen the method up close, personal, warts and all, been inside the spreasheet. If he's cool with it, OK. If not, let's slow this train down.

There's lots of reasons to agree or disagree with my estimates (Brent has often done so, for instance with things I can neither prove nor disprove and he's MUCH smarter than I am about this stuff), and that's why I'm always really hesitant to get too attached to things. I mean it works for me, I generally use my own numbers, but it's not perfect. Look at the post of mine above this one before we go off half cocked. Or fully cocked!
   107. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 20, 2007 at 03:47 AM (#2584871)
Given Lundy's stellar reputation, seeing him as the black Bobby Wallace (which these numbers make him out to be, IMO) is if anything conservative.
   108. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 20, 2007 at 04:01 AM (#2584873)
Again, that's why I am jumping headfirst on the Lundy bandwagon--remember that I've been inquiring about him for months. It's because given his reputation, I expected him to be great. These MLE's just confirm that. Bus Clarkson's MLE's may be as valuable or more than Lundy's, but given how shaky MLE's are the lack of reputation is just too much for Clarkson to overcome to get on my ballot.
   109. sunnyday2 Posted: October 20, 2007 at 04:47 AM (#2584899)
I've had Lundy near-ballot for many years, so bumping him up is not that bigofa change. I mean, I've been tempted for years to ballot him based on his old MLEs. I mean, he was at 330 WS even then, with a 92 and an A/A+ glove. He always sounded like Ozzie Smith to me. And I like the type--I already had Pesky and Rizzuto on my ballot. But even with my low consensus score I guess I needed a little reinforcement before going with my gut. I don't think balloting him is half-cocked at all and if it is, well, somebody's gotta have the low consensus. We can't all be above average, can we?

And I agree with Dan. We (or at least I) have a fair degree of confidence in MLEs derived from the NeLs because we've had a lot of cases to help get it calibrated. MLEs based on the places Clarkson played carry a lot more uncertainty.
   110. KJOK Posted: October 20, 2007 at 07:34 AM (#2584957)
KJOK should probably make an appearance now. He's the one other person in the world who has actually seen the method up close, personal, warts and all, been inside the spreasheet. If he's cool with it, OK. If not, let's slow this train down.


I think the method is solid, except for maybe the Win Shares portion, which I don't fully understand.

What I'm worried about is the big bump in walks using HOF data. I would trust Gary's data over anyone's. Looking at Gary's data:

Lundy
1921
PA - 138
BB - 3

1927 (Cuba)
PA - 133
BB - 5

1928
PA - 229
BB - 10

It's possible walks could be missed for certain team games, but his teammates have higher walk %'s than he does, so I don't think it's a matter of Gary's data missing walks - the man was a hacker at the plate, and most of his OBP was Batting Average.
   111. sunnyday2 Posted: October 20, 2007 at 04:32 PM (#2584992)
Doc said the new walk rates come from HoF data and Cuba.... But ironically his "old" MLE OBA was .335 compared to Ozzie at .337 and Concepcion at .332. Now he's at .360. Well, what if he is at .335? Or split the difference to .347. We were under-rating him anyway.
   112. Dr. Chaleeko Posted: October 21, 2007 at 12:41 AM (#2585393)
Doc said the new walk rates come from HoF data and Cuba....

Just the NgLs, actually. No BB data for Cuba except for Gary's. I use the HOF data not because I have anything against Gary but as a matter of internal consistency. Since they are aligned with other data for other candidates, I'm a little hesitant to mix and match. That said, Gary's and the HOF guys' discrepancy is making me uncomfortable.

Well, I'm glad to see that KJOK indicates that the my MLE protocols are working. There's lots of moving parts, so human error is ALWAYS in play. Given that I'm using letter-grade-based defensive FWS borrowed from Chris Cobb's original estimates, the defensive information should be pretty solid, too. In which instance, Dan, I feel much better about your estimating his WARP from the stat lines I provide and the FWS I provide. Carry on then, and sorry for my little freakout!

I think the method is solid, except for maybe the Win Shares portion, which I don't fully understand.

OK, let me explain what I'm doing here. I've been through several iterations because I keep finding issues, but I'm getting really close. Let's just say, WS is a big pain in the buttocks when you're trying to create theoretically average teams and avoid distortion. I think I'm there or close though.

The big idea here is to put the players onto an otherwise exactly league-average offensive team. Turns out that if you just stick him on a team with average RC/G, it throws the other players out of whack. Instead of being average players, they gain or lose production proportionate to Lundy's contribution above or below the league average. If the RC/G is 4.50 and Lundy is at 5.0, he gets too many WS because the other players' RC/G drop below the league average in order to maintain exactly a 4.50 team.

So what I've done instead is to allow the team's RC/G to float based on Lundy's contribution. Returning to our 4.50 team, now all the other players retain their relative theoretical avreageness versus the league, but the team's RC/G can move to, say, 4.60 if Lundy is a 5.00 guy (just making up a number here). Or to 4.40 if he's a 4.00 guy.

In order to make this work, I've used positional RC/G averages and swapped Lundy in for an average SS. This way the average LF remains an average LF and Lundy's increase in production changes only relative to the league's SS, which has the happy effects of keeping all other players on Lundy's team frozen at their league average levels, allowing me to swap him in or out with precision (if he only plays a half year, I can figure his replacement's contribution and factor it in), and letting me figure out how many wins Lundy contributes beyond an average SS and use that to reassign WS.

The big issue (maybe not so big?) is that if Lundy much exceeds the average SS, he starts looking like he's the most valuable guy. That's the point of post #95, though that might not be such a bad thing. The big picture here is that on the otherwise average team in the league I'm putting Lundy into, his WS value is more accurately reflected than I've been able to do before. I think.... ; )

To show you how I do it, let me take you through Lundy's 1924 season, his best by MLE, for which I put him into the NL. Just so you know, I have MLE'ed Lundy at 96.7 RC and 409 outs in 145.8 games with 627 PA.

A) FIND THE OFFENSIVE CONTEXT OF AN AVERAGE 1924 NL TEAM AND AN AVERAGE NL SS
1) Find the average total RC of a theoretically average team in the 1924 NL.
A theoretically average NL team in 1924 created 730.6 runs (per research with the SBE).

2) Find the average RC/out of a theoretically average batter at Lundy's position.
Per the SBE, 1924 NL shortstops created .154 runs per out (I have a big chart with these rates year by year.)

3) Find how many runs a typical SS would have created in Lundy's outs.
Lundy made 409 outs, .154 * 409 = 63 RC (rounded).

