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

Friday, February 18, 2005

Mule Suttles

Mule Suttles

John (You Can Call Me Grandma) Murphy Posted: February 18, 2005 at 03:04 AM | 95 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

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

   1. John (You Can Call Me Grandma) Murphy Posted: February 18, 2005 at 03:14 AM (#1151547)
hot topics
   2. Chris Cobb Posted: February 19, 2005 at 03:07 AM (#1153567)
Mule Suttles Data, mostly from Holway

Teams: 23-25 Bir Black Barons, 26-31 Stl Stars, 32 Det/Was, 33-35 Chi Am Giants, 36-40 Newark Eagles, 41 NY Black Yankees, 42-44 Newark Eagles

Park Effects: (as far as I know) Birmingham played in a very good pitcher’s park, St. Louis played in an all-time great hitter’s park, Chi Am Giants an all-time great pitchers’ park, Newark in a very good hitter’s park.

Seasonal Data from Holway

1923 .243 for Bir Black Barons; lf
1924 .317 for Bir; lf
1925 .428 for Bir; 1st in ba, 14 hr (3rd), 36 hr/550 ab (3rd); 1b
1926 .498 for Stl; ba 1st, 27 hr (1st), 70 hr/550 (1st), 27 2b (1st), 21 3b (1st), 11 sb; 1b; all-star, MVP
1927 .476 for Stl; (1/4 season, was beaned & missed the rest of the year) 53 hr/550 (1st); 1b
1928 .372 for Stl; ba 3rd, 19 hr (3rd), 38 hr/550 (4th), 9 3b (1st); 1b, all-star
9-27 in playoff vs. Chi Am Giants
1929 .351 for Stl; 20 hr (2nd), 35 hr/550 (4th), 29 2b (1st); 1b, all-star
3-13 in World Series vs. Homestead
7-20 vs. major-league competition
55-175 in Cuban Play
1930 .422 for Stl.; ba 1st, 12 hr (2nd), 43 hr/550 (1st), 9 3b (3rd); ut, all-star (as dh)
.389 for Bal Black Sox; 10 hr (1st), 73/hr 550 (1st) [57/550 combined], 7 2b (4th); 1b
1931 .349 for Stl; ba 4th; lf
2-4 vs. major-league competition
1932 .315 for Det, .322 for Was; 9 hr (3rd), 31 hr/550 (1st), 12 2b (3rd); 1b, all-star
1933 .254 for Chi Am Giants; 5 sb (4th); 1b
1934 .254 for Chi Am Giants; 12 hr (3rd), 32 hr/550 (4th); 1b
3-15 in playoff vs. Phi Stars
1935 .263 for Chi Am Giants; 13 hr (2nd), 40 hr/550 (1st); 1b
5-12 vs. major-league competition
1936 .365 for Newark Eagles; ba 4th, 15 hr (2nd), 57 hr/550 (3rd); 1b, all-star
1937 .345 for Newark Eagles; 16 hr (2nd), 80 hr/550 (1st); lf
1938 .282 for Newark Eagles; 14 hr (1st), 51 hr/550 (1st); 1b
1939 .282 for Newark Eagles; 12 hr (2nd), 46 hr/550 (3rd), 6 2b (5th), 2 3b (3rd); 1b
4-15 in playoffs
0-4 vs. major-league competition
31-142 in Cuban Play
1940 .262 for Newark Eagles; 8 hr (4th), 26 hr/550 (5th), 11 2b (5th), 3 sb (1st); lf
1941 .222 for NY Black Yankees; lf
1942 .385 for Newark Eagles; ut
1943 no data, manager of Newark Eagles
1944 .250 for Newark Eagles; ut and manager

Career Batting
1103-3230, .341
237 hr, 40/550 ab
58-170, .341 vs. major-league pitching
.329 mean avg. for 21 seasons with data

Career from MacMillan 8th ed.
830 g, 2962 ab, 1001 hits, 167 2b, 63 3b, 183 hr, .338 ba

Career from MacMillan 10th ed.
870 g, 3077 ab, 1011 hits, 171 2b, 65 3b, 190 hr, .329 ba

Black/Gray Ink
56/145

Black Ink: #2 behind Gibson. Both of these players benefit by the heavy weighting of available black ink information towards home runs.
Gray Ink: #4 behind Stearnes, Charleston, and Gibson

Career Fielding

1b 25-30, 32-36, 38-39
lf 23, 24, 31, 37, 40-41
ut/ph 42-44

Brief Comment: Suttles was, by all accounts, a tremendous power hitter. The difficult question here is, How much was he helped by his parks? I don’t think that’s a question that could move him across the in-out line, but it’s a question could move him ahead or behind Turkey Stearnes (my uninformed belief is behind), and that has bearing on little matters like, would he have hit 450 or 550 home runs in the major leagues.
   3. Chris Cobb Posted: February 19, 2005 at 03:13 AM (#1153572)
Sorry, forgot to include slugging percentages derived from the MacMillan data.

MacMillan 8th .622

MacMillan 10th .612
   4. KJOK Posted: February 19, 2005 at 05:22 AM (#1153657)
Not so sure Park Effects won't significantly move Stearnes down. For exmaple, here are the averages for the 1926 St. Louis Stars:

Suttles, Mule 0.498
Murray, Mitch 0.396
Ross, William 0.379
Wells, Willie 0.371
Bell, Cool Papa 0.355
Bobo, Willie 0.347
Creacy, Dewey 0.337
Redus, Frog 0.323
Russell, Branch 0.319
Russel, Johnny 0.313

Suttles DOES have the best average on the team, but you can see how inflated the averages are...
   5. David C. Jones Posted: February 19, 2005 at 06:01 AM (#1153702)
I think it's debatable whether Suttles was a better hitter than Stearnes or not. But in the other aspects of the game, fielding and baserunning, it seems clear that Turkey is well ahead of Mule.

Suttles had a reputation for being a free-swinger who struck out a great deal.
   6. Chris Cobb Posted: February 19, 2005 at 03:22 PM (#1154172)
On the St. Louis Park Effects: I'm hoping that Gary A. will be able to give us raw park factors (if he hasn't already -- I should go look) for 1928 to see if there is evidence that we should be looking at St. Louis more as a 120+ type park than as the 110-type that I would generally use as an estimate.

Suttles maintained his stature as a pure power-hitter in Chicago from 34-36, though his batting average dropped a long way.

While I'm creating a wish-list for comprehensive historical data, I'll add that it would be great to have some league-wide data for a season in the mid-30s, to see how NeL levels of offense were tracking compared to the majors. Setting MLEs for the 1930s will be tricky in any case because of the divergence in offense levels between the National and the American leagues.
   7. Howie Menckel Posted: February 19, 2005 at 05:27 PM (#1154273)
baseballlibrary.com excerpt:

Suttles began playing Negro League baseball at age 17 and lasted until 42......
During his five years with the St. Louis Stars (1926-30), he led the league in homers twice and in doubles, triples, and batting average once...... His average in 26 documented games against white major leaguers was .374, with five HR.
   8. Chris Cobb Posted: February 19, 2005 at 06:03 PM (#1154316)
Holway has Suttles as 58-170, .341, with 10 home runs, vs. major-league competition.
   9. karlmagnus Posted: February 19, 2005 at 11:18 PM (#1154725)
I'm becoming fairly convinced Stearnes>Suttles, but does Suttles>Beckwith, and if so by how much? Stearnes looks currently likely to be #3, after Simmons and Beckley but before Welch, Suttles probably on but at about 12-13, and Beckwith just off, probably at 18-19. Recognizing that you would probably rank all three higher, have I got the order and the relative positioning right?
   10. Chris Cobb Posted: February 19, 2005 at 11:43 PM (#1154755)
I'm becoming fairly convinced Stearnes>Suttles, but does Suttles>Beckwith, and if so by how much?

That's a very good question; from what I know so far, I don't think I know for sure yet.

In the absence of clear evidence to the contrary, Suttles obviously ranks higher. He has the better reputation (though we know that Beckwith's reputation has been [probably] undeservedly tarnished), and he definitely had a significantly longer productive career.

However, Beckwith was the more valuable defensive player and may have been a better hitter. I'd say it's possible that Beckwith's career NeL average was 25 points higher after Suttles is park-adjusted, and his power may turn out to be close to Suttles'. And then there's the matter of plate discipline. We've heard it that Suttles was a free-swinger; Beckwith appears to have been willing to take a walk. We'll see what the evidence shows.

Given those issues, if I had to submit a final 1946 ballot today, I'd put Suttles higher, but I'm keeping an open mind until I have a more reliable view of Suttles' career. It's possible that Suttles was a lot better than Beckwith; it's also possible that Beckwith was slightly better than Suttles.

Of course, the usual variations in ranking corner outfielders/first basemen against middle infielders, and ranking peak vs. career will also influence how one decides to rank the two.
   11. Gary A Posted: February 20, 2005 at 12:14 AM (#1154788)
1928 Mule Suttles
NNL St. Louis Stars

G-80 (team 83)
AB-310
H-112
D-19
T-11 (3rd)
HR-20 (tied for 3rd)
R-71
W-23
HP-1
SH-10
SB-5
TB-213 (2nd)
AVE-.361 (NNL .278; 2nd)
OBA-.407 (NNL .333; 5th)
SLG-.687 (NNL .384; 2nd)

There's also a double-header in which Suttles hit 3 homers, but I don't have box scores.

He was outhit by teammate Willie Wells, though it was close (.365/.425/.699).
   12. Gary A Posted: February 20, 2005 at 02:28 AM (#1154997)
1928 Mule Suttles
1b-78, 3b-1, rf-1

Fielding (1b)
G-78 (team 83)
DI
PO-852
A-29
E-15
DP-41
FPCT-.983 (NNL 1b .983)

Bill James subtracts pitchers' putouts from first base assists to try to measure first basemen's range; fwiw, the Stars' figure in 1928 (35 first base assists minus 9 pitcher putouts) is easily the best in the NNL, at a rate (.0325 per 9 innings) that's more than twice league average (.0129). Suttles did NOT have a reputation as a good first baseman, though.

