Baseball for the Thinking Fan

Login | Register | Feedback

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

Wednesday, December 22, 2004

Gavvy Cravath

“Cactus Gavvy” gets his own page. Go to town, Phillybooster!

John (You Can Call Me Grandma) Murphy Posted: December 22, 2004 at 02:36 AM | 291 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 3 pages  < 1 2 3 > 
   101. DavidFoss Posted: April 22, 2005 at 02:42 AM (#1278586)
What were the league averages at that time? Pretty low IIRC, no time to look it up now. I'd expect the league hit less than .250 during the early 1900s.

These are the pitching-removed MLB context numbers for the years in question:

1903 .270/.327/.359
1904 .255/.310/.332
1905 .256/.317/.335
1906 .254/.316/.325
1907 .253/.314/.319
1908 .247/.306/.315
1909 .253/.317/.323
1910 .257/.327/.336
1911 .275/.347/.370

These are the numbers I would have normalized against if I had gotten MLE's from Chris Cobb. Brent has already calculated OPS+'s against what appears to be a slightly different context (AL?,NL?, multiple year? -- he was the one who showed me how to do this originally so he knows what he's doing :-)).
   102. Brent Posted: April 22, 2005 at 03:21 AM (#1278672)
Thank you for questioning my numbers, Dr. Chaleeko. My walk projections did not include any PCL-specific information on walks. Instead, I was trying to pick up two factors - first, a tendency to draw more walks as players age; this is based on data that Tangotiger presents on his Web site. The other are trends in major league walk rates - there were fewer walks per AB during the aughts than the teens, so I didn't want to project the higher walk rate of Gravath's post-1912 career to the earlier period.

However, in implementing it, I see that I have overdone it. The biggest problem is that most of the increase in major league walk rates took place in the American League, which is really irrelevant to Cravath's post-1912 career in the NL. I will re-do these and re-post later this evening.

Another thing I would like to emphasize is that these MLE projections shouldn't be treated as facts. It is more appropriate to think of them as a way to organize our factual information from minor league records, but they also require a mixture of assumptions and opinions. For example, these MLEs reflect my opinion that the quality of play in the PCL from 1903-07 was closer to double AA than triple AAA quality, but that is only an opinion. I have to say after looking at the resulting MLEs it makes me wonder, because the PCL years really seem too low relative to the rest of his career. It's hard to think of another player who had many seasons of 130+ OPS+ who did not show signs of greatness before age 26. There is enough uncertainty about quality of competition that it might be reasonable to add 10 points of OPS+ to the PCL years. MLEs should be treated as a starting place for discussion, not as facts.

Another note - the writer who said he had little defensive skill or speed in his PCL years may have been inappropriately projecting his major league record, which mostly represented seasons past the age of 31, to his younger playing record. In the PCL Cravath regularly stole 30 to 50 bases each season and his fielding statistics look pretty good.

I'll be back in a few minutes with corrected MLEs.
   103. Brent Posted: April 22, 2005 at 04:30 AM (#1278788)
Here are corrected MLEs for Cravath. His walks and OBPs are now higher than my last set of MLEs. His projected BB/AB ratios for ages 22-23 are still quite a bit lower than his actual major league walk ratios, however, which is mostly because Tangotiger’s data show that major league players typically improve their walk ratios during these ages. If you disagree, you are welcome to substitute your own assumptions.

I’ve also changed my assumptions for playing time for 1907, showing his actual PCL playing time prorated to a 154-game schedule, rather than reducing it by 10 percent as I had done in the last round. These projections now show his 1906 PCL season as about major league average for a right fielder (especially if you assume that he had fairly good speed and fielding range at that time). And certainly there were several regular major league corner outfielders with OPS+ in the 97-101 range, as Cravath is projected for 1904-05. I’d say that by the end of 1904 he had demonstrated ability to play above major league replacement level, and by the end of 1906 at a major league average level (which is actually pretty good). I still think it's unusual that the first clear sign that he will be a great hitter doesn't come until 1907 at age 26.
Year  Lg Age   G  AB   R   H 2B 3B HR  BB  AVG  OBP  SLG OPS+
1903 PCL  22 104 413  42  99 23  4  3  41 .240 .308 .337   88
1904 PCL  23 101 383  41  90 17  1  5  41 .235 .309 .324   97
1905 PCL  24  98 357  35  85 15  3  4  45 .238 .323 .331  101
1906 PCL  25 133 498  64 120 28  5  4  65 .241 .329 .341  109
1907 PCL  26 138 440  61 119 29  2  6  59 .270 .357 .386  135
1909  AA  28 115 394  53 109 21  5  4  59 .277 .371 .386  136
1910  AA  29 150 567  77 165 35  8 10  97 .291 .395 .434  150
1911  AA  30 153 557 101 178 43  8 20 102 .320 .425 .533  167
1921 PCL  40  83 247  30  67 14  0  9  35 .271 .362 .437  106
1922  AA  41  33  58   6  14  2  0  2   9 .241 .343 .379   87
   104. Carl Goetz Posted: April 22, 2005 at 12:08 PM (#1279080)
Given his low Avg before 1907, I still question whether an ML manager would have given him so much PT. The extra walks improve him in our minds, but would it have done anything for the decision makers then?
   105. Dr. Chaleeko Posted: April 22, 2005 at 12:24 PM (#1279084)
Carl,

This is an interesting point you're raising. Right that an ML decision maker might not have been as impressed by Cravath as we might be, but the question each voter must ask him or herself is whether or not that's important to their consideration of the merit of the player. Just because he wasn't appreciated in his time doesn't mean he wasn't MLB-ready or Meritorious. But again, that's an individual voter's decision.

Also, I'll have WS estimates posted some time this weekend.
   106. Mark Shirk (jsch) Posted: April 22, 2005 at 01:23 PM (#1279122)
I don't think that 'Major League Replacement Level' is the standard that we shoudl be susing to determine if Cravath shold have been in teh Majors. This is an era with a highly inefficient feeder ssytem to MLB clubs (relative to today of course) and I presuem that there were scores of players who could have been Major League replacement level.

What was the MLB average for OPS+ among corner outfielders? 109 still seems low, I would guess that 115 is closer to the average. If the MLB average is close to 109 then maybe I wil give him 1907 credit, but for now I dont' think he deserves 1907 credit (even though he was of obvious MLB quality) because nothign he did before that would have caused MLB scouts/front offices to say, "Hey, we need this guy".