B) FIND THE OFFENSIVE CONTEXT OF AN AVERAGE TEAM WITH LUNDY AT SS
4) Subtract (3) from (1) to determine how many runs the rest of the team created without Lundy or an average SS.
730.6 - 63 = 667.6 RC

5) Find team total with Lundy at SS instead of average SS by adding Lundy's RC to (4).
667.6 + 96.7 = 764.3 RC

C) FIND OUT THE WS CONTEXT FOR AN AVERAGE TEAM WITH LUNDY AT SS
6) Add together (1) and (5) to find out out how many total runs the team created and theoretically allowed.
764.3 + 730.6 = 1494.9 RC

7) Divide (5) by (6) to determine claim percentage for offense. Since the RA will always be the league average, it's not problematic to do this, I think. If the RA floated could move upward or downward, then it would be problematic.
764.3 / 1494.9 = .511

8) Figure the R/W for the league.
In the 1924, using pythagenpat, I have calculated the R/W at 9.7 R/W.

9) Subtract (1) from (5) to see how many extra runs Lundy added to a theorectially average team.
764.3 - 730.6 = 33.7

10) Divide (9) by (8) and add to 77 to figure how many wins a team with Lundy at SS would have.
(33.7 / 9.7) + 77 = 80.5

11) To determine the teamwide BWS, multiply (10) times 3 WS, and multiply that product by (7).
80.5 * 3 * .511 = 123.4 BWS

D) FIGURE MARGINAL RUNS FOR TEAM AND PLAYER AND ASSIGN BWS AND FWS
12) Figure team's marginal runs per the Win Shares way by multiplying (5) * .48.
764.3 * .48 = 366.9 marginal runs

13) Figure Lundy's background runs per the Win Shares way by multiplying the league's RC/outs * Lundy's outs and multiplying that product by .52.
409 * .181 * .52 = 38.5 runs

14) Figure Lundy's marginal runs by subtracting (13) from his RC.
96.7 - 38.5 = 58.2 marginal runs

15) Figure Lundy's batting claim points by dividing (12) by (14).
58.2 / 366.9 = .159

16) Figure Lundy's BWS by multiplying (15) times (11).
.159 * 123.4 = 19.6 BWS

17) Figure Lundy's estimated FWS/G
In 1924 Chris (and so I) have him as an A+ level fielder, and it turns out his rate is .056 FWS/G, which I copped from Chris IIRC.

18) Figure Lunday's estmiated games by dividing his PA by 4.3 PA/game.
627 / 4.3 = 145.8 games.

19) Find his FWS by mutiplying (17) by (18).
.056 * 145.8 = 8.2 FWS

20) Add BWS in (16) to FWS in (19) to get total 1924 MLE WS.
19.6 + 8.2 = 27.8 (this is .01 higher than above due to rounding errors)

21) To get 162 WS multiply (20) by 1.05.
27.8 * 1.05 = 29.2

That's "it." Lot of steps, but I think it's getting closer to "theoretcally correct" than I've been before.
-SFWS makes no accoutning of the league's offensive conditions and the value of runs relative to outs, and many of the MLB years in question were extreme offensive environments.
-Just sticking Lundy on an average team causes him to stand out more than he should by suppressing the offensive performance of the rest of the team.
-Allowing RC and team wins to float allows Lundy's performance to be seen in a more realistic context.
   113. Chris Cobb Posted: October 21, 2007 at 02:50 AM (#2585937)
Given that I'm using letter-grade-based defensive FWS borrowed from Chris Cobb's original estimates, the defensive information should be pretty solid, too.

Just for the record, "estimates" is too strong a word for the fielding values in the system. They are not derived from any stats, although Gary A's fielding data has provided a few checkpoints that suggest whether defensive reputations were deserved or not. What I tried to do with the fielding win shares was give a realistic view of what a player's fielding value would have looked like in win shares terms over the course of his career, if he was about as good as his reputation made him out to be. The fielding win shares give a career total that fits a letter grade that fits the reputation, with the win shares distributed seasonally to fit the typical peak and decline of a fielder of a given quality at the player's position.

So call them my "realistically modeled quantifications of fielding reputation," not "estimates" :-) .
   114. Paul Wendt Posted: October 21, 2007 at 04:26 AM (#2586085)
"realistically modeled quantifications of fielding reputation," not "estimates"

excellent!

Good night,
See you on Veterans Day
   115. Dr. Chaleeko Posted: October 21, 2007 at 04:57 PM (#2586271)
Thought I'd respond to Karl's 5-10% comment from the ballot thread here, since there's a great deal of methodological talk above and we're talking aobut Lundy anyway. Crossposted from the ballot thread for 2006.

63. karlmagnus Posted: October 19, 2007 at 08:58 PM (#2584774)
Even with EricC's new projectiomns, Lundy doesn't make it. 104 OPS+ for a shortstop is about 129 for an outfielder, and that with 2400 hits wouldn't sniff the ballot. In any case, Eric's projections are usually 5-10% high and too long, compared with say Chris Cobb's, on which we elected most of the NgLs.


I will say this in my own defense. I think the WS are possibly puffy. But I think the MLE stat lines are not as much. I know I said otherwise recently, but I realized I had it backwards.

I claimed previously that because in some cases I had to estiamte RC/G for a league based on fewer, not complete, components there might be puffing as a result. In reality, this isn't true.

I take the est RC/G and compare to a 4.50 R/G league. But
a) I make adjustments along the way which specifically account for the difference between using RC/G and R/G as the conversion rate (because I sometimes must use both depending on the available data)
b) If the est. lg RC/G is a little higher than the actual R/G, that means the player is drawn further toward 4.50 than he should be.

To give an example. Let's say we have a player who created 10 R/G. Further say that the estimated RC/G for his league is 5.00. That means that 4.50 / 5.00 * 10.00 = 9.00 RC/G in the neutral environment. Now let's say that we found out later that the player's league was actually 4.95 R/G. Now 4.50 / 4.95 * 10 = 9.09 RC/G: 1% higher than before.

Flipping it around player is 10.00 again, but league is 4.00...
4.50 / 4.00 * 10 = 11.25 RC/G

But if we later find out the league was a 3.95 R/G league, then:
4.50 / 3.95 * 10 = 11.39 RC/G, or 1% higher.

Since having fewer components nearly always leads to a little puff in the lgRC versus actual RS, I think the MLE conversion ends up coming in slightly conservative in this regard.

Of course if you don'thave park factors that leads to some variability too, but that variability works in both directions, often within the same career.