When the Stars acquired George Giles, Suttles was moved to left. As far as I can tell, this was not because of Suttles' ability as an outfielder, but because Giles was considered a brilliant first baseman. In 1934, the Chicago Defender considered it a move that strengthened the western team's defense when manager Dave Malarcher once again moved Suttles to left to make room for Giles at first.

This is evidence that, well into the 1930s, first base was considered a more importance defensive position than at least left field (probably rf, too).
   13. Gary A Posted: February 20, 2005 at 02:37 AM (#1155022)
From the Beckwith thread, Stars Park raw factors (home/away):

1922: 103
1923: 155
1928: 124

I don't have detailed home/road splits for any Stars Park seasons yet. But here are the Stars' regulars (with two catchers) in 1928:

Cool Papa Bell 320/377/453
Branch Russell 303/372/452 (12 triples)
Willie Wells 365/425/699 (28 doubles, 23 hr)
Wilson Redus 330/390/612 (20 hr)
Dewey Creacy 327/378/508
Mule Suttles 361/407/687
J.H. Russell 276/312/355
Henry Williams 299/309/407
Spoony Palm 313/374/580

The team as a whole hit .313/.364/.502, leading the league with 569 runs in 83 games.
   14. Mark Shirk (jsch) Posted: February 20, 2005 at 02:41 AM (#1155035)
Gary,

Gary is it evidence that 1B was considered a more important defensive position than LF or is it evdence that Giles was that much better than Suttles?

For instance, if a team that already had, say Richie Sexson went out and acquired Derek Lee, would Lee be moved to LF or would Sexson? I would think that sexson woudl because Lee is just a really good 1B. And as they say in football, you don't move a blue (chip).
   15. Gary A Posted: February 20, 2005 at 02:46 AM (#1155047)
In post #12, I meant that 1b seems to have been considered more important than corner of IN THE NEGRO LEAGUES (of course).

Also, I miscalculated the rate of (1b a - p po)/innings for the Stars; their rate was .0352, compared to league average .0129. AND it's simply per inning, not per 9 innings. Not that it matters that much.
   16. Gary A Posted: February 20, 2005 at 02:54 AM (#1155062)
It's possible, although there's lots of other evidence to indicate that first base was given a great deal of importance in the NeL. I've remarked elsewhere how Charleston moved from center field to take over first base in 1922 when Ben Taylor was hurt, and how he seemed to be regarded as filling a gap (surely we'd today consider that an odd move for a brilliant, or even simply good, center fielder). I've seen almost as many remarks about first basemen's defense in 1920s newspapers as any other position; fans seemed in some instances to care about it a great deal. Poor first basemen were roundly abused--George Carr in 1928, for example, who was blamed by Bacharachs' fans for not being the smooth Chance Cummings.

Buck O'Neil, himself a slick first baseman, might give us more evidence for this attitude; from what I understand, he didn't back Suttles for the HOF because of his first-base defense.
   17. OCF Posted: February 20, 2005 at 03:19 AM (#1155115)
(surely we'd today consider that an odd move for a brilliant, or even simply good, center fielder)

Darin Erstad? Of course everyone around here did consider that an odd move, so your point stands.
   18. Gary A Posted: February 20, 2005 at 03:29 AM (#1155134)
Anyway, I'm not making an argument to raise Suttles's defensive standing or anything; this is just a little pet theory of mine about the NeL defensive spectrum. Giles was certainly considered a far better fielding first baseman than Suttles, and Suttles' oufield defense was surely nothing special.
   19. Mark Shirk (jsch) Posted: February 20, 2005 at 05:13 PM (#1155800)
Of course Ersatd moved because of worries about his staying healthy. Whether a healthy Darin Erstad playing 1B everyday is a good hting (it isn't) the Angels didnt' seem to think about.

And that Charleston comment is odd. Was Torriente on that team?
   20. Chris Cobb Posted: February 23, 2005 at 04:26 AM (#1160572)
I've started entering data for a first run at MLEs for Suttles, so it's time to talk park factors!

As I see it, we have four parks in Suttles career for which adjustments are or may be needed: Birmingham, St. Louis, Chicago, and Newark, pitcher, hitter, pitcher, hitter.

We have raw park factors for some seasons for St. Louis and Chicago from Gary A., and some analysis (inconclusive, if I recall correctly) from gadfly on Newark.

St. Louis, from above: 103, 155, 124
Chicago, from Beckwith thread: 85, 54, 66, 81

First question: how best to turn the "raw" park factors into adjustments of player stats? If they give the ratio of team scoring in home games to away games, as a first step should their variance from 100 be divided in half (assuming equal numbers of home-road games, which we know wasn't the reality, but must serve in the absence of seasonal home-road splits)? If we take the three park factors for St. Louis, take the average and divide by two, we get a factor of 114 by which Suttles totals would be reduced.

Third question: How to set park factors on a combination of evidence and anecdote? Given the reputation of the St. Louis park, I could see an argument to boost the 114 factor to 120 or so. I'd been using 90 for Chicago, but my sense from Gary A.'s data is that should go down to about 85.

Fourth question: Factors for Birmingham and Newark? In the absence of any data to guide, I'd probably set Birmingham at 95 and Newark at 105 (adjusted -- this is the factor I'd apply to Suttles' numbers). Does that seem reasonable, or is the evidence of ballpark side and anecdote insufficient to apply adjustments?
   21. Gary A Posted: February 23, 2005 at 06:27 AM (#1160688)
Birmingham home/away park factor for 1928: 102, with 38 games at home, 58 on the road.

Chicago also had a park factor of 57 in 1928 (i.e., the 54 in 1921 is not a lone outlier).

Chicago home/road games, pf:
1920: 34/20, 85
1921: 41/42, 54
1922: 50/19, 66
1923: 43/37, 81
1928: 78/33, 57

St. Louis home/road games, pf:
1922: 17/18, 103
1923: 58/23, 155
1928: 59/39, 124
   22. karlmagnus Posted: February 23, 2005 at 01:54 PM (#1160856)
Chris, given the tiny numbres of game data points, even over several years, you may want to regress the park factors season by season towards 100 by the square root of games played, divided by sqrt (154) just as you did for hitter stats. You have to assume that some of the extreme park factors are just random noise.

Since Suttles has some very high park factors and some low, the overall effect for him won't be that great; it may affect those who played mostly in pitchers' parks and have been reinflated excessively (Beckwith?) Conversely some who played in hitters parks and have been deflated (Poles, DeMoss? to think of 2 who came out worse than I expected) may need reflating a bit.
   23. Gary A Posted: February 23, 2005 at 03:49 PM (#1160959)
Beckwith played with Schorling Park as his home park in only 1922 and 1923, and I'm pretty sure Chris used a pf of 90 for those years.

DeMoss, as we've been discussing, probably played the bulk of his recorded career (in the league era) in Schorling Park. (He also played in Detroit for a while at the end of his career, but he was around 40 by then.)
   24. jimd Posted: February 23, 2005 at 09:07 PM (#1161658)
Paul Wendt recently posted a correction needed to adjust extreme "raw park factors". When I have some time, I'll look for it. Why is it needed?

Assume we have a park in an 8 team league that magnifies offense by 35%. The other 7 parks must then average out to 95% of league average, so that the whole league averages out to 100%. The measured "raw park factor" at the extreme park will then be 135/95 = 142; it is overestimated because the road sample is biased away from average. The correction changes that 142 to 135.

*******

For this I'm using the park factors in the Lahman DB, which IIRC are basically the Total Baseball factors (schedule adjusted and opposition quality adjusted as described in Total Baseball, though the latter adjustment really only matters for extreme offensive or defensive teams)

120 is EXTREMELY high. The only AL/NL park which has attained that level even once is Coors. Is St. Louis really another Coors?

85 is EXTREMELY low. There has never been a NL/AL park which has attained that level. (Some tiny-sample Union Association parks, that's it.) 89-90 is as low as it gets after 1890, and Milwaukee County Stadium is the only park to maintain that for 2-3 years, and that was PPF only (when the staff was penalized for not having to face that potent offense). Is Chicago more extreme than any historical AL/NL park?
   25. jimd Posted: February 23, 2005 at 10:02 PM (#1161792)
First question: how best to turn the "raw" park factors into adjustments of player stats?

In Win Shares (pp86+), James describes how he converts the "raw park factors" available to him from STATS. (Note that he does not use the correction noted by Paul Wendt, though he should.) His conversion performs the balanced-schedule adjustment (which he uses whether the schedule was balanced or not). He also describes how he "stabilizes" ("regresses") the raw factors pre-1908 before doing the schedule adjustement; this is a procedure that probably should also be applied to the Negro League factors in the absense of a more sophisticated treatment.
   26. Chris Cobb Posted: February 23, 2005 at 10:40 PM (#1161883)
jimd,

It's because these adjustments seem very extreme that they need discussion. Let me focus on Chicago first.

Using Gary A.'s data, I calculate that Chicago's adjusted park factor, when home and road games are averaged together, is for 396 games over 5 seasons, 80.5, with the seasonal factors calculated separately and then averaged, with each season receiving equal weight.

Making a correction of the sort you describe, if I understand it correctly, would, for an 8-team league, raise the factor of 80.5 to 83.3.

Given these numbers, it seems to me that using 85 as the adjusted park factor, though extremely low, is fully in keeping with the evidence we have, which includes a number of games equal to 2 2/3 major league seasons.

If, given this evidence, there's a reasonable argument to be made for being still more conservative and using a park factor of 90 as I have, I could be persuaded to do it, but, given the size of the sample and the numbers that derive from this sample, I think 85 is accurate.

Using the same procedure, I get an adjusted park factor of 118.4 for St. Louis over 214 games in three seasons. Adding the extreme-park adjustment lowers it to 1.158. The data, then, when fully adjusted, do not support a factor of 120; let me propose instead 115.

Here's an anecdotal but relevant example to test the appropriatness of these factors. Using Macmillan 8th data, Mule Suttles shows up as a .389 hitter in 999 at bats in 5 seasons in St. Louis and a .285 hitter in 309 at bats in Chicago over three seasons. Now, there may have been some changes in league offense level, and we could expect some decline in Suttles as a hitter, but that won't account for a 100-point swing in batting average (and Suttles avg. goes back up again when he goes from Chicago to Newark). We are dealing with some large park factors here! Adjusting the high average by a schedule-adjusted park factor of 115 and the low average by a factor of 85 bring the two averages to .338 and .335. Suttles career avg. in MacMillan 8 is .329. The seasons in St. Louis and Chicago are all in Suttles' prime, so they should be higher, probably, than his career average. Anyway, I think that the shape of Suttles' career suggests that we have to be open to considering park factors for these to parks that at or beyond those of the most extreme hitter & pitcher parks in the majors.