He does, however deserve full credit for 1909-1911. I have been pretty bearish on this PCL yeras but even I recognize this.
   107. Mark Shirk (jsch) Posted: April 22, 2005 at 01:24 PM (#1279124)
sorry about the typing errors, I am at work right now and dont' have time to edit my posts. Ugh.
   108. Carl G Posted: April 22, 2005 at 01:42 PM (#1279141)
He was a league average RF in 1906. He was still below average in '04-05 and probably around Replacement in '03. That is not a clear cut major leaguer or a clear-cut signing. I'm thinking that '06 is the first year that would attract attention and get him signed to an ML club. Remember, virtually everyone we'll be considering from here to eternity had at least 1 or 2 minor-league seasons that would translate to somewhere between replacement and average for their position; some maybe even better. Are we prepared to translate these couple seasons for everyone on the ballot? If we don't and we give Cravath credit for all of his above replacement minor league seasons, we are effectively giving him an unfair advantage. Cravath obviously was a star in '07 and deserves credit. It is also inexplicable to me that he wasn't a big-leaguer in '09-11; again, deserves credit. Beyond that, I think we're dipping too far into the well; unless, of course, we're going to undertake this analysis for every player with minor-league stats currently on the ballot and those who come in the future. This seems excessive when the net result will probably be a 'wash' anyway. I prefer to start Cravath in '07 and go from there. To be clear, I am a FOGC, but I want to get this right and the more I think about this, the more I'm convinced that giving him credit for anywhere between '04-06 would give him an unfair advantage over other candidates.
   109. Mark Shirk (jsch) Posted: April 22, 2005 at 02:26 PM (#1279203)
How many games should we give Cravath in 1907? He was only a part time player with the Red Sox in 1908 in part because he was underappreciated and in part because his team already had two decent corner OFers. Should Cravath recieve a full season's wort of credit for 1907 when it is entirely possilbe that he wouldn't have had more then 350 AB's?
   110. PhillyBooster Posted: April 22, 2005 at 03:46 PM (#1279379)
Given his low Avg before 1907, I still question whether an ML manager would have given him so much PT. The extra walks improve him in our minds, but would it have done anything for the decision makers then?

This strikes me as double counting against. When considering whether Player X, who was above replacement level but not in the majors, should get credit, I don't think it makes sense to say, "no, because the decision makers wouldn't have played him." Of course, the decision makers wouldn't play him. They didn't even sign him! That's why he's off in Los Angeles and we're asking whether to give him extra credit!

Also, in my opinion the ranking of the PCL at essentially at AA level is too low. I recognize that it was considered the lowest of the "high-minors", but working against that is the fact that it was the highest quality league within 2,000 miles. Essentially all the best players from the time zone passed through there. One great season in Buffalo will get easily noticed in Cleveland, but PCL players could get "lost" out there for a longer time. In addition, players would want to play close to home, so would be more likely to prefer a starting job in Oakland to a bench job in New York.

I can see a larger deduction than for the American Association or Eastern (Int'l) League, but only by a few percentage points, not almost all the way down to AA.
   111. Dr. Chaleeko Posted: April 23, 2005 at 02:28 AM (#1281325)
GAVY CRAVATH MLE WS, 1903–1911

YEAR__LG___OPS+*__PA_____OWS___FWS___WS
1903__AL___88____454___4.8____2.1___6.9
1904__NL___97____424___6.6____1.9___8.5
1905__AL___101___402___8.8____1.8___10.6
1906__NL___109___563___16.1___2.6___18.7
1907__AL___135___499___28.4___2.3___30.7
1909__AL___136___453___23.6___2.1___25.7
1910__NL___150___664___36.5___3.0___39.5
1911__AL___167___659___42.8___3.0___45.8
4118___167.6__18.8__186.4

*Cravath's OPS+ as MLE'ed by Brent

I followed the method I laid out in post #148 on the 1949 ballot results page. I should, however, point out a few things.

NOTES
I used the same method that Chris Cobb uses for league environment: even years NL, odd years AL. I used the SBE (ver. 2002) to find yearly OPS+ comps by league. For years where at least one or more players finished with the same OPS+ as projected for Cravath, I used a simple average of their OWS/PA to figure out the rate at which he would accrue OWS. In seasons where no one finished at his exact OPS+, I then moved on to those who finished within a point, and usually there were enough data points on either side of him to average out…however, there are a few funky years.

1910: Cravath’s MLE OPS+ led the NL by 11 points. Cravath’s figure led NL leader Sherry Magee by about 7%, so I prorated Magee’s OWS/PA rate and then rounded down from .0556 to .055 to be conservative. I chose to be conservative because I don’t know the exact nature of the relationship between OPS+ and OWS/PA.

1911: Cravath’s MLE OPS+ led the AL by 15 points. Cravath’s figure led AL leader Ty Cobb by about 9.8%, so I prorated Cobb’s OWS/PA rate and then rounded from .0693 to .065 to be very conservative.

For FWS, I first figured his FWS/est defensive inning. I estimated Cravath’s MLB defensive innings at around 9400 by dividing his MLB career plate appearances by 4.2 then multiplying the result by 8.5. It’s possible this would result in my giving him too many defensive innings. Clay Davenport estimates Cravath at around 1080 adjusted defensive games, which if multiplied by 8.5 would only net 9180 defensive innings. Then I divided his 19 FWS by 9400 to get .002 FWS/est defensive inning. I think this is probably a smidge too low for the reason cited above, but not off by so much it would dramatically change our picture of Cravath. Next I assumed that Cravath, like many players, would be a slightly better defender in his youth. I used trial-and-error to come up with an MLE FWS/inning rate that showed him as better but not so much better that it looked utterly out of context within his seasonal MLB FWS numbers. I ended up going with .00225 WS/inning, .00025 higher than his MLB days. This gives him two seasons off full-time play at ages 28 and 29 where he is at 3.00 WS for the season, a number he exceeded only in 1915, but which isn’t too far removed from the 2.6 and 2.7 he posted at ages 30 (in less than full-time play) and 31 (in full-time play).

Happy to share any of my information with anyone who wants to see it, or to go back and try again if anyone thinks I should change any of the assumptions I made.
   112. Dr. Chaleeko Posted: April 23, 2005 at 02:41 AM (#1281373)
Oops, that was post #48 not #148 on the 1949 results page. And since I'm not great with formatting, here's the totals from the last line in the stats part of my last post:

MLE PAs 4118
MLE OWS 167.6
MLE FWS 18.8
TOTAL MLE WS 186.4

So now my take on it. 1909-1911 are total no-brainers for giving him credit. I don't see 1903-1905 as meriting credit. In 1906, he's a solid regular getting close to All-Star level at the age of 25. He's also made a very sudden jump in productivity and accumulate more PAs as a result. I'm mixed on giving him 1906 at this point. 1907 is an MVP-caliber age-26 year, and I think it's just as easy a call as 1909-1911.

If my estimates are reasonable (and we can all differ on that point...) then he's adding at least four good years of peaky-primey goodness
1907: 30.5
1909: 25.7 (+2 MLB WS)
1910: 39.5
1911: 45.8
++++++++++
total 141.5

That's 35 a year. It brings up around 340 career WS.

Admittedly these WS look, at first glance, very high, but I don't think they necessarily are. He put up 29, 28, 35, 26, 26 from ages 32-36 (two years down the line from 1911 and onward), so a decline from that kind of height isn't out of the question by any means. It does suggest a late peak, but as Sunnyday2 has often said (or at least I think it's he) strong HOM candidates are often outliers.
   113. Brent Posted: April 23, 2005 at 03:12 AM (#1281417)
Dr. Chaleeko,

Thank you for doing the WS calculations. I will try to take a closer look at them over the weekend and see how they compare with what I would have come up with.

Obviously Cravath in 1910-11 is projecting much higher than any other minor leaguer we've looked at. As you point out, these totals are not out of line with his major league career, but I'm still a bit astonished. I'll doublecheck my formulas.

Also, I agree with you that based on the projections, 1903-05 could be dropped as representing a normal minor league apprenticeship, but I think 1906 and definitely 1907 and 1909-11 should be given credit.