So WS, possibly 5% puffy. Stat lines, I don't think so.
   116. Chris Fluit Posted: October 21, 2007 at 08:56 PM (#2586520)
The short hand is this: Lundy is possibly as good as Sewell and Trammell- two guys who are already elected into the Hall of Merit- and at least as good as Rizzuto and Concepcion- two guys who are picking up 10-15 votes per year. It's understandable if you're skeptical about the latest revision that has him up there with Sewell and Trammell, but even after a 5 to 10% discount Lundy is still a strong candidate for consideration.
   117. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 21, 2007 at 08:59 PM (#2586525)
Unfortunately, it's too late. Lundy's not going in this election--he was "rediscovered" after many ballots were already in--and we have Ripken, Gwynn, and McGwire coming up next year. What a shame.
   118. karlmagnus Posted: October 21, 2007 at 09:08 PM (#2586532)
If his OPS+ is 104 he's not as good as either Sewell or Trammell, and both are bottom-quartile HOMers, especailly Sewell who took decades to be elected.Dr C says he's happy about the solidity of his OPS numbers but not his Win Shares numbers. Even for NgL optimists, even if you take Dr C's OPS numbers as rock-solid, I think Lundy misses the cut. Stephens for example is clearly substantially better, by 15 points in OPS+.
   119. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 21, 2007 at 09:19 PM (#2586546)
Oh come on. Stephens only played for a decade; Lundy played for two. Lundy was renowned as the best defensive shortstop in NgL history if I'm not mistaken; Stephens was merely adequate. And three of Stephens's nine full seasons were against wartime competition.
   120. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 21, 2007 at 09:25 PM (#2586553)
karlmagnus, have you ever given any thought to Dave Orr? His lifetime OPS+ is a cool 161.
   121. karlmagnus Posted: October 21, 2007 at 11:57 PM (#2586711)
I've given considerable thought to Orr, but it's a very short career. Levi Meyerle is in my PHOM, however. His 1871 knocks Cash's 1961 into a cocked hat and it wasn't SLG-heavy like McGwire's.
   122. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 22, 2007 at 12:04 AM (#2586723)
You're right. You are nothing if not consistent.
   123. DL from MN Posted: October 22, 2007 at 12:49 AM (#2586948)
Updating using the new translations and DanR's work (can you give me your other NgL Candidates - Ben Taylor, Bus Clarkson, Alejandro Oms, Luke Easter as a sanity check on my swags?) Dick Lundy moves up from 60th to 10th based on changing his OPS+ translation from 94 to the current levels. He's still behind where I have Clarkson. I'm using Jay Bell as Clarkson's defensive comp.

When I asked Holway who we overlooked he immediately jumped to Dick Lundy so the reputation is clearly in Lundy's favor. I'm not quite as impressed with Lundy as Trammell or Ozzie but I'm liking him better than Willie Randolph.
   124. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 22, 2007 at 12:53 AM (#2586971)
Taylor doesn't have seasonal MLE's posted. I'll do the others as soon as I can.
   125. Gary A Posted: October 22, 2007 at 01:52 PM (#2588604)
A few things about Lundy:

1) As far as I can tell from reading contemporary newspapers (my subjective assessment of their subjective assessments), observers were more impressed by Lundy's defense at short than anyone else's. Reputation-wise, only Lloyd approached him in NgL history (and it was to accommodate Lundy that Lloyd finally moved to second base in 1924).

2) I'm currently rebuilding 1921 from the bottom up (though I haven't gotten to Bacharachs' games yet), so I might soon have a better explanation of the discrepancies between my numbers and the HOF's. I suspect the differences have something to do with the fact that Atlantic City box scores frequently lacked summary sections, meaning there were often no walk numbers. (Unfortunately, I won't be able to return to 1928 for a while.)

3) It's possible that in coming months I could have some limited stats for Lundy's 1922 season--his Original Bacharach Giants did in fact play a few games against Hilldale, the Baltimore Black Sox, the Richmond Giants, and other top-level teams, plus Lundy got in at least one game with the NY Bacharachs vs. Baltimore in October.
   126. DL from MN Posted: October 22, 2007 at 03:51 PM (#2588765)
When you rebuild 1921 can you "infer" a walk from Lundy having fewer AB than you would expect in a game?
   127. Gary A Posted: October 22, 2007 at 05:07 PM (#2588877)
Yeah, I already do that (you also have to make allowances for the possibility of sac hits and hit batsmen). Btw, there's also trouble with games played in Philadelphia (where the Bacharachs played quite a bit)-- Philadelphia box scores never carried at bats, and many didn't have summary sections, either.
   128. Mike Green Posted: October 22, 2007 at 05:17 PM (#2588893)
I suspect, Gary A, that an HBP is as good as a walk, for these purposes. That probably makes it a bit easier.
   129. Gary A Posted: October 22, 2007 at 09:29 PM (#2589442)
OK, mea culpa time. Bottom line is that I’ve revised Lundy’s walks from 10 to 15 for 1928, however much that matters. Here’s why:

I decided to go back over Lundy’s 1921 and 1928 seasons to see if there was anything I missed. For 1928, it turns out that I did NOT always do what DL from MN was referring to above; that is, infer walks (or other “non-AB events”) for obvious non-AB plate appearances when there are no walk totals given in the box score (that is, if Lundy, batting fourth, has 3 at bats but everyone else in the lineup has 4, you have at least one “non-AB event” such as a walk, HBP, or SH for Lundy--even if there's no summary section). It was actually quite easy to find, since my method of recording games includes recording the number of known plate appearances. Subtracting at bats, walks, hit batsmen, and sac hits from total plate appearances reveals whether any of the non-AB plate appearances have not been “assigned,” as it were, to a specific non-AB event.

In eastern Negro League games in 1928, I found 114 such “unassigned” plate appearances, mostly over 21 Bacharach Giants’ home games. These are 114 plate appearances out of 12,722 for all eastern games in 1928, less than one percent; but they do affect Bacharachs’ players disproportionately. More than half of them (65 of 114) belong to Bacharachs’ hitters.

If you “assign” non-AB events to the 114 unassigned plate appearances proportionate to those events’ occurrence in other games, you end up adding about 82 walks, 8 hit batsmen, and 24 sac hits to eastern “league” totals. This actually raises the eastern “league” OBP from .335 to .340, and the Bacharachs’ team OBP from .342 to .357. (It doesn’t have any effect on batting average or slugging pct, of course.)

Eight unassigned plate appearances belong to Lundy (George Carr has the most, with nine; Fats Jenkins has seven). If you assign them proportionately, you add about five walks, one hit by pitch, one sac fly, and one sac hit to Lundy’s totals. It raises his OBP from .368 to .382. This is how the comparison with the HOF numbers looks now (my corrected numbers first / HOF second):

G: 58 (team 62) / 53
AB: 217 / 195
H: 73 / 68
D: 13 / 12
T: 3 / 3
HR: 4 / 3
R: 36 / 34
BB: 15 / 16
HP: 3 / --
SF: 3 / --
SH: 2 / 1
SB: 8 / 8
AVE: .336 / .349
OBA: .382 / ---
SLG: .479 / .487

Revised eastern averages are: .283/.340/.387 (I have eight box scores for the Bacharachs playing against NNL teams in the west, which aren’t included in the eastern totals.)

(The HOF study may not count the two games Lundy played with the “Eastern League Stars,” which was basically the Bacharachs with a couple of Hilldale players thrown in, against the Lincoln Giants. Lundy hit 1 for 7 with a sac fly, the one hit being a home run.)

I think the reason I’d failed to deal with these non-AB events was that I was intending to assign them statistically (as I’ve just done), en masse after compiling the numbers, rather than assign them game by game. I just forgot to do that step (or else I did it for the HOF, and then misplaced the file). This problem doesn’t occur in any other year I’ve done that involves Bacharachs’ home games, because my methods now involve always accounting for every plate appearance at the game level.