What is reasonable, what is unreasonable, in the foregoing analysis? Would it be better, even given this evidence, to use 90 and 110 for St. Louis and Chicago, or some other number part-way between a convenient round sum and the number indicated by the data we have?
   27. Chris Cobb Posted: February 23, 2005 at 10:42 PM (#1161886)
jimd,

Just saw your post with page numbers from _win shares_ on regression. I'll take a look when I get home, and I look forward to comments from you and others on the above analysis.
   28. KJOK Posted: February 23, 2005 at 10:58 PM (#1161923)
For this I'm using the park factors in the Lahman DB, which IIRC are basically the Total Baseball factors (schedule adjusted and opposition quality adjusted as described in Total Baseball, though the latter adjustment really only matters for extreme offensive or defensive teams)

120 is EXTREMELY high. The only AL/NL park which has attained that level even once is Coors. Is St. Louis really another Coors?

85 is EXTREMELY low. There has never been a NL/AL park which has attained that level. (Some tiny-sample Union Association parks, that's it.) 89-90 is as low as it gets after 1890, and Milwaukee County Stadium is the only park to maintain that for 2-3 years, and that was PPF only (when the staff was penalized for not having to face that potent offense). Is Chicago more extreme than any historical AL/NL park?


Jimd, a slight nitpik, these are THREE YEAR factors you're using from Total Baseball 8, instead of the Total Baseball calculation for just one year. These parks all had extreme 1-year Total Baseball 8 Calc factors:

1 Year BPF: (Post-1899)
1934 Braves Field - 86
1950 Braves Field - 86
1976 Astrodome - 86
1926 Braves Field - 86
1906 Washington Park III - 86
1973 Oakland Colisuem - 87
1932 Comiskey Park - 87
1977 Jack Murphy Stadium - 88
1938 Braves Field - 88
1967 Dodger Stadium - 88
1981 Astrodome - 88
1972 Jack Murphy Stadium - 88
1946 Cleveland Combined - 88
1972 Anaheim Stadium - 88
1976 Jack Murphy Stadium - 88

and going the other way (exluding Coors):
1955 Fenway Park - 122
1900 South End Grounds - 120
1906 Hilltop Park - 120
1993 Mile High Stadium - 120
1970 Wrigley Field - 120
   29. KJOK Posted: February 23, 2005 at 11:05 PM (#1161935)
Here are the Park Factors for the Chicago Park when it was used in the AL by the White Sox:

1901 99
1902 96
1903 92
1904 99
1905 97
1906 92
1907 104
1908 94
1909 97
1910 95

and I think it could be argued that Negro League Parks had more variability (more extreme both ways) in how they impacted offense than AL parks.....
   30. karlmagnus Posted: February 23, 2005 at 11:16 PM (#1161958)
The fact that you don't get a run of years at these extreme park factors demonstrates that they're random fluctuations, because park owners don't change the fences etc. every year, so if Fenway was a hitters park in 1955 it should have been exactly equally a hitters' park in '54 and '56, absent fence movements and stand construction (which for Fenway, one could check.)

I would also distrust intensely very LOW park factors, because except for home runs there should be a limit, which would be that of playing on an open field with no fences at all. The lowest established park factors in MLB, in 106 seasons, are 86-88, even with random flucatuations; I don't see how an NL park could physically have a "real" park factor that was lower than this, say a "real" 88 that fluctuated down to 86 randomly.

High park factors are more likely, if you just had a small park.
   31. Gary A Posted: February 24, 2005 at 12:09 AM (#1162052)
About the notion of a lower limit to park factors: actually, there wouldn't be any such thing, since park factors are always *in relation to other parks.* A large park with a bad hitting background and large foul territory, placed in a league with seven bandboxes, could suppress offense enormously with respect to the league. I don't see why 88 or 86 or any other number represents a "true" lower limit. Park factors really measure variability among parks.

With respect to the variability of Negro League parks: they ranged from current/former major league parks, to minor league parks, to college parks, to amateur/semipro parks, to parks built by NeL teams, which didn't have nearly the financial resources of major league teams, of course, and sometimes had to accommodate their parks to some real geographical or architectural peculiarities (the trolley barn that cut into Stars' Park left field).

When you put higher actual variability among parks together with higher statistical fluctuations caused by short, unbalanced schedules--and it's not at all surprising that NeL park factors jump around and sometimes look extreme, when compared to major league park factors.
   32. jimd Posted: February 24, 2005 at 01:28 AM (#1162161)
Chicago home/road games, pf:
Gary A's Data     After Wendt  After Bal-Sched (8 teams)
1920: 34/20, 85        87          94
1921: 41/42, 54        57          80
1922: 50/19, 66        69          87
1923: 43/37, 81        83          92
1928: 78/33, 57        60          83
5 Yr average 69        71          87

Now, is their road data skewed by extra trips to St. Louis or other high context parks?

St. Louis home/road games, pf:
Gary A's Data     After Wendt  After Bal-Sched (8 teams)
1922: 17/18, 103       103         101
1923: 58/23, 155       145         119
1928: 59/39, 124       120         109
3 Yr average 127       123         110

Now, is their road data skewed by extra trips to Chicago or other low context parks?

Notes: The after-balanced number is analogous to what James calls the "Park Run Adjustment". If there aren't 8 teams in the league, the numbers above aren't right. (Never mind that they're stabs in the dark anyway due to the severely imbalanced schedules. 50Home 19Away: the balanced schedule adjustment is going to be close as the Virgo super-cluster with that one...)

I think it's more prudent to err on the side of caution, and then change the factor as more evidence becomes available. (I agree with James' caution on 19th century park-factors; they can make a huge difference in the evaluation. BP has some problems with this in the 1870's.)
   33. jimd Posted: February 24, 2005 at 01:47 AM (#1162177)
Jimd, a slight nitpik, these are THREE YEAR factors you're using

Well, if we're just talking about using 85 for Chicago 1921, I agree you have picked more than a nit. OTOH, if we're talking about using 85 for Chicago, all years, then I think looking at the averages is appropriate (and it WAS handy ;-). Extreme factors boost the disadvantaged side enormously while harming the other side and have to be used cautiously.
   34. Chris Cobb Posted: February 24, 2005 at 01:54 AM (#1162188)
While I'm not sure that the balanced-schedule adjustment is the best way to go, I think that 87 and 110 as the park factors for Chicago and St. Louis are reasonable for now. The 87, in fact, represents a change based on more evidence from the 90 that I had been using for Chicago.

I'm not going to finish the other preparations I need to run MLEs for Suttles until Friday, so there's more time for discussion, but right now I like the look of these numbers.
   35. karlmagnus Posted: February 24, 2005 at 02:59 AM (#1162301)
I don't think 87 is unreasonable; for reasons discussed above I'd be dubious about any lower than that. In a league of 7 bandboxes and a prairie, the distortions beocme mind-boggling -- you'd basically want an entirely different tean for the prairie -- give the home run hitters and the ground ball pitchers the day off.
   36. jimd Posted: February 24, 2005 at 03:52 AM (#1162374)
In a league of 7 bandboxes and a prairie

When originally constructed (1915), Braves field was a prairie. 401 down the foul lines, 550 at it's deepest point to the right of CF (nearly square, nearly double the fair territory of a modern park). It's designer thought that ITP-HR's were the most exciting play in baseball. (Maranville's early "power" was really speed.)
   37. Brent Posted: February 24, 2005 at 05:43 AM (#1162629)
Adjusting the high average by a schedule-adjusted park factor of 115 and the low average by a factor of 85 bring the two averages to .338 and .335.

Are the park factors for runs per game, or for hits per at bat? My understanding has always been that there is about a 2-to-1 relationship between these variables, so that if a park raises runs per game by 10 percent, it will raise batting average and other per-at-bat (or per-plate-appearance) averages by about 5 percent. At least that's how I interpret the way that OPS is mathematically constructed, and the explanation of conversions given by Bill James. (For example, in the Willie Davis comment of the NBJHBA, he says "if you increased each run element by ten percent and froze the batting outs, a player's runs created would increase by almost twenty percent.")

Some of you know much more about sabermetrics than I do, so please correct me if I'm off track. But it seems important to know what units the park factors are in, and not adjust everything by the same factor.
   38. jimd Posted: February 24, 2005 at 06:25 AM (#1162734)
Usually, when someone mentions "park factor", I interpret it as a run factor (and some info on how it was calculated because of the lack of standard terminology/method).

There are also HR factors available (STATS has published these, Total Baseball in some editions). Any stat element can be put through the park factor calculation process to see if there is some interesting variation, given the available retrosheet files, and a well-chosen rate basis.
   39. jimd Posted: February 24, 2005 at 06:27 AM (#1162738)
and WANT some info
   40. KJOK Posted: February 24, 2005 at 06:30 AM (#1162743)
Are the park factors for runs per game, or for hits per at bat?

Runs per game I'm quite sure.
   41. jimd Posted: February 24, 2005 at 06:35 AM (#1162758)
nearly double the fair territory of a modern park

Goofed on my calculator. It was nearly double the fair territory of Baker Bowl, and about 50% more fair territory than a modern park with 330 ft foul lines.
   42. Dr. Chaleeko Posted: February 24, 2005 at 02:37 PM (#1163171)
I'm still struggling generally with where to slot Suttles, and I'm wondering if anyone can give a general sense of his career shape. To that end, here's some possible scenarios, can anyone describe which one or two seem most likely to describe him?

As always, I'm talking in terms of WS: 3 yr peak, 10 year prime, career.