Phillybooster,

You could be right about me knocking too much off the PCL record for league quality. Cravath's projected PCL record seems too low relative to his subsequent AA and NL record. It seems to me that there are three possible explanations:

1) Cravath's batting skills may have been uniquely adapted to bandbox-type parks like Nicolet Park in Minneapolis and the Baker Bowl in Philadelphia, so that his career really took off in 1909 when he arrived in that environment.

2) My estimates of PCL league quality for the aughts may be too low.

or 3) He may have been an unusual late developer as a hitter.

Phillybooster, I recall you saying that you now have Coast League Statistics. I think we could be convinced that the league quality was higher if you could pull several examples of young players from the aughts (say with at least 1200 ABs in both the PCL and the majors) whose batting average and slugging didn't drop too much when they transitioned to the majors. If you can find 6 to 10 players like that, and while looking don't run into other players who did fall off a cliff, then I think I'd agree that the PCL quality should be rated higher. Because I had seen indications that the quality may have been lower than in the 1920s I was trying to be conservative.
   114. Paul Wendt Posted: April 23, 2005 at 04:02 AM (#1281474)
The first number of Baseball Magazine was dated May 1908. "A Prospectus of the Baseball Magazine" was received by the Library of Congress in December 1907. The "Initial Bow" (p4) begins: "We offer this, not as a finished publication, but simply as a sixteen-page prospectus, giving some idea of what the finished Baseball Magazine will be. The first regular and complete number will appear April 11."

Jacob C. Morse was President of Baseball Magazine, Inc., and Editor of the magazine, as well as sporting editor of the Boston Herald daily newspaper and Boston correspondent to Sporting Life. I consider him responsible for the unattributed content of the Prospectus, including the chief article, "In the World of Baseball," pages 5-9. Maybe he wrote it. It covers the Boston Americans, Boston Nationals, and New England League.

Baseball Magazine (Jacob C. Morse?) on Gavy Cravath, Fall 1907
. President John I. Taylor of the Boston Americans is extremely optimistic as to the outlook for the season of 1908. The country has been scoured ... . No money has been spared ... .
. Undoubtedly the changes in the team will be of the most sweeping character. ... Being aware that his team ranked last in stolen bases for two seasons he directed his agents to secure no player who was not fast.
. Outfielder Thoney ... is conceded by many authorities to be the fleetest man in the country. ...
. Another sensation is promised in outfielder Cravath of the Los Angeles Club, who is a find of Hugh McBreen, during his extended search of last season. Cravath was a bulwark of strength to his club last year and hit the ball in a way that was a revelation to the other outfielders of his league. ...
. With Thoney and Cravath assured outfield places it will be interesting to watch the battle for the remaining outfield position. Three members of last year's outfield remain and six other players have been drafted or purchased.


Cravath "hit the ball in a way that was a revelation to the other outfielders of his league." Thanks for the description, Jake.

I have noted for myself a few other tidbits on Cravath that I will post soon.
   115. Paul Wendt Posted: April 23, 2005 at 05:07 AM (#1281541)
AA1909 (168-game schedule)
<u>Lastname g atbat r h 2b 3b hr sac sb avg (tb slug)</u>
Cravath 125 413 60 120 23 7 4 16 21 .290 (169 .409)
Hickman 167 644 70 183 49 7 5 13 21 .284 (261 .405)
Beckley 113 428 41 126 16 3 1 18 12 .280 (151 .353)
Collins 153 556 61 152 21 3 2 12 13 .273 (185 .333)


Among the 21 regulars with batting averages .260 and above (this "name" quartet ranks 2,4,6,11), Cravath is 6th in steals per atbat. I guess that is better than people expect, even if several of the 21 are elder statesmen. On the other hand, his batting is merely excellent, not off the charts; comparable to ol' Piano Legs.
   116. Chris Cobb Posted: April 23, 2005 at 01:28 PM (#1281737)
I haven't looked over Cravath's MLEs and WS fully yet, but one reason the 1910-11 AA win shares may be so high is that Cravath is credited with a very large number of plate appearances: 4.4 & 4.3 per game, not counting sacrifice hits.

That may be right, based on his actual AA stats, but it's too high for the majors at that time, where 4-4.1 would probably fit a clean-up hitter, assuming that's how Cravath would have been used.

If his PA are dropped to around 600, his WS by the basic pro-rating method would drop by about 10% each season, which makes his totals closer to his documented career best. 1910 would still match his best ML season, and 1911 would be his career year.

Incidentally, if one doesn't want to drop the PA totals, one might then prorate win share estimates by games rather than PAs: players in high-offense environments get fewer WS per pa than players in low-offense environments.
   117. ronw Posted: April 23, 2005 at 05:10 PM (#1281978)
Sorry, but I don't see Cravath as better than, for example, Jimmy Ryan (38th last year)

We have a few precipitous drops, but the highest seems to be Jimmy Ryan, who has been relegated to the worst of the glut.

It looks like OPS+ is making the case for Cravath, so what is that like?

Cravath..1912-1917..119..172..160..171..147..153
Ryan.....1887-1892..109..174..142..131..129..144

Here average is Cravath 154, Ryan 138.


Now lets throw fielding into the mix:

Year......WARP1.....WARP3
1887.......7.1.......4.6
1888......11.6.......8.3
1889......11.1.......8.5
1890.......9.0.......6.2
1891.......7.6.......5.8
1892.......9.1.......6.8

Then in 1893 we have the train wreck. Before it, playing full time, he had an average five-year peak at 9.25 WARP1 and 6.7 WARP3. Of course, he was a CF at this time, so his WS numbers are outstanding. Paul can correct my schedule numbers, I just took the maximum games played by any one team in the season

Year......MaxGames...RawWS.....AdjWS
1887......124........18........22
1888......135........34........39
1889......134........25........29
1890......137........23........26
1891......137........22........25
1892......154........25........25

So Ryan averaged 24.5 RawWS and 28 AdjWS during his 6-year pre-accident prime.

Cravath, who has now vaulted ahead of Ryan in the voting, has the following 6-year ML prime, according to WARP

Year......WARP1.....WARP3
1912.......5.3.......3.8
1913.......8.6.......7.1
1914.......7.9.......6.4
1915......10.6.......9.5
1916.......7.2.......5.9
1917.......7.3.......5.7

Average 7.8 WARP1, 6.4 WARP3

And WS

Year......MaxGames...RawWS.....AdjWS
1912......154........15........15
1913......154........29........39
1914......154........28........38
1915......154........35........35
1916......154........26........26
1917......154........26........26

Average (no adjustment needed) - 26.5

Even with Cravath's great 6-year prime, he doesn't beat Ryan on WARP1, WARP3 or AdjWS. If you don't adjust Ryan's WS, then Cravath beats him there.

Of course, we debate whether to give Cravath MLE credit for 3-5 minor league seasons, but outside Ryan's 6-year ML prime, he was playing in the consolidated 1890s NL, earning another 53.2 WARP1, 38.8 WARP3, and 154 RawWS over 10 more seasons as a solid to good outfielder.

Here's the argument against me as I understand it. These are Ryan's 24-29 seasons. Cravath was in the minors during those ages, and would have put up similar (or better) numbers at those ages that he did in the majors from 31-36 if he had just been given a chance. We can't argue that Jimmy Ryan ages 31-36 was a better player than Cravath ages 31-36. The claim is that if MLB was remotely smart, it would have realized that Cravath ages 24-29 was better, so we should give credit for those ages.