So I checked out 1921, and there is no such problem there. And it’s completely irrelevant to any Cuban statistics, including 1927/28, as those are all based on good box scores (and in 27/28 play-by-play accounts were published for many of the games).
   130. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 22, 2007 at 09:35 PM (#2589454)
Gary, how could you possibly have gotten Lundy's walks wrong by 5 in 1928. That's just completely unforgivable; a crime against humanity, really. You should hang your head in shame for all eternity.
   131. sunnyday2 Posted: October 22, 2007 at 10:09 PM (#2589534)
Well, he didn't know we were BB obsessive ;-)
   132. Chris Fluit Posted: October 22, 2007 at 10:40 PM (#2589606)
Thanks for the extra work, Gary. It certainly explains some of the differences we've been trying to account for.

And thanks for the laugh, Dan.
   133. DL from MN Posted: October 22, 2007 at 11:14 PM (#2589652)
Does the change to the league OBP affect the MLEs as to revise Lundy downward?
   134. Dr. Chaleeko Posted: October 23, 2007 at 02:23 AM (#2589799)
Does the change to the league OBP affect the MLEs as to revise Lundy downward?

The MLEs I recently posted remain as-is. I was using the HOF data and its league averages. To reiterate, this is nothing against Gary's work at all, just a matter of trying to keep apples together with apples as much as possible.
   135. Paul Wendt Posted: October 23, 2007 at 01:53 PM (#2590056)
This problem doesn’t occur in any other year I’ve done that involves Bacharachs’ home games, because my methods now involve always accounting for every plate appearance at the game level.

Hi, Gary
I understand the 1928 method.

Using the current game-level method, do you make assignments of non-AB events that depend on assignment history?
player A has 2 assigned BB and 0 Sac so far this season, so he gets the Sac
player B has 1 BB and 0 Sac, so he gets another BB . . .

I suppose box scores that do report AB by player generate three subclasses classes of non-AB events
- counts reported by batter
- counts by team (or opposing pitchers)
- no counts reported
And then you have some box scores that do not report AB by player, but some of them do report non-AB events by team or by batter.
Right?
   136. Gary A Posted: October 23, 2007 at 03:12 PM (#2590156)
Using the current game-level method, do you make assignments of non-AB events that depend on assignment history?
player A has 2 assigned BB and 0 Sac so far this season, so he gets the Sac
player B has 1 BB and 0 Sac, so he gets another BB . . .


For the very specific situation of a box score with ABs but absolutely no summary data on non-ABs (which is almost entirely confined to Atlantic City games--and not all of those), I assign each non-AB fractionally, using a general ratio of BB to SH based on all the NeL seasons I've done. I think it's 4 to 1 (though I haven't done it for a while and my notes are at home), meaning each non-AB counts as .8 walk, .2 SH (I ignore HBP, which, as Mike Green points out above, are functionally equivalent to walks anyway). Obviously, when the game account gives some scrap of information ("in the seventh Lundy laid down a bunt to put Jenkins in scoring position..."), you adjust for that. When compiling the season totals I round the fractions off.

I suppose box scores that do report AB by player generate three subclasses classes of non-AB events
- counts reported by batter
- counts by team (or opposing pitchers)
- no counts reported
And then you have some box scores that do not report AB by player, but some of them do report non-AB events by team or by batter.
Right?


Yes. You can find just about every possible combination of present and missing data. There are many Philadelphia box scores in the 1920s that have every everything you could ask for--EXCEPT at bats. In the 1930s the Philadelphia Tribune started printing the crudest box scores imaginable--just "r h e," no summary section, nothing else. It's like they stepped out of the 1850s. Luckily the Afro-American, Courier, and other papers often picked up the games with better box scores.

Nobody should think, though, that box scores like these represent the majority of Negro League data. At least in the 1920s, most box scores (60% or so) are excellent, with HBP, SH, double play details, fielding stats, even PB & WP. Another 25% or so have maybe a couple of deficiencies in the summary categories (HBP and SH are the most frequently missing). Only the remaining 15% or so have serious deficiencies like missing ABs, or walk & K totals, or whatever. And probably the vast majority of these are in Philadelphia or Atlantic City. Chicago box scores in 1920 & 1922 are mostly missing ABs, SHs, double plays, and HBP--but 1921 & the rest of the decade has ABs, and is much better overall; Pittsburgh box scores in 1921-22 are missing ABs. Box scores everywhere else--KC, St. Louis, Baltimore, NY, Detroit, Indianapolis, Cleveland, Washington, etc.--are basically good. And we do have the Hilldale scorebook for much of 1926.

Meanwhile, Cuba is a different situation altogether, with great box scores (and often PbP accounts) for nearly all Havana games.
   137. Chris Cobb Posted: October 24, 2007 at 02:09 AM (#2590994)
For the record, let me note that I have rerun Lundy's MLEs in my conversion system using the HoF numbers, except for 1918, 1919, and 1922. I get the following career line:

Dick Lundy
1918-32 8501 PA, 643 BB, 2343 hits, 2949 totabl bases
.298 BA, .351 OBP, .375 SLG, 93 OPS+

I can provide season-by-season MLEs if they are desired.

It looks to me like he compares very closely to Phil Rizzuto in terms of career length (if Rizzuto receives war credit) and offensive value.
   138. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 24, 2007 at 03:06 AM (#2591025)
say Wha?? Dr. Chaleeko says he's a 104 OPS+. You say 93. Can the two of you please settle this before the 2008 election? Your friendly shortstop supporters would appreciate it.
   139. Chris Cobb Posted: October 24, 2007 at 03:43 AM (#2591042)
Dr. Chaleeko and I are unlikely to reconcile our results, which rest on some different assumptions about how MLEs ought to be constructed. However, I have re-run my numbers, applying the slugging adjustment only to isolated power, and not to slugging average as a whole, a change that treats players whose SA is mostly BA more fairly. This gets results that split the difference between my original estimate above and tthe good Doctor's:

Dick Lundy
1918-32 8501 PA, 643 BB, 2343 hits, 3115 total bases
.298 BA, .351 OBP, .396 SLG, 99 OPS+
   140. Chris Cobb Posted: October 24, 2007 at 03:47 AM (#2591043)
My .298, .351, .396 is rather congruent to Dr. Chaleeko's .303, .360, .406.

It's clear when I make the switch to ISO that we're profiling the same player, but Dr. Chaleeko's estimates are just a bit higher across the board.
   141. TomH Posted: October 25, 2007 at 01:17 PM (#2592893)
a reminder: OPS+ is a terrible tool to compare Lundy to Ozzie Smith. Ozzie is the ultimate underrated-by-OPS+ player, given his high OBP, great SB/CS & baserunning. Ozzie has a lifetime OPS+ of 87 and a OWP from bb-ref of .481. Dave Kingman had years with that OWP range when his OPS+ was 100.
   142. TomH Posted: October 25, 2007 at 01:18 PM (#2592894)
...not that he needs to be Ozzie to get votes ....
   143. Dr. Chaleeko Posted: October 25, 2007 at 11:07 PM (#2593664)
My .298, .351, .396 is rather congruent to Dr. Chaleeko's .303, .360, .406.