1) 110+ 3 yr, 300+ 10 yr, 375-400 career
2) 100+ 3 yr, 275+ 10 yr, 350+ career
3) 100+ 3 yr, 250+ 10 yr, 300-325 career
4) 95+ 3 yr, 250+ 10 yr, 300 career
5) 95+ 3 yr, 250+ 10 yr, 275 career
6) 95+ 3 yr, 270+ 10 yr, 375+ career

Obviously, I'm patterning these after MLB players:
1 is roughly Al Simmons
2 is roughly Jimmy Sheckard
3 is roughly Dale Murphy
4 is roughly George Sisler
5 is roughly Don Mattingly
6 is roughly Roberto Clemente

Thank you!
   43. Chris Cobb Posted: March 01, 2005 at 05:30 AM (#1172077)
Here are revised MLEs for Beckwith and new MLEs for Suttles, using new park factors, the .90/.82 conversion factor, slightly adjusted league offense level factors, and a new regression (I added a ghost-year of 0 value at the beginning of the career to model career development more accurately in the regression – it brought the regressed career totals much closer to the straight MLE totals.

Left two columns are straight conversion of the NeL statistics; right two columns are regressed values.


John Beckwith Revised MLEs

Year MLE BA MLE SA Reg. BA Reg. SA
1919 0.188  0.233  0.250   0.368
1920 0.215  0.260  0.288   0.377
1921 0.387  0.550  0.359   0.527
1922 0.336  0.469  0.337   0.497
1923 0.321  0.568  0.332   0.571
1924 0.370  0.561  0.357   0.571
1925 0.376  0.663  0.366   0.610
1926 0.335  0.567  0.343   0.552
1927 0.333  0.455  0.337   0.493
1928 0.312  0.411  0.330   0.481
1929 0.342  0.565  0.338   0.538
1930 0.403  0.663  0.361   0.594
1931 0.308  0.558  0.323   0.564
1932 0.300  0.541  0.319   0.549
1933 0.400  0.380  0.307   0.479
1934 0.141  0.174  0.200   0.218
1935 0.000  0.000  0.083   0.136

car. 0.337  0.534  0.339   0.537


Mule Suttles MLEs

Year MLE BA MLE SA Reg. BA Reg. SA
1923 0.237  0.301  0.253   0.346
1924 0.289  0.369  0.299   0.417
1925 0.397  0.592  0.371   0.583
1926 0.420  0.943  0.384   0.794
1927 0.380  0.720  0.368   0.687
1928 0.308  0.530  0.315   0.561
1929 0.291  0.523  0.297   0.536
1930 0.349  0.662  0.331   0.614
1931 0.254  0.416  0.282   0.505
1932 0.290  0.476  0.290   0.500
1933 0.275  0.527  0.278   0.494
1934 0.260  0.446  0.270   0.467
1935 0.266  0.490  0.273   0.507
1936 0.329  0.582  0.302   0.547
1937 0.311  0.656  0.289   0.583
1938 0.255  0.459  0.255   0.480
1939 0.254  0.492  0.249   0.464
1940 0.234  0.383  0.232   0.387
1941 0.135  0.176  0.147   0.204
1942 0.375  0.410  0.253   0.338
1943 0.000  0.000  0.131   0.172
1944 0.245  0.363  0.260   0.360

car. 0.298  0.518  0.294   0.515



Suttles’ projected career MLE totals, 1923-41
2420 games, 9680 ab, 2855 hits, 5073 total bases
.295 ba, .524 sa

The projected career averages are higher because 1) the seasons are slightly reweighted when projected into full major-league seasons and (more important) 2) seasons at the end of his career in which Suttles was clearly not a major-league quality player have been dropped, raising his career averages.

I intend to do similar projections for Beckwith to see their effect on him. When I have that data I’ll do a full post on Beckwith on the Beckwith thread.

I also haven’t had time to do win shares for Suttles or revised win shares for Beckwith.

I’ll work on those after I get MLEs for Turkey Stearnes done. Those I should have by Wednesday night; win shares for the weekend. But with the seasonal averages and career totals I expect you all will know how you want to rank Suttles, insofar as you use this data.

I'm happy to answer questions about the MLEs and the projections, but I'm in a hurry tonight just to get the data up, so no detailed description of these matters for now.
   44. Brent Posted: March 01, 2005 at 01:19 PM (#1172450)
Hmmm... a long career, .295, .524, a bit of a free swinger. It's hard to think of a comparable player from the 1920s and 30s. Would Reggie Jackson be a good comparison?
   45. karlmagnus Posted: March 01, 2005 at 01:55 PM (#1172465)
He's Jackson's numbers but in a better hitting era, so not quite Jackson. Borderline HOM, for me just below, I think, but a tiny bit above Beckwith. Others may be more generous.
   46. andrew siegel Posted: March 01, 2005 at 02:03 PM (#1172468)
Those numbers are quite a bit lower than my image of Suttles (and my voting position for him). I appreciate Chirs's hardwork. Now I (and we) need to walk the tigthrope of neither ignoring nor blindly defering to his study. That's not easy to do.
   47. Dr. Chaleeko Posted: March 02, 2005 at 12:32 AM (#1173787)
The MLEs for Beckwith and Suttles put them in some very interesting company. Using the SBE, I hunted for players contemporary with JB and MS with similar AVG and SLG. And I tried to find players who did so in a simmilar number of games.

Here's what I found

NAME.......AVG...SLG...G
Beckwith...339...537...1900ish
===========================
Simmons....334...534...2215
Heilman....342...520...2148
Klein......320...543...1753
Averill....318...533...1669
Bb Herman..324...532...1552
K Williams.319...531...1397
Hafey......317...526...1203
O'Doul.....349...532...970

Not bad, he's right there with the two big boys an his G aren't higher than the borderline HOFs and HOVG guys. Notice also, all these guys are outfielders and first basemen, not 3B/SS like Beckwith.

NAME.......AVG...SLG...AB
Suttles....294...515...9680 ABs
===========================
Ott........304...533...9456
Bottomley..310...500...7471
B Johnson..296...506...6920
Berger.....300...522...5763
Meusel.....309...497...5475
Troskey....302...522...5161
Heath......293...509...4937
H Wilson...307...545...4760
Keller.....286...518...3970

(a couple of the at-bat totals might be off, I had a little trouble with my own handwriting). Suttles is a more complex case. He fits this group neatly because he is also a 1B/OF. The good news is that Ott is the most similar of the hitters in terms of sustained excellence. The not as good news is that there's a lot of HOVG types on the list too, though they all have much shorter careers. So you've got a long-career guy who's not quite as good as Mel Ott but got a career longer than Sunny Jim. Looks good to me.
   48. Chris Cobb Posted: March 02, 2005 at 05:39 AM (#1174138)
Since I've seen the new numbers on Beckwith, I've been thinking "Harry Heilmann" as a hitting comp.

It should be remembered in comparing Suttles to Ott that Ott had excellent plate discipline, where, as far as we know, Suttles did not. I believe that as a power hitter he was probably a match for Ott (and perhaps even a match for Jimmy Foxx).
   49. Howie Menckel Posted: March 02, 2005 at 06:00 AM (#1174151)
Like everyone else here, I greatly admire Chris Cobb's work.
I will say, though, that I almost get the feeling that some voters will rank Suttles and others almost solely on what Chris does. And I greatly doubt he wants that.
I suspect he hardly minds that I happen to be a little skeptical on Beckwith even while I've been highly supportive of many Negro Leaguers. Chris is supplying very valuable grist for the mill.

But none of us here wants to BE the mill.
   50. DavidFoss Posted: March 02, 2005 at 06:58 AM (#1174256)
It should be remembered in comparing Suttles to Ott that Ott had excellent plate discipline, where, as far as we know, Suttles did not. I believe that as a power hitter he was probably a match for Ott (and perhaps even a match for Jimmy Foxx).

Well, Foxx led the AL in strikeouts seven times, but he also walked a lot -- semi-intentionally or not.

What kind of equivalent OBP's would we expect from Suttles?
   51. John (You Can Call Me Grandma) Murphy Posted: March 02, 2005 at 03:12 PM (#1174507)
But none of us here wants to BE the mill.

Agreed, but I have enough faith that he'll be closer to the "truth" than I will, so I'll lean on his Negro League MLEs more than my collective wisdom on the subject.
   52. Dr. Chaleeko Posted: March 02, 2005 at 05:28 PM (#1174779)
I agree with you John. And the constructive suggestions that Chris has gotten seem to really be helping all of us understand the process and the outcomes better as well as sharpening CC's results.

I rely heavily on the MLEs and WSMLEs to get a sense of the player, and to suggest where he MIGHT have fallen within the realm of MLB. I think the discussion of Suttles lack of walks and how it would show him as less valuable than Ott is a good example of how we take information from MLEs, see where it falls within what we know about the MLB talent pool, and then each draw our own conclusions from there.

For my part, I had Suttles third in pre-balloting, but seeing the MLEs actually LOWERED my estimation of his value, as it seems to have for numerous voters. That suggets to me that, to paraphrase Howie M., we're taking the grist and running it through a very strong mill indeed, and that the MLEs don't overrule our personal judgements, they inform them.
   53. Chris Cobb Posted: March 02, 2005 at 05:34 PM (#1174797)
What kind of equivalent OBP's would we expect from Suttles?

It's very hard to project this due to lack of data on Suttles' actual walks. He has the "free swinger" reputation, but we have only 1928 data from Gary A. that actually shows walks.

League average walk rate for that year is 7.6 walks per 100 pa; Suttles had 6.9.

He also had a lot of hits, though. The league bb rate in plate appearances not resulting in a hit was .105; Suttles' was .104.

So in our one data slice, he looks pretty much average in terms of walk rates. Given that this is still fairly early in his career, we should expect those to increase somewhat.

Foxx, by comparison, had 17.5 bb/100 pa in 1935 (his age 27 season to compare with Suttles' age 27 season in 1928), in a league with a rate of 9.6 bb/100 pa.

Foxx's bb rate was .245 in a league with a .129 bb rate.

The available evidence suggests that Suttles was not comparable to Foxx in terms of walk rates. Whether it would have been possible in the Negro Leagues to walk like Foxx did is a serious question. Certainly there are indications that other players were walking more frequently than Suttles was.

Stearnes in 1928, for example, had 9.0 walks/100 PA and a bb rate of .127, both substantially better figures than those of Suttles. Stearnes had 31 walks, which was 7th in the league: the league leader had 38. I think, therefore, that it is highly unlikely that the style of play (and the nature of umpiring) in the NeL would have permitted any player to reach a Foxx-like bb rate.