Sure, luck is a factor in making the MLB, but during 24-29, Cravath had a shot, and didn't make it. From the MLB records we have, they had comparable 6 year MLB peaks. Outside of that, we have Cravath's fabulous 4 years of high-minor league time, and Ryan's solid 10 years of mostly 12-team ML time. I guess I have to go with the solid ML time.
   118. OCF Posted: April 23, 2005 at 05:26 PM (#1282020)
MLE:
1910 .291/.395/.434 MLE WS 39.5 ????
1911 .320/.425/.533 MLE WS 45.8 ????

1911: Cravath’s MLE OPS+ led the AL by 15 points. Cravath’s figure led AL leader Ty Cobb by about 9.8%,

I'm sorry, but something does not compute here.

Ty Cobb in 1911 was .420/.467/.621 for 48 WS. How is .320/.425/.533 supposed to even be in the same time zone? And what does "Cravath's MLE OPS+ led the AL by 15 points" mean? That's not AL league-leading stuff.

The MLE's are also well behind Magee in 1910.
   119. Dr. Chaleeko Posted: April 23, 2005 at 06:04 PM (#1282178)
OCF,

It just means that the OPS+ listed by Brent led the the OPS+ I found when sorting OPS+ by league on the SBE. Perhaps I'm misinterpreting the SBE's numbers?
   120. Dr. Chaleeko Posted: April 23, 2005 at 06:36 PM (#1282301)
Actually, I think I did misinterperet what I was looking at. The SBE numbers were simply, I think, an unadjusted versus-average calculation. I've just now cross-checked against bb-ref.com's OPS+ numbers, and they are, as OCF implies much different.

I'll rerun the WS estimates with bb-ref's OPS+ numbers, and wipe the copious egg off of my face as I do.

OCF, thanks! This was my first real stab at this, and now I'll have a better system going forward in case I try it with other players.
   121. Dr. Chaleeko Posted: April 23, 2005 at 07:13 PM (#1282515)
REVISED CRAVATH MLE WIN SHARES

YEAR__LG___OPS+_PA__OWS___FWS___WS
------------------------------------------------------------------
1903___AL____88___454___6.0____2.1____8.1
1904___NL____97___424___7.8____1.9____9.7
1905___AL___101___402___7.9____1.8____9.7
1906___NL___109___563___14.1___2.6___16.7
1907___AL___135___499___19.0___2.3___21.3
1909___AL___136___453___15.2___2.1___17.3
1910___NL___150___664___27.9___3.0___30.9
1911___AL___167___659___31.1___3.0___34.1
======================================
TOTALS__________4118__129.0__18.8__147.8

NOTES
I used the same method as previously discussed but this time used bb-ref.com to find yearly OPS+ comps by league. This looks much smoother than the previous iteration compared to his subsequent MLB career. Thanks for your patience.

1906: His nearest OPS+ comps were all at 108/107, see:

NAME____OPS+__OWS/PA
Leach____107____.02529
Nealon___108____.02469
P Bennet__107___.01706

Due to this observation of variance within OWS rates versus OPS+, I chose to place Cravath’s OWS/PA rate at .02500, just below Leach’s rate.

1907: Same sort of situation in 1907; Lajoie plays the role of Leach 06, with a .04014 OWS/PA rate, and so I placed Cravath at .038 since the other two comps had lower OWS/PA rates than Lajoie.

1909: Cravath is MLEed to a 136 OPS+. Nearest comps

NAME____OPS+__OWS/PA
Murphy___133____.030519
Heitmuller_136____.033891

So I put Cravath at .0335.

1910: Cravath is MLEed to a 150 OPS+. No player in the league came within three points. Two finished at 154, so I prorated their OWS/PA back to a 150 OPS (3 percent reduction) to get .042.

1911: Cravath is MLEed to a 167 OPS+. Nearest comp is Crawford at 163. Cravath’s OPS+ is 2.45% higher than Crawford’s rate, so I prorated Sam’s OWS/PA rate upward by that much to get .04725.
   122. Chris Cobb Posted: April 23, 2005 at 07:37 PM (#1282613)
Dr. Chaleeko,

Revised WS look good! and the season-by-season account of your choices is quite helpful.

I wonder about the causes of the rate differences among the 3 comps in 1906: maybe the fact that St. Louis missed their pythagorean record by 6 wins, and were a bad team on top of that has something to do with it.

I also wonder about Gravath's PA totals for 1910-11: I hope Brent will address that at some point.
   123. Dr. Chaleeko Posted: April 23, 2005 at 07:59 PM (#1282673)
Thanks Chris!

FYI: If Brent ends up revising his PA totals for 1910-1911, I'll go ahead and repost the WS. In the meantime, anyone can easily play along at home by simply dividing the OWS by the PAs and taking the resulting rate and multiplying it to whatever PAs you feel are appropriate.

For fielding take the PAs you thought were appropriate, divide by Chris's suggested total per game (4 or 4.1) to get est. defensive games, then multiply by 8.5 to generate est. defensive innings, and finally multiply the est. defensive innings by .00225 to replicate the FWS rate that I used (see comments above).
   124. Brent Posted: April 23, 2005 at 10:01 PM (#1283013)
Chris Cobb wrote:

I haven't looked over Cravath's MLEs and WS fully yet, but one reason the 1910-11 AA win shares may be so high is that Cravath is credited with a very large number of plate appearances: 4.4 & 4.3 per game, not counting sacrifice hits.

That may be right, based on his actual AA stats, but it's too high for the majors at that time, where 4-4.1 would probably fit a clean-up hitter, assuming that's how Cravath would have been used.


Good point. On some earlier MLEs, it was pointed out that I had my PA/G too low, so I put in a formula to boost them, but it looks like I overshot. In future I'll try to compare PA/G to comparable MLB players of the time.

After looking at several comparables (including the ones listed by Dr. Chaleeko above), I've limited Cravath's (AB+BB)/G to 4.03 for 1910 and earlier, and to 4.13 for 1911 (when the cork-centered ball caused offenses to pick up).

So, here are Gavy Cravath's MLEs, take 3:
Year  Lg Age   G  AB   R   H 2B 3B HR  BB  AVG  OBP  SLG OPS+
1903 PCL  22 104 380  39  91 22  4  3  38 .239 .309 .342   90
1904 PCL  23 101 368  39  87 17  1  5  39 .236 .310 .329   99
1905 PCL  24  98 350  35  83 15  3  4  44 .237 .322 .331  101
1906 PCL  25 133 473  60 114 26  5  4  62 .241 .329 .342  109
1907 PCL  26 138 440  61 119 29  2  6  59 .270 .357 .386  135
1909  AA  28 115 394  53 109 21  5  4  59 .277 .371 .386  136
1910  AA  29 150 517  70 151 32  7  9  89 .292 .396 .433  150
1911  AA  30 153 534  97 171 42  7 19  98 .320 .426 .532  166
1921 PCL  40  83 247  30  67 14  0  9  35 .271 .362 .437  106
1922  AA  41  33  58   7  14  2  0  2   9 .241 .343 .379   87

Some of the rate statistics change slightly due to rounding.

Dr. Chaleeko - you're latest win share estimates look more like what I had been expecting. Thank you for tackling them.
   125. Brent Posted: April 24, 2005 at 02:20 AM (#1283399)
I promised to post some information on Cravath’s minor league fielding statistics. These come from guides, and I am missing 1906.