It's clear when I make the switch to ISO that we're profiling the same player, but Dr. Chaleeko's estimates are just a bit higher across the board.


Also, there are potentially differences about which leagues we're teleporting Lundy into: Chris may be using alternating leagues (AL then NL), while I use just NL for Lundy. Therefore there may be differences in the actual stats we each crank out.

I think the effective conversion rate for mine are probably a little higher than Chris', which obviously makes a difference.

One of us could be right. Neither of us could be right. I prefer to think we're both right.
   144. DL from MN Posted: October 26, 2007 at 01:29 AM (#2593938)
99 OPS+ and solid defense looks a lot like contemporary Dave Bancroft.
   145. David Concepcion de la Desviacion Estandar (Dan R) Posted: October 26, 2007 at 01:48 AM (#2594011)
Take the average of Chris and Chaleeko and you get 101.5 OPS+, in a longer career than Bancroft's. I like Bancroft plenty, always have, but I like Lundy more.
   146. DL from MN Posted: October 26, 2007 at 02:40 AM (#2594227)
I agree - a little bit better than Dave Bancroft.
   147. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 13, 2007 at 05:25 PM (#2613388)
I'd be interested to hear what the following voters think of Dick Lundy. Have you examined his new MLE's based on the HoF data? Are you aware that his reputation was so strong that one Negro League historian (can't remember if it was Gary A, Holway, or someone else) was stunned to hear he was left out of the HoM? Would you vote for Bobby Wallace if he were eligible today?

John Murphy
Sean Gilman
Rob Wood
Eric Chalek
Howie Menckel
Andrew M
DanG
Mike Webber
Jim Sp
jimd
KJOK

Cross-posting to Lundy's thread.
   148. DavidFoss Posted: November 13, 2007 at 06:52 PM (#2613499)
I'd be interested to hear what the following voters think of Dick Lundy. Have you examined his new MLE's based on the HoF data? Are you aware that his reputation was so strong that one Negro League historian (can't remember if it was Gary A, Holway, or someone else) was stunned to hear he was left out of the HoM? Would you vote for Bobby Wallace if he were eligible today?

Why are you only asking a subset of the electorate? :-)

Bobby Wallace had some peak seasons mixed in and a long decline phase that hurt his career rates. That depressed career rate still beats Lundy's by a point, though. I would still rank Wallace over Lundy.

Its no surprise that we left Lundy out with the previous translations, though. Dock 10 points of OPS+ off of any candidate and it will hurt him. Sorry if I'm seeing this late, but how were the guys we inducted (Moore, Wells, Beckwith) affected by the new HOF data?
   149. DL from MN Posted: November 13, 2007 at 07:16 PM (#2613521)
Here's the quote from Holway

"Clarkson is a nobody. Lundy is one of the Big Three shortstops. Hit 30
points higher than Wells. Took (an older) Lloyd's job away from him. Won
three flags as a manager. Considered the smoothest fielding shortstop. A
travesty that he didnt make the Hall of Fame. I'd have named him in the
first five on the latest [1996] ballot.

Pop Foul"
   150. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 13, 2007 at 07:22 PM (#2613531)
I'm asking the subset whose ballots seem to me to show a) some minimal understanding that playing first base and playing shortstop are not the same thing b) a similar embracing of the concept that catching the ball is part of the game c) a willingness to look past pure peak value and d) didn't mention Lundy on their 2007 ballots.

As you are obviously aware, (one) problem with MLE's is that they flatten peaks. We don't know what Lundy's Best Years would have looked like in the bigs. What we do know is that for him to have accumualted the career value he projects to without having had a peak at least as high as Wallace's, he would have to have been a historic outlier, the Jake Beckley of shortstops. That *could* be the case, but it is not the most likely one. I chose Wallace as his comp because he was a long-career, low-peak, defense-first, prewar shortstop whose career value is very similar to Lundy's MLE's. I too have Wallace ahead of Lundy, but by a hair.

Lundy was not really a viable candidate until the new data came along, which put him as a very strong HoM'er, welll above the backlog. I don't know if anyone else's MLE's moved nearly as much--my understanding is that the old MLE's had Lundy with extremely low walk rates because walks were not recorded for some of the teams he played on. So if any of those guys played on the same teams as Lundy, or on other teams with missing data, then we'd expect their MLE's to move as well, although it's just as likely (leaving Lundy's teams aside) that they'd go down as it is that they'd go up.
   151. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 13, 2007 at 07:23 PM (#2613532)
Thanks very much, DL from MN!! That's tremendously helpful.
   152. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 13, 2007 at 08:12 PM (#2613585)
DavidFoss, I mention Wallace as a comp, but, of course, Wallace was elected long ago. What I really want to know is how you see Lundy compared to the backlog, in light of the new MLE's and reputational evidence.
   153. DavidFoss Posted: November 13, 2007 at 08:47 PM (#2613643)
Well, Lundy certainly looks better with a career OPS+ of 104 than he does with a career OPS+ of 94, that's for sure. If it turns out that not inducting Lundy was a mistake, than not having complete information would be one of the biggest reason why he was underrated during the time him and his contemporaries were being considered.

Looking at the 1943 balloting and discussions. Wallace had been long inducted by then. It looks like voters were struggling whether to place him with Sewell (high backlog, eventually inducted) or with Bancroft & Maranville (glove-only, still uninducted). Ten points of OPS+ would certainly have helped him at the time. I'll take another look at him.

You are certainly right about the peak-smoothing with the MLE's. That's something that's worth reminding people of as its one of the things I forgot about when looking back '50 years' later.
   154. Dr. Chaleeko Posted: November 13, 2007 at 09:03 PM (#2613664)
Well, Lundy certainly looks better with a career OPS+ of 104 than he does with a career OPS+ of 94

I just want to caution that as you can see upthread, it's possible that I'm coming in a little bit over. I don't think I am, but it's possible. If you think I am coming in over, then you should make an appropriate allowance.
   155. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 13, 2007 at 09:59 PM (#2613747)
Is it *more* likely you're coming in high on Lundy than with any other MLE? If you have missed on him, is it more likely you've over- rather than under-estimated him?
   156. Dr. Chaleeko Posted: November 13, 2007 at 10:09 PM (#2613765)
Is it *more* likely you're coming in high on Lundy than with any other MLE? If you have missed on him, is it more likely you've over- rather than under-estimated him?

Not to my knowledge. It's possible that I generally come in a little over. Though I'm not 100% sure of that either. But in Lundy's specific case, I don't have any reason to believe that he's more over than anyone else might be.