The data, such as they are, do not, it seems to me, support a view of Suttles as wildly undisciplined at the plate. He appears, in 1928, to have taken about his expected share of walks, and that's it. I think if I were to attempt OBP projections for him, I'd peg him to the league-average bb rate.

Stearnes would be projected above that.
   54. DavidFoss Posted: March 02, 2005 at 05:59 PM (#1174875)
Thanks for the sanity check, Chris. With Suttles' power, it would have taken a Kingman-esque level of plate discipline for him not to make my ballot. I'm projecting at least a league average level of plate discipline for him then because of the amount of 'fear factor' semi-intentional walks I would expect. Around 60-70 a year sounds reasonable.

So, its likely that he was slightly lower than Kiner & Killebrew at their peaks because both them had very good plate discipline dominating their leagues' HR races.
   55. Chris Cobb Posted: March 05, 2005 at 06:45 PM (#1182303)
Mule Suttles Projected Win Shares

Version 1.0

Here are my projections for Mule Suttles’ win shares.

BA and SA are the regressed MLE values as published earlier.

Games is my estimate of Suttles games per season. There is no formula, but in general I examined his biography for records of serious injuries, compared his listed games played to records of team games played, and considered aging patterns of similar players.

PA = games * 4.2 . This factor perhaps ought to vary with league offense level, but I’m not ready yet to engage in that level of exactness in these estimates.

BWS – derived by finding the most similar hitter in the majors, adjusting the win shares up or down slightly if the match is not exact, and prorating from that player’s PA to Suttles’. I generally assumed a league-average walk rate to estimate Suttles’ OBP. Flaw Alert: Suttles’ batting win shares alternate between very good and completely ordinary in the 1930s. This is due to the difference in offense levels between the AL and NL. Given the large discrepancies between the leagues, simply alternating between them (which works pretty well for the 1920s, when the leagues are close) does not work well here, and further adjustments of the MLEs are necessary to fit with the differing league levels. In the interests of getting these win shares out before the 1946 election is over, I have gone ahead with them using the old method, but I will have to handle the 1930s differently. Also, Gary A.’s 1934 data showing lower offensive levels in the NeL in 1934 than I had estimated has not yet been accounted for, and more study of offensive levels in the NeL in the 1930s is needed. On the whole, translations for the 1930s need more work, so I’ll be redoing Suttles’ 1931-1940 seasons when I have a better information.

FWS – Based on Suttles’ reputation as a mediocre defender, I have calculated his fielding win shares as if he were a B- first baseman for his career and a C- outfielder, starting him above that level early and dropping him below it late. If you want a major-league image for Suttles as a fielder, I would suggest Willie Stargell. I have Suttles as a better first baseman than Stargell was, because Stargell moved to first late in his career. I see Stargell as a pretty good comp for Suttles overall, actually – tremendous raw power, lots of Ks, average walks, good but not great average, indifferent fielding, some monster seasons that never line up into a huge peak because of injuries. I hadn’t thought of Stargell as a comp for Suttles before I did the win shares, but I think it’s a good one.



Year BA    SA    Games  PA    BWS  FWS  Total Pos
1923 0.253 0.346   75   315   6.3  1.7   8.0  lf
1924 0.299 0.417  154   647  10.4  3.5  13.9  lf
1925 0.371 0.583  154   647  25.7  2.7  28.4  1b
1926 0.384 0.794  154   647  41.6  2.5  44.1  1b
1927 0.368 0.687   51   214  11.5  0.8  12.3  1b
1928 0.315 0.561  120   504  19.3  1.7  21.0  1b
1929 0.297 0.536  145   609  16.7  2.2  18.9  1b
1930 0.331 0.614  150   630  28.5  2.4  30.9  1b
1931 0.282 0.505  148   622  18.8  2.8  21.6  lf
1932 0.290 0.500  140   588  16.6  2.0  18.6  1b
1933 0.278 0.494  145   609  21.0  2.2  23.2  1b
1934 0.270 0.467  140   588  12.3  2.3  14.6  1b
1935 0.273 0.507  154   647  19.3  2.2  21.5  1b
1936 0.302 0.547  145   609  15.0  1.9  16.9  1b
1937 0.289 0.583  140   588  23.2  2.1  25.3  lf
1938 0.255 0.480  140   588  10.5  1.9  12.4  1b
1939 0.249 0.464  130   546  13.0  1.7  14.7  1b
1940 0.232 0.387  100   420   4.5  1.6   6.1  lf
1941 0.147 0.204   35   147   0.0  0.5   0.5  lf
car. 0.295 0.524 2266 10165 314.2 38.7 352.9 

   56. John (You Can Call Me Grandma) Murphy Posted: March 05, 2005 at 06:50 PM (#1182311)
Thanks, Chris! Now I can figure out where Mule belongs on my ballot for next time.
   57. Mark Shirk (jsch) Posted: March 05, 2005 at 07:50 PM (#1182384)
The one flaw in my system, one that I do correct manually, is that a big season can look just as valuable as a series of great seasons, if that makes sense. Anyways, this is the erason I want to ask about Suttles 1926. Was he really, really good or should I spread some of those Win Shares around.

Thank a lot Chris. If these are correct, and I have every reason to think they are acurate, Suttles will most likely jump over Jennings (from #8 to #3) in 1947.
   58. favre Posted: March 05, 2005 at 11:14 PM (#1182805)
I mentioned this on my ballot, but given Chris's research (kudos, Chris, once again), I think Willie Stargell is a pretty good comp for Suttles.

Stargell: .282/.360./529, 370 Career WS in 21 seasons
Suttles: .295/.524, 355 Career WS in 19 seasons

Both played left field and first base, both had a reputation for free swinging. (Stargell had 1,936 career strikeouts to 937 walks). With era adjustments, I would say Pops was better, but I still think the two are pretty comparable.
   59. Chris Cobb Posted: March 10, 2005 at 01:37 AM (#1190873)
I don't have time to post season-by-season breakdowns for Suttles with the revised 1930s projections (which take account of the difference between NeL offensive levels and ML offensive levels and project the NeL to the National League level of offense), but here are career projections for him:

1923-1941

2420 g, 9248 ab, 2787 hits, 915 bb, 4967 tb, .301 ba, .364 obp, .537 sa

For ready reference, here are Wilson's 1922-1938 totals:

2352 g, 8466 ab, 2845 hits, 1413 walks, 3789 total bases, .336 ba, .431 obp, .448 sa

Suttles has 20 points of OPS on Wilson, but Wilson's is OBP-heavy.

Suttles has 70 games on Wilson, but 2 more seasons.

Wilson will have an edge in defensive value. He was adequate at an important defensive position, while Suttles was adequate at the least important defensive positions.
   60. Dr. Chaleeko Posted: March 10, 2005 at 01:53 AM (#1190906)
Chris,

So if Suttles is revised up from .295 to .301 and from .515 to .537, will Beckwith also see increases (though perhaps not quite as substantial since less of his career seems to be in the 30s)?
   61. Chris Cobb Posted: March 10, 2005 at 03:18 AM (#1191055)
Suttles' revision isn't quite that far: the .295 and .515 figures a) don't account for projection into major-league seasons, b) include 1942-44, seasons which I have left off of my estimate of Suttles' likely major-league career, and c) include a few small errors in data entry and regression that I caught after I published the original Suttles' figures, which weren't large enough to justify reposting about (2% raise in career slugging or something like that but which do contribute to the size of the change now.

Suttles' 1923-41 projected MLE figures have changed from .298/.524 to .301/.537.

I'll do new projections for Beckwith, but they will not change nearly so much. Beckwith only has a couple of big seasons in the 1930s, and he was in the East in 1930, which had a level of offense equal to the NL in its highest-octane year. The West, where Suttles spent most of that season, was 10% below the NL. That's a considerable boost for Suttles in a big year for him.
   62. Dr. Chaleeko Posted: March 10, 2005 at 02:49 PM (#1191577)
Chris C,

In the immortal words of Inspector Columbo... "Just one more question, sir."

How does the recalibration of Suttles's MLEs and the change to a single-league context effect his MLE WS projections?

Thanks, as always!
   63. Chris Cobb Posted: March 10, 2005 at 03:27 PM (#1191615)
How does the recalibration of Suttles's MLEs and the change to a single-league context effect his MLE WS projections?


That is a good question. The general answer is: they will go up, mostly in his even-numbered seasons in the 1930s, when he was matched to AL players in a much higher offensive environment. But there will be some changes in other seasons as well.

Because the process of determining win shares from the projections is not automated (I do it by finding the nearest match among ML players for a given season and adjusting from there), recalculating win shares takes a _lot_ longer than making changes in the automated calculations that do the MLEs.

I did one season, just to provide an example. In 1934, Suttles version 1 had a BA of .269 and an SA of .467. His closest comp in the AL was E. Coleman. With adjustments for playing time and offensive differences, I estimated Suttles at 12.3 bws. In 1934, Suttles v2 has a .284 BA and a .496 SA. HIs closest comp in the NL is Chick Hafey. With adjustments, I estimate him at 15.0 bws, a gain of 2.7.

Looking at the raw numbers, I could imagine a gain of up to 5 win shares in some season, but this looks typical: I'd guess that the revisions will gain Suttles somewhere between 10 and 15 win shares for his career.
   64. Dr. Chaleeko Posted: March 10, 2005 at 03:45 PM (#1191640)
Chris,

Wow, that's interesting stuff. I thought you used a single comparable career to shape the arc of the WS estimates rather than season-by-season comparables.

Seems like some of the WS Suttles would gain will also raise the profile of his peak/prime estimates as well, smoothing out that 44/31 drop-off somewhat.
   65. DavidFoss Posted: March 10, 2005 at 03:48 PM (#1191646)
I did one season, just to provide an example. In 1934, Suttles version 1 had a BA of .269 and an SA of .467. His closest comp in the AL was E. Coleman. With adjustments for playing time and offensive differences, I estimated Suttles at 12.3 bws. In 1934, Suttles v2 has a .284 BA and a .496 SA. HIs closest comp in the NL is Chick Hafey. With adjustments, I estimate him at 15.0 bws, a gain of 2.7.