The records for 1903-04 are the only ones I’ve seen that show the records for each outfield position separately; unfortunately, they do not show defensive games played by position, so I based range factors for those two years on total games played.

Year  Lg   G  PO  A  E   FP  Rng 
1903 PCL  -- 316 25 22 .939 1.63*
1904 PCL  -- 324 53 29 .929 1.79*
1905 PCL 204 277 33 14 .957 1.52
1906 PCL  --  -- -- --   --   --
1907 PCL 173 287 41  9 .973 1.90**
1909  AA 125 198 15 11 .951 1.70
1910  AA 164 300 20 33 .907 1.95
1911  AA 167 313 19 21 .941 1.99
1921 PCL 112 139 12  2 .987 1.35
1922  AA  25  25  2  2 .931 1.08

* = Right field only; Range factor calculated using batting games
** = In 1907, also played 9 games at 1B, with 96 PO, 14 A, 5 E, .956.

Comparisons:
For 1903, ranked 4th in fielding percentage of 6 full-time right fielders.
For 1904, fielding record shown is for right field only; record for all positions is 341 PO, 55 A, 30 E. Reach Guide does not indicate what his other position(s) were.
For 1904, ranked 2nd in fielding percentage and 1st in assists of 6 full-time right fielders.
For 1905, ranked 9th in fielding percentage and 5th in assists among 19 outfielders with 100+ games.
For 1907, ranked 1st in fielding percentage and 3rd in assists of 10 outfielders with 100+ games.
For 1909, ranked 15th in fielding percentage of 21 outfielders with 100+ games.
For 1910, ranked last in fielding percentage of 22 outfielders with 100+ games.
For 1911, ranked 15th in fielding percentage of 20 outfielders with 100+ games.
   126. PhillyBooster Posted: April 25, 2005 at 04:52 AM (#1285678)
I think we could be convinced that the league quality was higher if you could pull several examples of young players from the aughts (say with at least 1200 ABs in both the PCL and the majors) whose batting average and slugging didn't drop too much when they transitioned to the majors.

Tall order! Without a data base, it'll take some searching. Lots of the guys I'm finding played fewer than two years in one league or other. There are some interesting players there, though!

It seems to me that using the PCL might actually be useful in gauging league strength, since more players went from the PCL to the AL or NL (and vice versa) than went between the majors.

I'll start in 1903 and see what I come up with . .
   127. PhillyBooster Posted: April 25, 2005 at 06:16 AM (#1285832)
Okay, let's start at the very beginning (since I've got insomnia tonight) and look at the 1903 PCL players . . . A lot of "old" major league talent joined the new PCL right in 1903 -- much more than future ML talent that year. Some of the "talent" was more talented than others. I was trying to compare AVG and SLG from 1901-02 majors to 1903-04 PCL for players who were hitters, but the numbers don't usually break down that easily.

In terms of league quality, it shouldn't make too much of a difference which direction the players are going, so I'll post these numbers for now.

1903, Los Angeles Looloos.

Dummy Hoy:

ML: .293/.393 (806 ABs, age 39-40)
PCL: .257/.334 (808 ABs, 1903 only)

Pop Dillon:

ML: .254/.324 (1042 AB, 1901-1902 and 1904)
PCL: .250/.341 (1530 AB, 1903 and 1905, age 29 and 31)

1903, San Francisco Stars

Charlie Irwin:

ML: .249/.304 (960 AB, 1901-1902, age 32-33)
PCL: .288/.351 (1558 AB, 1903-1904, age 34-35)

Hitters who will be considered in later years, as their ML careers followed their PCL careers: Gavy Cravath, Tommy Sheehan

Some hitters who crossed over but not considered due to minimal playing time in one league or other in relevant years. Could be part of a more complete study that I might do later: Jud Smith, George Wheeler, Herman Spies, George Hildebrand, Charlie Graham, Harry Lumley, Joe Dolan, Parke Wilson, Truck Eagan, Henry Krug, Danny Shay, Tom Leahy, Dave Zearfoss, Deacon van Buren, Art Kruger, Moose Baxter
   128. TomH Posted: April 25, 2005 at 02:31 PM (#1285988)
repeat request, from a guy who is way behind in reading many 100s of posts

Gavy Cravath is back in my consideration after the fine work done by my esteemed colleagues. But I am still waiting for someone to convince me via direct comparison why he belongs ahead of P Browning, H Wilson, and C Jones (none of whom made my ballot last week). You can add in years of reasonable minor league credit for Cravath, but also add in C Jones blacklist missing time, and the shorter 19th cent schedules. Additionally, Cravath took unusual advantage of the Baker Bowl, which distorted his power numbers. Go ahead...convince me!
   129. Dr. Chaleeko Posted: April 25, 2005 at 02:52 PM (#1286014)
TomH,

I think it depends quite a bit on your preferences. First, let's look at how Cravath stacks up with minor league credit. I'm currently giving him 1907 and 1909-1911. These seasons add roughly 21.3, 17.3 (plus his 2 MLB WS), 30.9, and 34.1 WS to his existing WS. Adjusting all seasons to 162 games, yields this:

CAREER TOTAL: 325
BEST ANY 3: 105
BEST ANY 5: 164
BEST ANY 10: 279
BEST ANY 15: 325

That's a really nice peak and a not insubstantial career. It places him around 40th all time among corner outfielders in my rankings. Among eligibles, it puts him a shade above George Burns who looks like this

Car. 314
3 105
5 161
10 279
15 314

Almost indistinguishable, but a smidge more career.

He's very similar also to Hugh Duffy:

car. 337
3 106
5 166
10 289
15 336

Duffy's a little better and was a CF. I rank Duffy around #15 among all-time CFs in my rankings. Hack Wilson is around 35th by dint of
car. 236
3 103
5 160
10 231
15 236

With MiL credit, Cravath surpasses Wilson easily by vaulting over him in career.

I'll be honest, Browning and C. Jones are two guys that I'm having a lot of trouble with because I'm not sure yet about league-quality issues. I think Browning's career is probably about 30-50 WS shorter than Cravath's (with MiL credit), and his peak is better than or equal to Gavy's. Same for Jones. Those early NL and AA leagues were rather sketchy in terms of total quality.

As for the Baker Bowl issue, not to be too flippant, but I've got two words: Mel Ott. Seriously though, the fact that Cravath also had big years in Minneapolis should quell some of this doubt. I'm taking the he-adapted approach, and, of course, his OPS+ and WS are park-adjusted to remove at least some of the residue of the Baker Bowl.

So that's where I'm at with him. He's safely on my ballot. Second or third best corner OF eligible (Waner and maybe Dihigo ahead), slotting right in behind the best available CF (Duffy). He's mid-ballot material this year.
   130. PhillyBooster Posted: April 25, 2005 at 03:54 PM (#1286125)
1904 PCL gives a couple of players who vaulted from the PCL to "careers you've actually heard of."

Orval Overall had a 2.94 ERA/ 364 K(led league), 173 BB in 510.7 IP at age 23. His K rate dropped a little in the switch to the majors in 1904, but he had a 115 ERA+ and a respectable 108-71 major league career.