I've laid out my MLE process in headache-producing detail on other threads, so all voters are encouraged to check it out and determine for themselves if they think there's any place where my numbers may be puffing things up.

Karlmagnus thinks so, though I'm not sure he's ever given a convincing argument why. Brent's argument is the most convincing, namely that my guesses at league-quality adjustments aren't always empirically driven...they can't be without a level of research I'm not able to commit to (time, data needs, would need to learn how to use dbs effectively, etc). That's why I backpeddle on their absolute accuracy.
   157. DavidFoss Posted: November 13, 2007 at 10:17 PM (#2613769)
Is it *more* likely you're coming in high on Lundy than with any other MLE?

The issue here is that it changed. In post #51 (April 2005), it was 94 (92 including the decline phase). In post #97 (Oct 2007), it was 104 (including the decline phase).

300 WS estimate in Jan of 2005 and 347 WS estimate in Oct of 2007.

We leaned on these MLE's heavily in the 40s and 50s balloting as we tried to rank the Lundy's, Wells's, Sewell's and Maranville's on the same ballots. Oral traditions were hard to sift through as every candidate had superlative anecdotes associated with them.

Is this bump in Lundy's MLE's due to new raw data that was found since then? Or a change in the MLE algorithm? How does the new raw data and/or new algorithm affect the evaluation of Lundy's contemporaries?
   158. DavidFoss Posted: November 13, 2007 at 10:24 PM (#2613778)
Karlmagnus thinks so, though I'm not sure he's ever given a convincing argument why. Brent's argument is the most convincing, namely that my guesses at league-quality adjustments aren't always empirically driven...they can't be without a level of research I'm not able to commit to (time, data needs, would need to learn how to use dbs effectively, etc). That's why I backpeddle on their absolute accuracy.

Oh, I'm not questioning their exact accuracy. I'm just curious about how the new 2007 MLE's are quite a bit different than the older 2005 MLE's that we used for all the other NeL candidates. Of course they won't be perfect, but the ones we used back then at least formed a self-consistent group.

It could be that we've got better data now and that research is still actively going on and that we'll have even better data in the future. In that regard, the NeL constituency of the HOM is very much a "2004-2005 snapshot" of the best data we had available at the time.
   159. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 13, 2007 at 10:27 PM (#2613782)
The change in MLE's is definitely due to the new raw data. I hope Dr. Chaleeko can tell you whether it changes his evaluation of Wells, Moore, or Beckwith.\

It could be that we've got better data now and that research is still actively going on and that we'll have even better data in the future. In that regard, the NeL constituency of the HOM is very much a "2004-2005 snapshot" of the best data we had available at the time.


My understanding is that this is indisputably the case.
   160. Dizzypaco Posted: November 13, 2007 at 10:27 PM (#2613783)
Brent's argument is the most convincing, namely that my guesses at league-quality adjustments aren't always empirically driven...they can't be without a level of research I'm not able to commit to (time, data needs, would need to learn how to use dbs effectively, etc). That's why I backpeddle on their absolute accuracy.

I've said this before, but I believe league quality adjustments are critical, yet virtually impossible with any real degree of accuracy. We're talking about comparing the quality of two leagues, in which players of the two leagues never played against each other in any truly meaningful game. I personally don't think its possible to create an empirically driven league quality adjustment, regardless of the time, effort, skills, or anything else, so you ending up taking a guess. Its an educated guess, so its useful, but its still just a guess. And given the high level of uncertainty, reputation matters.
   161. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 13, 2007 at 11:55 PM (#2613871)
Dr. Chaleeko, I suppose the real question is, even if you don't trust Holway, do you trust your own MLE's? And if you do, why isn't Lundy on your ballot? :)
   162. ronw Posted: November 14, 2007 at 12:40 AM (#2613901)
Is Lundy the right candidate? I realize Dan R likes shortstops, but is he focusing on Lundy because he has MLE's. I have the following data for Negro Leaguers, taken from the HOF website:

** means that they have been elected to the HOM
First is the first year of data (some players played before the first year)
Last is the last year of data (players may have played past this year)
A1 is the age at First data
A2 is the age at Last data

Player...............First..Last.....G....AB.....R.....H...2B...3B...HR..RBI...BB...AVG...OBP...SLG....OPS..A1..A2
**RobinsonJackie....1945..1945....14....53....11....23....7....1....1...15....8...434...508...660..1.169..26..26
**GibsonJosh........1930..1946...510..1855...467...666..109...41..115..432..255...359...435...648..1.083..18..35
**StearnesTurkey....1920..1940...917..3514...713..1211..203..105..183..718..388...345...405...618..1.023..19..39
**CharlestonOscar...1920..1941...814..3027...697..1054..185...61..128..502..408...348...417...576....993..23..44
**BeckwithJohn......1920..1935...452..1647...331...575..103...29...77..302..148...349...396...587....983..20..36
**IrvinMonte........1938..1948...159...587...125...210...34....9...23..146...57...358...408...564....972..19..29
**BrownWillard......1935..1958...475..1806...373...634..111...47...67..287..109...351...387...576....963..19..42
**SuttlesMule.......1923..1944...762..2727...561...893..167...50..133..493..257...327...380...572....951..22..44
**LeonardBuck.......1934..1948...412..1472...352...471...73...26...60..275..257...320...413...527....940..26..40
**TorrienteC........1920..1932...581..2040...368...691..132...47...46..309..283...339...409...517....926..26..35
**WilsonJud.........1922..1945...850..2954...602..1038..182...33...71..485..354...351...415...507....923..28..51
**RoganBullet.......1920..1938...672..2022...363...684..104...59...45..251..243...338...400...515....915..30..48
Scales
George........1921..1946...666..2215...423...710..154...32...66..363..303...321...395...508....904..19..43
**WellsWillie.......1924..1948...756..2879...644...918..171...43...98..399..347...319...385...510....896..15..39
**MooreDobie........1920..1926...438..1767...296...611..108...49...32..201..109...346...375...517....892..25..31
**DihigoMartin......1923..1945...379..1404...292...431...61...17...64..227..143...307...366...511....877..18..40
**DobyLarry.........1942..1947....88...329....62...101...14....9....9...64...48...307...391...486....877..19..24
**CampanellaRoy.....1937..1945...137...466....86...151...25....8...14..107...39...324...372...502....874..15..23
Dixon
Rap............1922..1937...478..1790...398...564...90...37...51..246..199...315...373...492....865..21..34
**OmsAlejandro......1923..1935...275..1042...202...331...60...11...31..173..119...318...377...486....862..28..39
Parnell
Roy..........1927..1943...529..1982...398...646..109...41...31..224..166...326...373...469....842..21..37
**MackeyBiz.........1920..1947...863..3120...503..1028..143...51...59..433..270...329...374...465....839..22..49
**LloydJohn.........1920..1932...444..1692...304...580...86...19...19..162..140...343...384...450....833..36..48
**SantopLouis.......1920..1926...190...584...100...189...29....6...13...87...42...324...361...461....822..30..36
Taylor
Ben...........1920..1929...467..1695...226...545...83...29...21..191..173...322...375...442....817..31..40
**HillPete..........1920..1925...129...404....54...119...18....8....7...40...66...295...386...431....817..39..43
**MinosoMinnie......1946..1948....74...298....52....89...19....8....4...40...23...299...348...456....804..23..25
**BellCool Papa.....1922..1946...865..3444...736..1092..158...53...36..237..339...317...374...425....799..19..41
Jenkins
Fats.........1924..1940...484..1951...432...633...70...13...15..119..207...324...385...397....781..26..42
Hughes
Sammy T.......1930..1946...316..1184...233...357...65...20...16..113..108...302...350...431....781..19..35
Lundy
Dick...........1920..1937...614..2291...409...700...90...26...37..297..204...306...357...416....773..21..38
Dandridge
Ray........1933..1944...186...710...105...224...29...11....3...53...38...315...349...400....749..19..30
Day
Leon.............1934..1946...148...425....66...123...19....9....4...59...29...289...331...405....736..17..30
O
'Neil, Buck..........1937..1954...387..1285...177...386...53...19...13..160...64...300...331...402....732..25..42
Johnson, Judy.........1921..1936...784..2983...476...875..126...48...27..294..159...293...324...395....719..20..35
Allen, Newt...........1923..1947...914..3538...561..1017..160...47...18..204..312...287...337...375....712..19..43
Marcelle, Oliver......1920..1931...443..1674...269...491...48...18....9..144..163...293...345...360....704..24..35
Smith, Hilton.........1932..1948...140...282....35....83...19....2....2...38....7...294...304...397....701..25..41
**Mays, Willie........1948..1948....18....61.....9....16....2....0....0...12...12...262...384...295....679..17..17
Poles, Spot...........1923..1923....27....99....16....25....1....1....0...10....2...253...262...283....545..35..35
**Trouppe, Quincy......N/A...N/A...N/A...N/A...N/A...N/A..N/A..N/A..N/A..N/A..N/A...N/A...N/A...N/A....N/A.N/A.N/A 