I'm not sure I follow what happened here.

I understand that .269/.467 is worth more in the 1934-NL than it is in 1934-AL.

But, why would his raw projection go *up* to .284/.496 when switching from using only the NL in the 1930s?

I mean, if Arky Vaughn's career line were really the NL-MLE's of a NeL-er, then his AL-MLE's (and thus also his MLB-MLE's) would be quite a bit higher, no?
   66. Chris Cobb Posted: March 10, 2005 at 04:12 PM (#1191695)
David,

The other factor at play here is that the Negro Leagues were generally _below_ even the National League in offensive level during the 1930s.

In 1934, as Gary A.'s dat shows, the NeL EW-league average was .261, compared to NL .279 and AL .279.

For this season in particular, Suttles' gain is all from the offensive-level conversion from NeL to NL.

When I first did the projections for Suttles, I didn't take account of the frequent difference in offensive levels between the NL and the AL and I didn't have any data on offensive levels in the NeL, so I just applied the straight competition-level discount of .90/.82 BA/SA and found AL and NL matches for Suttles in alternating years. As I began to look more closely at the 1930s data from the majors and to get 1930s data for the NeL, it was clear that two corrections to the process were needed.
   67. DavidFoss Posted: March 10, 2005 at 04:29 PM (#1191719)
The other factor at play here is that the Negro Leagues were generally _below_ even the National League in offensive level during the 1930s.

OK... make sense now. Thanks!
   68. Brent Posted: March 11, 2005 at 03:52 AM (#1193134)
Chris Cobb wrote:

Because the process of determining win shares from the projections is not automated (I do it by finding the nearest match among ML players for a given season and adjusting from there), recalculating win shares takes a _lot_ longer than making changes in the automated calculations that do the MLEs.

Chris, I've always admired your work on MLEs - you clearly have gained the respect of the voters for the care with which they are put together. But I'll suggest that when you don't have time to do what appears to be a very labor-intensive calculation to obtain win shares, you might consider using some modification of the short-form calculations I've used with the minor leaguers. I think you've observed that they come out pretty close to the results of your more sophisticated method (with the exception of a problem with high values, which I think I have now largely fixed). The nice thing is that once I've translated the batting statistics, the WS calculation is pretty much an automated set of spreadsheet formulas.

Just a suggestion for the weeks when you may have a few other things going on besides HoM. :-)
   69. Chris Cobb Posted: March 11, 2005 at 05:07 AM (#1193293)
Brent,

As I understand it, even the short form of Win Shares requires some basic, reliable team data that is simply not available for the Negro Leagues at present.

I don't have a thorough knowledge of your modifications of the win-shares short form, but can it be used in the absence of any team data?

If so, I'd be very interested.
   70. Chris Cobb Posted: March 25, 2005 at 02:40 AM (#1215485)
Mule Suttles’ updated MLEs

I’ve calculated new MLEs for Suttles, using revised offensive level adjustments for the 1930s and handling walks on a systematic seasonal basis like the one used for Bell.


Year Team G    PA   BB Hits   TB  BA   OBP  SA
1923 Bir  75   315  22   74   99 .253 .306 .346
1924     154   647  47  179  245 .299 .350 .417
1925     154   647  44  226  343 .374 .417 .583
1926 Stl 154   647  45  234  467 .389 .431 .794
1927      51   214  16   73  134 .368 .415 .687
1928     120   504  41  147  259 .317 .373 .565
1929     145   609  52  167  300 .300 .359 .542
1930 *   150   630  52  203  382 .351 .404 .666
1931 **  148   622  57  167  299 .296 .361 .528
1932 *** 140   588  55  162  281 .304 .369 .525
1933 Chi 145   609  61  150  279 .274 .347 .503
1934     140   588  59  150  265 .284 .356 .496
1935     154   647  66  164  309 .282 .355 .525
1936 Nwk 145   609  62  172  319 .315 .385 .575
1937     140   588  62  156  318 .296 .370 .594
1938     140   588  65  139  261 .265 .346 .488
1939     130   546  58  121  233 .247 .328 .468
1940     100   420  44   87  148 .232 .313 .388
1941 NY   35   147  17   19   27 .147 .246 .205
tot     2420 10163 924 2791 4967 .302 .366 .538



*Stl, Bal
**Stl, Was
*** Det, Was

These MLEs are set to major-league seasonal norms for 1923-29, to National League norms for 1930-41.
   71. DavidFoss Posted: March 25, 2005 at 03:11 AM (#1215538)
MULE SUTTLES

-First you have Year, Team(s), PA.
-Second you have Chris's MLE's
-Third, in parentheses, you have pitchers-removed offense context. MLB for the 20s, then NL
-Fourth, you have AVG+/OBP+/SLG+
-Lastly, is the OPS+

1923  Bir 315  0.253/0.305/0.346   (0.292/0.356/0.405)    87/ 86/ 85     71
1924  Bir 647  0.299/0.349/0.417   (0.294/0.356/0.406)   102/ 98/103    101
1925  Bir 647  0.374/0.417/0.583   (0.300/0.364/0.425)   125/115/137    152
1926  Stl 647  0.389/0.431/0.794   (0.289/0.355/0.402)   135/121/198    219
1927  Stl 214  0.368/0.416/0.687   (0.292/0.355/0.406)   126/117/169    186
1928  Stl 504  0.317/0.373/0.565   (0.290/0.355/0.412)   109/105/137    142
1929  Stl 609  0.300/0.360/0.542   (0.298/0.363/0.432)   101/ 99/125    125
1930  S-B 630  0.351/0.405/0.666   (0.312/0.370/0.464)   113/109/144    153
1931  S-W 622  0.296/0.360/0.528   (0.285/0.344/0.403)   104/105/131    136
1932  D-W 588  0.304/0.369/0.525   (0.284/0.337/0.412)   107/110/127    137
1933  Chi 609  0.274/0.346/0.503   (0.274/0.327/0.376)   100/106/134    140
1934  Chi 588  0.284/0.355/0.496   (0.287/0.342/0.408)    99/104/122    125
1935  Chi 647  0.282/0.355/0.525   (0.286/0.341/0.407)    99/104/129    133
1936  Nwk 609  0.315/0.384/0.575   (0.286/0.345/0.400)   110/111/144    155
1937  Nwk 588  0.296/0.371/0.594   (0.280/0.342/0.397)   106/108/150    158
1938  Nwk 588  0.265/0.347/0.488   (0.275/0.339/0.391)    96/102/125    127
1939  Nwk 546  0.247/0.328/0.468   (0.280/0.346/0.401)    88/ 95/117    111
1940  Nwk 420  0.232/0.312/0.388   (0.272/0.337/0.391)    85/ 93/ 99     92
1941   NY 147  0.147/0.245/0.205   (0.266/0.337/0.375)    55/ 73/ 55     27
   72. DavidFoss Posted: March 25, 2005 at 03:18 AM (#1215544)
That 219 is very impressive. The 186 is good but its a short season. Then there are 12 seasons between 125 and 158.

Looks like the OBP values on my chart are rounded differently (plus/minus 0.001). I still have a (H+BB)/PA calculation in my script left over from when Chris was just reporting those values and not OBP. Probably doesn't matter, but I'll fix that for next time and just use Chris's values. With translated stats, I'm not sure which values are rounded.

Anyways, thanks Chris!
   73. Dr. Chaleeko Posted: March 25, 2005 at 04:00 AM (#1215621)
DavidFoss,

Nice work! What do you have Mule's career OPS+ working out to?
   74. John (You Can Call Me Grandma) Murphy Posted: March 25, 2005 at 04:25 AM (#1215648)
Nice work! What do you have Mule's career OPS+ working out to?

That was also my question, Doc. I keep reading about comparisons to Stargell, but unless I'm wrong, Suttles' OPS+ won't be close to Stargell's 147+.
   75. John (You Can Call Me Grandma) Murphy Posted: March 25, 2005 at 04:26 AM (#1215650)
BTW, nice work, David!
   76. Chris Cobb Posted: March 25, 2005 at 05:50 AM (#1215726)
My math may be off, but using David's seasonal figures I get a career OPS+ of 137 for Suttles.
   77. Brent Posted: March 25, 2005 at 06:41 AM (#1215774)
Here are a few players with long careers and career OPS+ near 137:

R Jackson: 139
Cash: 139
J Clark: 137
McGriff: 134
B Powell: 134
B Williams: 132
Palmeiro: 132
   78. andrew siegel Posted: March 25, 2005 at 06:44 AM (#1215776)
Among his contemporaries, Chuck Klein and Ken Williams are two guys with 137 OPS+'s. Except that Suttles projects to 40% more plate appearances than Klein and 80% more than Williams. By my calculations, Suttles is somewhere between 130 and 170 all-time and safely in the HOM.
   79. DavidFoss Posted: March 25, 2005 at 07:02 AM (#1215794)
My math may be off, but using David's seasonal figures I get a career OPS+ of 137 for Suttles.

Yup... assuming we use all the years listed. (Reasonable for Suttles, but maybe not for Wilson or Bell?)

Counting stats (+/- 2 for rounding?)
10165 PA
9240 AB
2790 H
4968 TB

Percentages
Suttles -- 0.302/0.365/0.538
Context --(0.288/0.348/0.408)
Plusses -- 105/105/132

OPS+ = 137
   80. John (You Can Call Me Grandma) Murphy Posted: March 25, 2005 at 03:41 PM (#1216119)
Here are a few players with long careers and career OPS+ near 137:

R Jackson: 139
Cash: 139
J Clark: 137
McGriff: 134
B Powell: 134
B Williams: 132
Palmeiro: 132


Yes to Reggie, probably no to Cash, not sure about Clark, not sure about McGriff, no to Powell, probably yes to Bernie, probably yes to Palmiero.
   81. Brent Posted: March 25, 2005 at 10:52 PM (#1216768)
B Williams: 132

I'm sorry, that's Billy Williams, not Bernie (whose OPS+ is currently 130). I should have spelled it out.
   82. karlmagnus Posted: March 25, 2005 at 11:10 PM (#1216806)
Canseco 131?
   83. Michael Bass Posted: March 25, 2005 at 11:14 PM (#1216813)
For my PHOM...