The other guy who made the jump from PCL in 1904 to ML regular was Hal Chase, who was .279/.352 in the PCL for Tacoma as a 21 year old in 702 ABs, and .249/.329 for the Yankees as a 22 year old.

Also worth noting in Doc Newton, who played for Los Angeles for 2 year (1903 and 1904) in the middle of his major league career (looks like he was cut for behavioral problems after his best season in 1902 -- he pitched better when he was drunk). His ERA for those years were:

1902 Dodgers -- 2.42 ERA in 265 IP (114 ERA+)
1903 L.A./Portland -- 2.54 in 413 IP
1904 L.A. -- 2.17 in 478 IP
1905 Yankees -- 2.11 in 60 IP (139 ERA+)

Meanwhile, Ike Rockenfeld parlayed a .222/.333 in the PCL for Portland to a .221/.260 for the Brown in 1905/1906. I guess you get what you pay for.

Going the other way, Tim Flood was a career .233/.290 with the Dodgers for (mostly) 1902 and 1903 in 816 ABs. In 1904/1905 with Los Angeles, he went .255/.333 in 1500 at bats.

George van Haltren went .270/ .342 in 933(!) ABs in 1904, after going .257/.286 in his last year with the Giants at age 37.

The great John Gochnauer did not have his career revived in California. A .187 career batting average from 1901 to 1903 in the majors converted into a .161 in 285 at bats for the San Francisco Seals in 1904

Included in later years: Charlie Hall, Bobby Ganley, Larry Schlafly
   131. jimd Posted: April 25, 2005 at 06:08 PM (#1286399)
Integrating Dr.Chaleeko's minor-league numbers with Bill James' major league numbers.

1903___AL____88___454___6.0____2.1____8.1
1904___NL____97___424___7.8____1.9____9.7
1905___AL___101___402___7.9____1.8____9.7
1906___NL___109___563___14.1___2.6___16.7
1907___AL___135___499___19.0___2.3___21.3
1908_*_AL___136___327___11.4___1.0___12.4
1909_*_AL____92____77____2.0___0.5____2.5
1909___AL___136___453___15.2___2.1___17.3 (19.8)
1910___NL___150___664___27.9___3.0___30.9
1911___AL___167___659___31.1___3.0___34.1
1912_*_NL___119___502___12.8___2.6___15.4
1913_*_NL___172___594___25.9___2.7___28.5
1914_*_NL___160___604___26.2___1.5___27.7
1915_*_NL___171___621___30.2___4.4___34.6
1916_*_NL___147___532___23.9___1.9___25.8
1917_*_NL___153___590___24.0___2.2___26.2
1918_*_NL___106___489____9.5___2.0___11.5
1919_*_NL___213___255___15.2___0.2___15.5
1920_*_NL___145____54____2.0___0.0____2.0
_______________________312.1__37.8__349.9

Notes on Win Shares:

1903-5 are very optimistic. In 1903 he's below replacement, and in 1904-5 he's at replacement level. Full-time playing time is possible (with a really bad team like Washington) but not likely. He was better off in the PCL.

1906: Is this the year he gets noticed? He's still below major-league average hitter for a corner-OF. The fielding estimates may be 20% too high because they do not reflect his ML performance in 1908/9 (1.5 DWS in 404 PA)

1907: The year that did get him noticed. This year is major league quality, somewhat above average. A "star" would be stretching the definition.

1908: Looks good, why didn't Boston keep him? Competition in RF: Doc Gessler, who posted a 162 OPS+; flash-in-the-pan career year, but them's the breaks. Competition in LF: Jack Thoney, 96 OPS+ but, to quote Paul Wendt's quotes from the annual guides of the times: Outfielder Thoney ... is conceded by many authorities to be the fleetest man in the country.

1909: White Sox tryout: they're looking for a replacement for Fielder Jones, long-time gold glove CF. Auditioning for the wrong part. They should have used Cravath to replace RF Eddie Hahn, but Gavy didn't hit like he could. Senator tryout: 1 game? C'mon.

1910-11: It's been noted before, but the PA's are too high, by maybe 10%. Win Shares estimates similarly.

Bottom line: I'd say the total is 25-50 Win Shares too high, depending on how much credit is given pre-1907. Still looks like a candidate to me.
   132. PhillyBooster Posted: April 25, 2005 at 06:45 PM (#1286473)
Movin' right along. . . .

Jim Nealon: .286/.403 in 787 ABs for S.F. in 1905 (led the team with 48 doubles and the league with 11 triples).
.256/.304 (103 OPS+) in 937 Pirate ABs in 1906-07 (third in triples in 1906).

Tommy Sheehan: .261/ .325 in his 25-27 years in Sacramento (1903) and Tacoma (1904-05) over 2341 at bats.
He was .235/.280 in the majors, mainly 1906-1908 for the Pirates and Dodgers.
Oddly, 1905 was the weakest of his three years in the PCL, so that he actually improved from 1905 to (.229/.284) to 1906 (.241/.289).

Larry McLean was .285/.352 for Portland in 1905. This fills in a blank spot in the middle of a 13 year major league career in which he was .262/.323. He was, however, particularly bad in 1904 and 1906 before breaking through in 1907 for the Reds.

Larry Schlafly was .258/.345 in the 1904-05 PCL in 1349 ABs, playing for three different teams (Oakland in '04, Portland and L.A. in '05).
He was .246/.329 for the Senators in 1906. That year was about 2/3 of his major league experience, and he was a career .240/.330 in the majors in 658 ABs.

Going the other way was Ned Garvin, who left (I have no idea why) after a 1.72 ERA (158 ERA+) performance with the Dodgers and Yankees in 1904 to play in Portland in 1905. He went 25-20, with a 1.96 ERA in 399 innings. In 1906, he split time with Portland and Seattle, going 20-19, with a 2.56 ERA.

Coming up: Bill Sweeney, John Kane
   133. Dr. Chaleeko Posted: April 25, 2005 at 08:03 PM (#1286697)
Phillybooster,

I'm just following along with your lists and doing simple division with your numbers. In 1903, the batting averages are coming in a little under .9 of big league avgs. From 1904 onward, averages seem to go up to around .9 of MLB averages with more guys getting over .9 than in 1903. Small samples, of course.

Looks like the varation in SLG between leagues is much wider (not a surprise I suppose) with one guy losing as much as 25% of his SLG but others losing just 5%. Naturally, they track AVG as well.
   134. PhillyBooster Posted: April 25, 2005 at 09:07 PM (#1286845)
I'm not much for figuring out MLEs or reductions. I'm just the guy with the book of raw numbers. I started giving relevant pitchers too, becuase I thought it might help someone distinguish between league quality and "hitters/pitchers league" effects.

It is entirely possible that the PCL was lower than the other high minors, but they were all three higher than AAA affiliates would be today due to inefficient scouting.

I'll go through the rest of the aughts over the next several days. Don't really know what the numbers will say. I'm pretty convinced myself on Cravath, and I don't know if these numbers will help or hinder my case, but I figure they should be out there.
   135. jimd Posted: April 25, 2005 at 11:02 PM (#1287095)
but they were all three higher than AAA affiliates would be today due to inefficient scouting.

This point is debatable. What is clear is that the range of talent in each league was much greater than today.

On the high-end, there were players not yet "discovered" by ML teams, as well as useful "has-beens" that modern "farm" clubs have no use for, they not being "prospects".