This is raw data, and encompasses only Negro Leagues, but a couple of things jump out:

1. George Scales sticks out as a guy who hit like a HOMer. Are we somehow attributing too many of his St. Louis years (that also affect Wells and Suttles, among others) to him? Scales also played all of the infield positions, but doesn't appear to have the reputation as a terrific fielder.

2. Bullet Rogan sure hit well. karlmagnus should like him better than Caruthers as a pitch/hit HOM combo.

3. Remember that at the bottom, Oms, Lloyd, Santop and Hill all had significant pre-1920's careers. (So did Torriente and Charleston, but they look like easy HOMers even without their pre-1920 time).

4. Lundy fits right in with the rates of Sammy T. Hughes and Ray Dandridge, guys who we don't have in the HOM. Now Lundy played shortstop and was an excellent fielder, but then again, Dandridge was pretty well-known for his fielding at third.

5. Josh Gibson was just awesome.

6. This may be the only list you will ever see with Willie Mays at the bottom.
   163. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 14, 2007 at 01:11 AM (#2613927)
Thanks very much for posting that, ronw.

1. Were seasonal MLE's ever done for Gibson? He's the one that always seemed to me to have a place in the greatest-player-ever discussion.

2. If you have two players with the same hitting, one of whom played SS and one of whom played 2B or 3B, the difference in their value is absolutely tremendous. Would we ever elect a 3B with Sewell's offense? Of course not. The fallacy of just calling them "gloves" is, in my opinion, one of the main reasons that led the group to err on Nellie Fox (the other being that he didn't have rival 2B candidates to split his vote). If Dandridge and Lundy were similar as hitters, that seems entirely consistent to me with Lundy being a strong selection and Dandridge a forgotten backlogger.

3. I'm focusing on Lundy because he has the MLE's-reputation combination that I like to see in a NgL candidate.

4. I must confess I had literally never heard of George Scales. Skimming his thread, it does seem like a 130 park factor might have something to do with the group deciding to pass him over. But it doesn't look like MLE's were done using the HoF data, although they were indeed requested.
   164. sunnyday2 Posted: November 14, 2007 at 01:26 AM (#2613939)
This data also reinforces the notion that electing Oms and Pete Hill and maybe Cool Papa (unless he was Tris Speaker in CF, which he might have been) to the HoM were mistakes.
   165. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 14, 2007 at 01:30 AM (#2613944)
Well, that's the very end of Hill's career--no fair. And if I'm not mistaken a lot of Oms' best work was done in Cuba, which I don't think would be included in these numbers.
   166. Chris Fluit Posted: November 14, 2007 at 01:45 AM (#2613951)
3. Remember that at the bottom, Oms, Lloyd, Santop and Hill all had significant pre-1920's careers. (So did Torriente and Charleston, but they look like easy HOMers even without their pre-1920 time).


So did Taylor. This chart only picks up on his decline phase from age 31 to 40 and misses his best years in the '10s.
   167. DavidFoss Posted: November 14, 2007 at 02:29 AM (#2613980)
1. Were seasonal MLE's ever done for Gibson? He's the one that always seemed to me to have a place in the greatest-player-ever discussion.

Post #61-63 of the Josh Gibson Thread. Career OPS+ of 180. Peak seasons in the 220s. Seemed to jive with the innermost MLB batting circle.
   168. mulder & scully Posted: November 14, 2007 at 02:58 AM (#2614006)
I did some basic ones for Scales based on the revised numbers from the HOF study. Based on the HOF's numbers, Scales does not fall off a cliff in the mid-30s like previous numbers indicated. I'll post more later, but time to head home.
   169. EricC Posted: November 14, 2007 at 03:08 AM (#2614015)
Day, Leon

Pitching data?
   170. ronw Posted: November 14, 2007 at 10:35 PM (#2614951)
This data also reinforces the notion that electing Oms and Pete Hill and maybe Cool Papa (unless he was Tris Speaker in CF, which he might have been) to the HoM were mistakes.

You do know that Oms had several years before 1923 when he wasn't in the Negro Leagues, but could easily have been there. He was 28 in 1923.

Of course, Pete Hill also had a significant career before 1920. He was 39 in 1920, and had started playing baseball in about 1903. Players (especially deadball players) don't usually last 20+ years without being a HOMer.
   171. ronw Posted: November 14, 2007 at 10:35 PM (#2614952)
Pitching data?

I'll post it on Redding's thread.
   172. Dr. Chaleeko Posted: November 15, 2007 at 02:00 AM (#2615194)
The other biggest trouble in my MLEs is the WS analysis, which is tricky and hasn't ever been perfect. I believe it the least of all. I'd say my raw totals are more likely to be OK than the WS.

But Lundy's case is generally pretty square about the competition levels. I run the NgLs at .90 throughout and use Cobb's Cuba estimates (see Oms thread) for the Carribean play.