Reggie - Yes
Cash - Probably yes (he certainly will be well higher than, to take an example, Beckley)
Clark - No
McGriff - Probably no
Powell - No
Williams - Yes
Palmeiro - Probably yes

I have to say, that's a wide divergence of value, and shows the downside to ranking players, even at similar positions, just by career OPS+.

As for Suttles, he has the peak that the "no"s on my list don't have, so I'm comfortable with him where he is on my ballot.
   84. John (You Can Call Me Grandma) Murphy Posted: March 25, 2005 at 11:22 PM (#1216832)
I have to say, that's a wide divergence of value, and shows the downside to ranking players, even at similar positions, just by career OPS+.

I agree, Michael.

As for Billy Williams, I would say yes at this time. Canseco gets a no from me.
   85. DavidFoss Posted: March 25, 2005 at 11:27 PM (#1216843)
Lots of non-contemporaries on the lists above. I really have no idea how I will vote for future guys until I see who else is on the ballot when they become eligible.

Andrew's suggestion of CKlein & KWilliams look like decent comps at 137, but Suttles of course played longer.

O'Doul 143
Jack Fournier 142
Babe Herman 140
Jeff Heath 139
Wally Berger 138
Bob Johnson 138
Dolph Camilli 136
Bill Terry 136

Suttles would appear to have longevity to rank over all this group (plus Klein & KWilliams), but maybe these are better comps?

Also:

Sherry Magee 137
Mike Tiernan 138

Unfornuately the bb-ref leaderboard cuts off at 136. :-)
   86. DavidFoss Posted: March 25, 2005 at 11:30 PM (#1216847)
maybe these are better comps?

I mean, project them to Suttles's career length and they become better comps... obviously, some of these careers are too short. Looking for just a general idea of contemporaries with similar offensive impact.
   87. Michael Bass Posted: March 25, 2005 at 11:36 PM (#1216859)
Canseco gets a no from me.

Agreed on that. 131 OPS+ or not, he's not even a marginal candidate. Horrid defense, always hurt, only one big year.
   88. Brent Posted: March 26, 2005 at 02:01 AM (#1217064)
maybe these are better comps?

I'd say no, at least if by "comp" we're talking about the type of hitter he was. Suttles had home run power, probably projects to 30-35 home runs per year, but (at least according to Chris's MLEs) not an unusually high batting average or OBP. Most of the contemporary players on David's list had high batting averages and maybe 20-HR/yr power. (Berger is an exception with 29-HR/yr and probably makes a pretty good comp except for career length. Klein was able to reach 30+ HRs while playing in the Baker Bowl.) Very few pre-WWII MLB players averaged 30-35 home runs; those who did, like Gehrig and Mize, were much better hitters than Suttles overall. Suttles appears to have been a type of hitter that didn't become common in MLB until after 1950.
   89. Mark Shirk (jsch) Posted: March 26, 2005 at 04:21 PM (#1217873)
What Brent said, that Suttles was a very unique player in pre WWII baseball, brings up a question. Since Suttles is so unique would playing in the Major Leagues have forced him to become a different player than he was in the NeL? Would they have forced him to change his swing or undervalued him, etc.?

This is purely theoretical, or course, and something that doesn't affect his ranking, but I just wonder.
   90. Gary A Posted: March 28, 2005 at 10:28 PM (#1221430)
I forgot I had this...it's a little different from what Holway's got.

1923 Mule Suttles (rookie season)
Birmingham Black Barons
G-34 (team 42)
AB-127
H-36
D-4
T-3
HR-1
R-20
RBI-16
W-6
SB-4
AVE-.283
OBA-.313
SLG-.386

Compare Turkey Stearnes in 1923, also his rookie season (they were both born in 1901, only two months apart). Even considering park differences, it's clear that Stearnes was already a complete major-league player and immediately stepped in as one of the best hitters in the league, whereas Suttles had some adjustments to make.

1923 Turkey Stearnes
Detroit Stars
G-69 (team 71)
AB-279
H-101 (5th)
D-18* (1st)
T-14
HR-17 (3rd)
R-70
RBI-85 (3rd)
W-17
SB-2
AVE-.362 (5th)
OBA-.398 (7th)
SLG-.710 (2nd)
   91. Gary A Posted: March 28, 2005 at 10:29 PM (#1221431)
Sorry--Stearnes led the league in TRIPLES, not doubles.
   92. KJOK Posted: June 30, 2005 at 02:05 AM (#1440556)
This is from John Holway, Feb. 2004 - thought it might have some bearing for some people:



In 1925 Suttles played in the most hitter-hostile park in America, Rickwood Field in Birmingham. He still hit .400 with power.

Players I've talked to, like Bell, say the LF pole in St Louis was about 250 feet. But balls hit down the line were ground-rule doubles. The fence receded quickly to a very distant CF, possibly even 600 feet, and remained relatively deep in RF.

How many homers did Mule lose on clouts pulled to left? How many did he lose on clouts hit straight-away to center? And why did the pitchers let him pull?

(The same could be asked of Mel Ott's opponents in the Polo Grounds. I've seen Ott uppercut seven-iron golf shots into the upper deck, which had an overhang making it even closer than the 257 -foot foul pole on the ground.)

From St Louis, Suttles spent three years in Chicago with Comiskey as a home park. It was one of the tougher hitters'parks in the majors.

He ended his career in Newark, which was shaped like the Polo Grounds, with short foul lines (ask Bobby Thomson) and unreachable territory from the left-center corner to that in right-center.

In KC Bullet Rogan was two hitters -- one in the Monarchs' first park, and another after they moved to Muelebach Field, which the Athletics inherited.
   93. sunnyday2 Posted: July 16, 2005 at 06:10 PM (#1476936)
Bump. First from #2 and #3 above.

Posted by Chris Cobb on February 18, 2005 at 10:07 PM (#1153567)
Mule Suttles Data, mostly from Holway

Teams: 23-25 Bir Black Barons, 26-31 Stl Stars, 32 Det/Was, 33-35 Chi Am Giants, 36-40 Newark Eagles, 41 NY Black Yankees, 42-44 Newark Eagles

Park Effects: (as far as I know) Birmingham played in a very good pitcher’s park, St. Louis played in an all-time great hitter’s park, Chi Am Giants an all-time great pitchers’ park, Newark in a very good hitter’s park.

Seasonal Data from Holway

1923 .243 for Bir Black Barons; lf
1924 .317 for Bir; lf
1925 .428 for Bir; 1st in ba, 14 hr (3rd), 36 hr/550 ab (3rd); 1b
1926 .498 for Stl; ba 1st, 27 hr (1st), 70 hr/550 (1st), 27 2b (1st), 21 3b (1st), 11 sb; 1b; all-star, MVP
1927 .476 for Stl; (1/4 season, was beaned & missed the rest of the year) 53 hr/550 (1st); 1b
1928 .372 for Stl; ba 3rd, 19 hr (3rd), 38 hr/550 (4th), 9 3b (1st); 1b, all-star
9-27 in playoff vs. Chi Am Giants
1929 .351 for Stl; 20 hr (2nd), 35 hr/550 (4th), 29 2b (1st); 1b, all-star
3-13 in World Series vs. Homestead
7-20 vs. major-league competition
55-175 in Cuban Play
1930 .422 for Stl.; ba 1st, 12 hr (2nd), 43 hr/550 (1st), 9 3b (3rd); ut, all-star (as dh)
.389 for Bal Black Sox; 10 hr (1st), 73/hr 550 (1st) [57/550 combined], 7 2b (4th); 1b
1931 .349 for Stl; ba 4th; lf
2-4 vs. major-league competition
1932 .315 for Det, .322 for Was; 9 hr (3rd), 31 hr/550 (1st), 12 2b (3rd); 1b, all-star
1933 .254 for Chi Am Giants; 5 sb (4th); 1b
1934 .254 for Chi Am Giants; 12 hr (3rd), 32 hr/550 (4th); 1b
3-15 in playoff vs. Phi Stars
1935 .263 for Chi Am Giants; 13 hr (2nd), 40 hr/550 (1st); 1b
5-12 vs. major-league competition
1936 .365 for Newark Eagles; ba 4th, 15 hr (2nd), 57 hr/550 (3rd); 1b, all-star
1937 .345 for Newark Eagles; 16 hr (2nd), 80 hr/550 (1st); lf
1938 .282 for Newark Eagles; 14 hr (1st), 51 hr/550 (1st); 1b
1939 .282 for Newark Eagles; 12 hr (2nd), 46 hr/550 (3rd), 6 2b (5th), 2 3b (3rd); 1b
4-15 in playoffs
0-4 vs. major-league competition
31-142 in Cuban Play
1940 .262 for Newark Eagles; 8 hr (4th), 26 hr/550 (5th), 11 2b (5th), 3 sb (1st); lf
1941 .222 for NY Black Yankees; lf
1942 .385 for Newark Eagles; ut
1943 no data, manager of Newark Eagles
1944 .250 for Newark Eagles; ut and manager

Career Batting
1103-3230, .341
237 hr, 40/550 ab
58-170, .341 vs. major-league pitching
.329 mean avg. for 21 seasons with data

Career from MacMillan 8th ed.
830 g, 2962 ab, 1001 hits, 167 2b, 63 3b, 183 hr, .338 ba

Career from MacMillan 10th ed.
870 g, 3077 ab, 1011 hits, 171 2b, 65 3b, 190 hr, .329 ba

Black/Gray Ink
56/145

Black Ink: #2 behind Gibson. Both of these players benefit by the heavy weighting of available black ink information towards home runs.
Gray Ink: #4 behind Stearnes, Charleston, and Gibson

Career Fielding

1b 25-30, 32-36, 38-39
lf 23, 24, 31, 37, 40-41
ut/ph 42-44

Brief Comment: Suttles was, by all accounts, a tremendous power hitter. The difficult question here is, How much was he helped by his parks? I don’t think that’s a question that could move him across the in-out line, but it’s a question could move him ahead or behind Turkey Stearnes (my uninformed belief is behind), and that has bearing on little matters like, would he have hit 450 or 550 home runs in the major leagues.
Posted by Chris Cobb on February 18, 2005 at 10:13 PM (#1153572)
Sorry, forgot to include slugging percentages derived from the MacMillan data.