OTOH, the inefficient scouting cut both ways. The leagues at the AA level at that time would also have players that hadn't been "discovered" by the AAA teams, players that today would be in AAA or maybe even in the majors. So the AAA teams then also had a number of AA level players that they were always looking to replace.
   136. Cblau Posted: April 26, 2005 at 01:46 AM (#1287978)
As for the Baker Bowl issue, not to be too flippant, but I've got two words: Mel Ott.
No, it was Shibe Park that he never homered in.

Ott didn't take unusual advantage of the Polo Grounds, except in HRs. Career OPS H/A: .979/.917 for a difference of 6.7%, vs. the normal 5% advantage at home. (Source-Total Baseball I)

As for why Garvin was dropped from the majors: I believe it was due to his drinking.
   137. Cblau Posted: April 26, 2005 at 01:47 AM (#1287985)
As for the Baker Bowl issue, not to be too flippant, but I've got two words: Mel Ott.
No, it was Shibe Park that he never homered in.

Ott didn't take unusual advantage of the Polo Grounds, except in HRs. Career OPS H/A: .979/.917 for a difference of 6.7%, vs. the normal 5% advantage at home. (Source-Total Baseball I)

As for why Garvin was dropped from the majors: I believe it was due to his drinking.
   138. Paul Wendt Posted: April 26, 2005 at 05:28 PM (#1289430)
Ron Wargo #117
Then in 1893 we have the train wreck.
I don't know it.
Before it, playing full time, [Jimmy Ryan] had an average five-year peak at 9.25 WARP1 and 6.7 WARP3. Of course, he was a CF at this time, so his WS numbers are outstanding. Paul can correct my schedule numbers, I just took the maximum games played by any one team in the season

Year......MaxGames...RawWS.....AdjWS
1887......124........18........22
1888......135........34........39
1889......134........25........29
1890......137........23........26
1891......137........22........25
1892......154........25........25


I understand that "everyone" prorates to the number of games scheduled, 126 140 154. It isn't clear to me what should be done. Ties and "never played" games are not uniformly distributed.

Win Shares treats ties and losses equally, so WS/162 punishes a player in proportion to his ties rate.

jimd
1908: Looks good, why didn't Boston keep him? Competition in RF: Doc Gessler, who posted a 162 OPS+; flash-in-the-pan career year, but them's the breaks. Competition in LF: Jack Thoney, 96 OPS+ but, to quote Paul Wendt's quotes from the annual guides of the times: "Outfielder Thoney ... is conceded by many authorities to be the fleetest man in the country."

That quotation is from the Prospectus for Baseball Magazine and I have read similar material in Sporting Life. By the way, some(*) considered Thoney the best player in the minor leagues (Toronto, EL 1907), fleetest or not.

Anyway, Cravath in Boston must be a complex story. BBMag/Morse? did not put Thoney and Cravath specifically in the corners, so Denny Sullivan CF is part of the story. When did Cravath last play for Boston? Was he reserved for 1909? What was expected from Speaker and perhaps Hooper by October 1908?
Also, I recall that Thoney was injured in 1908.

Philly #134:
It is entirely possible that the PCL was lower than the other high minors, but they were all three higher than AAA affiliates would be today due to inefficient scouting.

and multi-year contracts, draft, and reserve rules. MLB clubs could not simply sign minor league players.
   139. karlmagnus Posted: April 26, 2005 at 05:49 PM (#1289484)
The 1908 Red Sox appear to have been a classic "rebuilding" club -- some famous names, but finished below .500. Pitching was an ancient but still extremely effective Cy Young, a young Eddie Cicotte, and a few innings of a very young Smokey Joe Wood. They tried a lot of young guys, of whom Cravath and Speaker are the other ones one's heard of. My guess is, they looked at BA, where Cravath was nothing special and already 27, knew they had something special in the 20 year old Speaker, and guessed wrong as to which of the "youth movement" to back. Trading Cicotte a couple of years later was an even more eccentric decison, IMHO.
   140. PhillyBooster Posted: April 26, 2005 at 06:59 PM (#1289695)
For the record, my statements above regarding the relative quality of the PCL to today's AAA level was not an argument based on data, but merely an possible explanation for Dr. C's observation that stats dropped about 10% from the PCL to the majors (against, I believe about 15% today). I'll leave it to the number crunchers to determine what the exact discount rate to apply is:

Anyway, continuing the great march through the early years of the PCL, we get to the Class of 1906 . . .

Beginning with Mike Mitchell, who seems to be something of a "Cravath-lite". Mitchell played in Portland in 1905 and 1906. He was a combined .310/.407 in those two years, but was much better in 1906 (.339/.466), where he led the league in hits (207) and homers (7).

He debuted in the majors in 1907 at age 27, and immediately gave a .292/.382 performance as a rookie before going on to a very successful career, putting in 8 seasons of .278/.380 (114 OPS+). He had four great seasons in his first five in the majors (age 27-31), including three years with a SLG over .400, and then tapered off.

Next is Bill Sweeney, who also played for Portland in 1905-06 as a 19-20 year old shortstop (he appeared in fewer than half the team's games in '05). He was .261/.338 in the PCL in 857 ABs (.276/.359 for '06 only) before going on to an 8 year ML career for the Cubs and Braves, going .272/.344 (101 OPS+). Sweeney was .258/.296 in his first two years in the majors.

John Kane was .264/.332 for the 1905-06 Seattle Siwashes in 1438 ABs. After a good year for the Cubs as a 24 year old rookie in 1907 (.248/.347, 107 OPS+), he quickly faded and ended up .220/.297 (89 OPS+).

Harry Wolter was .307/.381 for the 1906 last-place Fresno Raisin Eaters. His batting average was 50 points higher than second best on that team, but most of his offensive aid was counteracted by his bad pitching in 288 innings for a 12-21 pitching record. In 1907, he gave up pitching and bounced around to 3 different teams in the majors as a 22 year old, hitting .286/.286. He caught on again in 1909, and ended up .270/.369 in his 1907 AB career.

Going to other way was Oscar Jones, cut by Dodgers after 875 innings over 1903-05 with was 3.20 ERA (92 ERA+), he played in Seattle in 1906 and S.F. in 1907 where he put up ERAs of 2.07 and 2.06 in a combined 900+ innings in his age 26 and 27 seasons.

Coming up: Rube Ellis, Sleepy Bill Burns, Heinie Heitmueller, Jack Bliss
   141. PhillyBooster Posted: April 27, 2005 at 03:36 PM (#1292328)
Moving right along . . .

1907. Down from 6 teams to four, so the searching should get a little easier. Also, one would assume that league-quality improved (as two teams worth of dead-weight dropped out). I think the (admitted small sample of) numbers bear that out.

Gavvy Cravath -- Let us not experiment upon our constant. 'Nuff said.



Sleepy Bill Burns -- was 16-16, 2.55 in 1906, and 24-17, 2.15 ERA in 1907 for Los Angeles. In 1908, he debuted for the Washington Senators at age 28, and went 6-11 with a 1.69 ERA (135 ERA+). He had a brief 5 year career as a "hard luck pitcher", finishing 30-52 over 719 IP -- a losing record every year -- despite a career 100 ERA+.

Catcher Gabby Street put in a full season in the PCL in 1907 for San Francisco, hitting .231/.279. He went back to the majors in 1908, hitting .206/.279 (also for the Senators). His career ML numbers were .208/.256 before he became a manager.