I haven't done Wells and them guys with the new data, by the way, so I can't position Lundy relative to them.
   173. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 15, 2007 at 04:37 AM (#2615316)
The WS calculations shouldn't be an issue for people who use my WARP. And for BP WARP voters, just use my BWAA+BRWAA+FWAA to get wins above average, and then just add on the BRAR-BRAA and FRAR-FRAA for the era and position, taking the numbers from another player of the same time period and position and multiplying for playing time differences.
   174. Paul Wendt Posted: November 17, 2007 at 10:49 PM (#2618672)
Davids Foss and Concepcion dlDE #158-159
> It could be that we've got better data now and that research is still actively going on and that
> we'll have even better data in the future. In that regard, the NeL constituency of the HOM is very much a
> "2004-2005 snapshot" of the best data we had available at the time.

My understanding is that this is indisputably the case.


Absolutely indisputably!
   175. Paul Wendt Posted: November 17, 2007 at 11:16 PM (#2618700)
The snapshot metaphor is apt only in the "snap" emphasizing that it was merely "this point in time" or one short interval in a long historical process. It is not apt if the "shot" is deemed to represent truth routinely, one old understanding of the photograph.

Many participants openly recognized the great uncertainty and differed in their responses and arguments.
- 1 - some for honoring those who clearly belong, given the uncertainty. Essentially that means let's develop standard low enough that we get the right size HOM by electing everyone who is, say, 90% probably above the standard.
- 2 - others for honoring those who are the best measured by highest point estimates, say a median estimate. Essentially that means between every two candidates let's elect the one who is 50% probably greater.

MLEs are point estimates. I think the whole group relied heavily on MLEs and thereby inclined to the latter position. Pressed to name regular longtime participants (not pressed very hard), I would say the group generally declined to follow Karlmagnus in this and generally followed Chris Cobb, not only adopting MLEs that he produced or coordinated but agreeing with him on how MLEs should be used.

The latter position goes further, however. Certainly Frank Grant, maybe Rube Foster and Dobie Moore, perhaps others, were elected only by going further. Marc Sunnyday's current argument for Don Newcombe is an argument for going further. There may be "probably meritorious" arguments for these players without 90% certainty that they are among the top 1000.

Marc has sometimes suggested going even further regarding the 1940s players. What if we all agree that 1-4 more players would have clear HOM playing careers except for WWII, although they may not really have playing careers that make them plausible candidates. Shouldn't we make our best guess and elect the one who seems to be the most likely loser of a HOM playing career?

--
Dick Lundy played a long time, many full seasons in the leagues, with a stellar reputation. There is fairly high floor on his merit, his downside isn't very far down. If he is elected that will be only with the help of MLEs but I wouldn't be surprised to hear that someone who takes the first approach (-1-) ranks him above ten pre-1950 black and pre-1890 white HOMers who are heavily shrouded.
   176. jimd Posted: November 21, 2007 at 12:26 AM (#2621861)
Players (especially deadball players) don't usually last 20+ years without being a HOMer.

Like Rabbit Maranville.

Dick Lundy played a long time, many full seasons in the leagues, with a stellar reputation.

Like Rabbit Maranville.

********

Based on the original MLE's, I had him as similar to Maranville, more bat, less glove, slightly behind overall. During the mid-60's, Rabbit made the bottom of my ballot, Lundy reaching the "extra credit" slots (during our 16-20 ballot experiment). Lundy appears to have gotten lost when I put all my ballot info into a spreadsheet during the late 1960's. Based on the old info he should be listed in my ballot mentions, probably in the 30's somewhere (Rabbit is currently 26th).

I'll have to evaluate the new MLE's and see how this significant boost in Lundy's bat affects his ranking in my system.

Dan has Lundy ahead of Concepcion (which is high praise coming from him). I have Dave at 10th, so if I have Dick ahead of Dave, he'll be mid-ballot or higher.
   177. sunnyday2 Posted: November 21, 2007 at 01:31 AM (#2621965)
If LundyMLE looks like Rabbit, you can probably deduce that Lundy had a higher peak in real life.
   178. jimd Posted: November 21, 2007 at 02:47 AM (#2622059)
And it wasn't just me making that particular comparison either.

21. Eric Chalek (Dr. Chaleeko) Posted: January 14, 2005 at 10:32 AM (#1079448)

Chris's WS estimateds paint a picture of a player who is very close to Rabbit Maranville, but not quite as meritorious.

Their 162-adjusted peak/prime/extended prime are all quite close, though Rabbit's is a little better overall. And he's got just a little more career to boot.

Another guy to compare Lundy to, in a more modern sense, might be Dave Concepcion. Better than average hitters in their good years. I see Concepcion's peak as a little higher, but his career and extended prime as a little bit lower.
   179. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 21, 2007 at 09:27 AM (#2622365)
Right. The old MLE had Lundy right around Maranville; the new one makes him look like Bobby Wallace. Big difference.
   180. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 26, 2007 at 10:32 PM (#2625520)
Voters are jumping on the Lundy bandwagon left and right. Don't get left behind!
   181. karlmagnus Posted: November 26, 2007 at 11:22 PM (#2625589)
Let its tyres down. Stop the runaway while we can!
   182. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 27, 2007 at 02:06 AM (#2625703)
The baseball acumen of anyone who spells tires with a Y can't be trusted!
   183. 'zop sympathizes with the wrong ####### people Posted: November 27, 2007 at 02:13 AM (#2625711)
Seriously. What are you, French?
   184. John (You Can Call Me Grandma) Murphy Posted: November 27, 2007 at 02:21 AM (#2625714)
Seriously. What are you, French?


He's from jolly old England.
   185. Dr. Chaleeko Posted: November 27, 2007 at 03:37 AM (#2625761)
He's from jolly old England.

And, despite a penchant for argumentative language, he's jolly!
   186. sunnyday2 Posted: November 27, 2007 at 03:42 AM (#2625763)
And, despite a penchant for supporting the most obscure players imaginable, he's jolly! More obscure than yest's. Better, but more obscure.
   187. karlmagnus Posted: November 27, 2007 at 01:24 PM (#2625908)
Obscurity is itself an indicator of quality. If Chuck Finley, despite the advantage of a doppelganger called Steve, had not had that run-in with Tawny Kitaen, he might have remained obscure enough to make my ballot :-)
   188. KJOK Posted: November 29, 2007 at 05:53 PM (#2628101)
Gary A. recently posted some very interesting info on his site from passports, including passport pictures of Lundy, Redding, Charleston, etc. Funny that on some of the passport photos, the guys are in their baseball uniforms..

1920 Bacharachs Passport Info and Photos
   189. KJOK Posted: September 17, 2011 at 06:50 AM (#3927857)
   190. KJOK Posted: September 17, 2011 at 08:46 PM (#3928237)
CORRECTED LINK:

Dick Lundy's Real Stats
Page 2 of 2 pages  < 1 2

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
Eugene Freedman
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 1.0588 seconds
49 querie(s) executed