MacMillan 8th .622

MacMillan 10th .612

Then from #43. (Chris, is this the latest Beckwith data or is the data on his (Beckwith's) thread more current???)

John Beckwith Revised MLEs

Year MLE BA MLE SA Reg. BA Reg. SA
1919 0.188 0.233 0.250 0.368
1920 0.215 0.260 0.288 0.377
1921 0.387 0.550 0.359 0.527
1922 0.336 0.469 0.337 0.497
1923 0.321 0.568 0.332 0.571
1924 0.370 0.561 0.357 0.571
1925 0.376 0.663 0.366 0.610
1926 0.335 0.567 0.343 0.552
1927 0.333 0.455 0.337 0.493
1928 0.312 0.411 0.330 0.481
1929 0.342 0.565 0.338 0.538
1930 0.403 0.663 0.361 0.594
1931 0.308 0.558 0.323 0.564
1932 0.300 0.541 0.319 0.549
1933 0.400 0.380 0.307 0.479
1934 0.141 0.174 0.200 0.218
1935 0.000 0.000 0.083 0.136

car. 0.337 0.534 0.339 0.537

Then from #70:

I’ve calculated new MLEs for Suttles, using revised offensive level adjustments for the 1930s and handling walks on a systematic seasonal basis like the one used for Bell.


Year Team G PA BB Hits TB BA OBP SA
1923 Bir 75 315 22 74 99 .253 .306 .346
1924 154 647 47 179 245 .299 .350 .417
1925 154 647 44 226 343 .374 .417 .583
1926 Stl 154 647 45 234 467 .389 .431 .794
1927 51 214 16 73 134 .368 .415 .687
1928 120 504 41 147 259 .317 .373 .565
1929 145 609 52 167 300 .300 .359 .542
1930 * 150 630 52 203 382 .351 .404 .666
1931 ** 148 622 57 167 299 .296 .361 .528
1932 *** 140 588 55 162 281 .304 .369 .525
1933 Chi 145 609 61 150 279 .274 .347 .503
1934 140 588 59 150 265 .284 .356 .496
1935 154 647 66 164 309 .282 .355 .525
1936 Nwk 145 609 62 172 319 .315 .385 .575
1937 140 588 62 156 318 .296 .370 .594
1938 140 588 65 139 261 .265 .346 .488
1939 130 546 58 121 233 .247 .328 .468
1940 100 420 44 87 148 .232 .313 .388
1941 NY 35 147 17 19 27 .147 .246 .205
tot 2420 10163 924 2791 4967 .302 .366 .538

Then from David Foss in #71 and #79

-First you have Year, Team(s), PA.
-Second you have Chris's MLE's
-Third, in parentheses, you have pitchers-removed offense context. MLB for the 20s, then NL
-Fourth, you have AVG+/OBP+/SLG+
-Lastly, is the OPS+

1923 Bir 315 0.253/0.305/0.346 (0.292/0.356/0.405) 87/ 86/ 85 71
1924 Bir 647 0.299/0.349/0.417 (0.294/0.356/0.406) 102/ 98/103 101
1925 Bir 647 0.374/0.417/0.583 (0.300/0.364/0.425) 125/115/137 152
1926 Stl 647 0.389/0.431/0.794 (0.289/0.355/0.402) 135/121/198 219
1927 Stl 214 0.368/0.416/0.687 (0.292/0.355/0.406) 126/117/169 186
1928 Stl 504 0.317/0.373/0.565 (0.290/0.355/0.412) 109/105/137 142
1929 Stl 609 0.300/0.360/0.542 (0.298/0.363/0.432) 101/ 99/125 125
1930 S-B 630 0.351/0.405/0.666 (0.312/0.370/0.464) 113/109/144 153
1931 S-W 622 0.296/0.360/0.528 (0.285/0.344/0.403) 104/105/131 136
1932 D-W 588 0.304/0.369/0.525 (0.284/0.337/0.412) 107/110/127 137
1933 Chi 609 0.274/0.346/0.503 (0.274/0.327/0.376) 100/106/134 140
1934 Chi 588 0.284/0.355/0.496 (0.287/0.342/0.408) 99/104/122 125
1935 Chi 647 0.282/0.355/0.525 (0.286/0.341/0.407) 99/104/129 133
1936 Nwk 609 0.315/0.384/0.575 (0.286/0.345/0.400) 110/111/144 155
1937 Nwk 588 0.296/0.371/0.594 (0.280/0.342/0.397) 106/108/150 158
1938 Nwk 588 0.265/0.347/0.488 (0.275/0.339/0.391) 96/102/125 127
1939 Nwk 546 0.247/0.328/0.468 (0.280/0.346/0.401) 88/ 95/117 111
1940 Nwk 420 0.232/0.312/0.388 (0.272/0.337/0.391) 85/ 93/ 99 92
1941 NY 147 0.147/0.245/0.205 (0.266/0.337/0.375) 55/ 73/ 55 27

Counting stats (+/- 2 for rounding?)
10165 PA
9240 AB
2790 H
4968 TB

Percentages
Suttles -- 0.302/0.365/0.538
Context --(0.288/0.348/0.408)
Plusses -- 105/105/132

OPS+ = 137

Finally, WS from Chris #55

Year BA SA Games PA BWS FWS Total Pos
1923 0.253 0.346 75 315 6.3 1.7 8.0 lf
1924 0.299 0.417 154 647 10.4 3.5 13.9 lf
1925 0.371 0.583 154 647 25.7 2.7 28.4 1b
1926 0.384 0.794 154 647 41.6 2.5 44.1 1b
1927 0.368 0.687 51 214 11.5 0.8 12.3 1b
1928 0.315 0.561 120 504 19.3 1.7 21.0 1b
1929 0.297 0.536 145 609 16.7 2.2 18.9 1b
1930 0.331 0.614 150 630 28.5 2.4 30.9 1b
1931 0.282 0.505 148 622 18.8 2.8 21.6 lf
1932 0.290 0.500 140 588 16.6 2.0 18.6 1b
1933 0.278 0.494 145 609 21.0 2.2 23.2 1b
1934 0.270 0.467 140 588 12.3 2.3 14.6 1b
1935 0.273 0.507 154 647 19.3 2.2 21.5 1b
1936 0.302 0.547 145 609 15.0 1.9 16.9 1b
1937 0.289 0.583 140 588 23.2 2.1 25.3 lf
1938 0.255 0.480 140 588 10.5 1.9 12.4 1b
1939 0.249 0.464 130 546 13.0 1.7 14.7 1b
1940 0.232 0.387 100 420 4.5 1.6 6.1 lf
1941 0.147 0.204 35 147 0.0 0.5 0.5 lf
car. 0.295 0.524 2266 10165 314.2 38.7 352.9
   94. Paul Wendt Posted: April 12, 2009 at 06:39 PM (#3135135)
not Mule Suttles but Park Factors :-(

Let me return to so-called "park factors" and "park" adjustments to playing records, discussed at #20-32 especially by jimd. I wonder whether it is now possible to do better than three years ago --no, it's four years! Question one begins in that direction. (I will also write to Gary A whose 2005 phone number is out of service.)

I have looked at the section of Win Shares by Bill James, beginning page 86, cited by jimd #24. I agree with jimd concerning what James overlooks (mistakes that must pervade the win shares rating system with small biases). On the other hand I'm not sure how he has used the Wendt correction in combination with the calculation by Bill James (jimd #32). Points 2a and 2b give some explanation of the James oversights, and otherwise leave this hanging. Maybe jimd will reply, but I welcome anyone who sees through the rounding error to the calculations that link #21 and #32.

1.
How many league-seasons do we have now, with home/away splits for all teams? We need at least joint measures of runs scored and playing time.

The measure of playing time may be games. If we do have runs scored and runs allowed separately then it is better to have innings batted and innings pitched separately (innings and outings).

2.
a.
Bill James makes an important theoretical mistake overlooking what jimd calls the case of extreme raw park factors; that is, neglecting the correction that jimd attributes to me (#24,25). In practice for eight-team leagues, that oversight causes him to overestimate park factors and park adjustments by one-seventh, or in the proportion 8 to 7. For example, given the record of a ballpark that truly boosts run-scoring by 14% (park factor 114), he tends to estimate 16% (116).

(That example uses a linear approximation that jimd does not use in #24. Probably I didn't use the linear approximation where jimd learned the point from me. I'm not sure just what jimd uses in #32 because there is a lot of rounding error in those numbers. Some of those numbers look wrong to me, even granting slack for rounding error.)

b.
Bill James also overlooks that a team's batters do not face its own pitchers and vice versa. That is another cause of bias in his use of raw park factors. As jimd says (#24), adjustment for teammates "really only matters for extreme offensive or defensive teams." That's true. That's true. Certainly the mistake just explained (2a) causes a greater bias. In order to get the same magnitude of bias illustrated above, 2%, the team's own offense must be about 14% better than 8-team league average.

(Overlooking a team offense 14% better than average would be the source of 2% bias in "pitcher park factor" and 2% overestimate of the quality of team defense. Overlooking team defense better than average would cause bias in "batter park factor". Bill James does not distinguish the two park factors (a misnomer) and it would be unusual to distinguish them while missing the point.)
   95. Paul Wendt Posted: April 12, 2009 at 08:10 PM (#3135316)
On the other hand I'm not sure how he has used the Wendt correction in combination with the calculation by Bill James (jimd #32). Points 2a and 2b give some explanation of the James oversights, and otherwise leave this hanging. Maybe jimd will reply, but I welcome anyone who sees through the rounding error to the calculations that link #21 and #32.

Now I reproduce the column "After Wendt" (with rounding error), so I think I know that calculation and agree with it (granting the rounding error). I don't reproduce the column "After Bal-Sched". For example,

1921: 41/42, 54 57 80
From 54.00 the result of the first stage (Wendt) is 57.29, or
54.00 => 57.29.
Then the second stage (Bal Sched) is either
57.29 => 81.69
or
57.00 => 81.57 (with intermediate rounding).
Either way, the two-stage result with final rounding is 82.

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
Phil Birnbaum
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 0.9318 seconds
49 querie(s) executed