Jack Bliss was another poor-hitting catcher who caught on in the majors. Bliss was a steady .225/.275 in both years with L.A. and Oakland. In 1908 he hit the majors at age 26, and was .213/.265 for the year, and .219/.274 for his 774 PA career. (No walks comparison available, but a large part of Bliss's "value" came from his OBP.)

Also, inadvertently omitted from post 132 above was Charlie Hall, who was 29-19, 3.32 in 1904 and 23-27, 2.62 in 1905, both for Seattle. At age 20, he joined the Reds, and went 4-8, 3.32 in 95 IP (85 ERA+). He was a career 54-47, 3.09 in 910 IP (95 ERA+).


Coming up: Rollie Zieder, Bob Groom
   142. sunnyday2 Posted: April 27, 2005 at 05:52 PM (#1292783)
I admire Gavy Cravath a bunch, expecially being here in Minnesota where we claim a part of his legacy.

But the problem I have with this whole Gavy Cravath boomlet is the selectivity of it.

Lots of players have a minor league track record. If we considered all of them, we could shake up the ratings quite a lot I would reckon depending on the discount.

But we are not considering all of them. So when we select Cravath and Averill for special analysis, we unfairly disadvantage all the other players whose minor league records are not being considered for extra credit.

Now certainly there are reasons to why Cravath's and Averill's minor league records beg for consideration. But vaulting them ahead of other players whose minor league records have not been considered seems to me like a basis for an appeal.
   143. John (You Can Call Me Grandma) Murphy Posted: April 27, 2005 at 06:12 PM (#1292866)
But we are not considering all of them. So when we select Cravath and Averill for special analysis, we unfairly disadvantage all the other players whose minor league records are not being considered for extra credit.

No, we're not. If anything, we're delaying other players' induction, that's all. But even if you're right, why should we have to ignore their professional records when we have documentation that they were better than their major league records suggest?

BTW, who else would fall in this category, Marc (besides Ike Boone, of course :-)?
   144. Dr. Chaleeko Posted: April 27, 2005 at 06:20 PM (#1292911)
Fournier seems like the big unknown in terms of those whose minor league performances might suggest a second look.

That said, I wasn't around for two other guys whose career paths might have merited a second look: Perry Werden and Maury Rathe (sp???). I'm not sure that either of them do require that kind of reconsideration, I'm just curious to know if they might or what conversation was had about them. (of course, my query about Rathe is based solely on James's mention of him in the NHBA).

Brent recently said that based on the PCL numbers for Williams that KW clearly didn't merit further consideration beyond his MLB numbers, so that's one less player whose minor league career might require further scrunity.
   145. PhillyBooster Posted: April 27, 2005 at 06:39 PM (#1292995)
Lots of players have a minor league track record. If we considered all of them, we could shake up the ratings quite a lot I would reckon depending on the discount.

But we are not just considering his "minor league track record". We are considering his minor league statistics for the time when he was wrongfully excluded from the major leagues. I also consider Averill's minor league statistics (although I do not find them as impressive.)

Gavy Cravath played almost 8 FULL YEARS in the top-level minor leagues. That is essentially unprecedented among players who have acquired, say, 150 or more win shares in the majors. Does anyone else in the 20th Century even have five?

The "if we include Cravath we have to include everyone" argument has been made several times, and I just don't see it. I am perfectly happy to include "everyone". That "everyone" is just a much smaller group than I think people are realizing.
   146. Paul Wendt Posted: April 28, 2005 at 12:20 AM (#1294175)
The Phillies will send "Donlin or Cravath" to Minneapolis for catcher Owen
-SL 1913-03-15 News Items
   147. PhillyBooster Posted: April 28, 2005 at 02:27 PM (#1295110)
1908. Still a four team league. Contraction seems to have been based upon ridiculousness or team nickname. No more "Seattle Siwashes" or "Fresno Raisin Eaters." Los Angeles survived contraction by switching from "Looloos" to "Angels" in 1906. Continuing, however, is the Oakland Franchise, that has been known since 1904 as the "Oakland Commuters." (In 1903, they were the "Oakland Recruits," a nickname that was destined to last for only one year.) Did they all have day jobs in San Francisco? Were they all just milling around waiting for the BART to be constructed? Anway. . .

Rebel Oakeswas the offensive leader for the Angels in 1908. He was .288/.344 in 736 ABs. Switching to the Reds in 1909 as a 25 year old, he went .270/.340 and was a career .279/.346, including two years in the Federal League.

Rube Ellis was an Angel in 1907-08. He made a large jump between ages 21 (.239/.310) and age 22 (.269/.362) and kept most of it when he went to the Cardinals at age 23 (.268/.332, 112 OPS+). He was a career .260/.344 (101 OPS+) over four seasons in St. Louis.

Also in L.A. was Ted Easterly, who put up a great .309/.399 in only 376 ABs (about 2/3 of the games). Picked up by Cleveland at age 24 he hit .260/.390 (112 OPS+) in 287 ABs, and ended up with 7 major league season, hitting .300/.394, 116 OPS+ (including 2 FL seasons).

Heinie Heitmuller had similar numbers in 1907 and 1908, and in 1908 led in most offensive categories (hits, doubles and homers). He was a combined .278/.382 over those two years in 1527 ABs. Going from Oakland to the Philadelphia Athletics (who would have known . . .) he continued to improve at age 26 in limited PAs (239), hitting .286/.405. He struggled in 1910, in a limited 121 PAs, and was not seen again.

Fred Beck went .242/.346 as a 21 year old in San Francisco playing 98 games. Picked up by the Braves, he struggled in 1909 (.198/.266) but rebounded to .275/.410 in 1910 and finished .262/.360 (including FL stats).

Among the pitchers, only Dolly Graycrossed over, and did not fare as well. He was 34-14, 1.75 in 1907 and 26-11, 2.17 in 1908. Moving from Los Angeles to the Washington Senators at age 30, his record plummeted to 5-19, 3.59 ERA (67 ERA+), and in 3 seasons (568 IP), ended up with a 75 ERA+.
   148. PhillyBooster Posted: April 29, 2005 at 02:17 PM (#1297505)
Finishing up the "aughts" with 1909. Did anyone else find this helpful? I had fun doing it mostly for finding the stuff I excluded: Players who played 3 games in the 1890 American Association popping up a decade later as PCL regulars; filling in the rest of the careers of the guys who went 0 for 1 for Milwaukee's AL team in 1901; as well as ponering imponderables (was San Francisco's Rex Ames Red Ames's brother? was Oakland Commuter pitcher Vess Loucks the grandfather of Astros Scott Loucks?).

Anyway, 1909 is an expansion year, with the Sacramento Senators and Vernon Tigers added, and Oakland changing its name to the "Oaks".

Rollie Zieder led San Francisco to its first championship. Over 1908-09 in 1447 ABs, he was .267/.338. Picked up by the White Sox in 1910 at age 26, he struggled (.217/.243, 75 OPS+), and finished a career .240/.286 (79 OPS+)

Jack Graney was a good hitting pitcher for Portland, unfortunately, he wasn't really a good-pitching pitcher. He took a .247/.3378 in 1908-09 to Cleveland where he gave up pitching and put up a .236/.311 (88 OPS+) in 1910 at age 24, and a .250/.342 (100 OPS+) over his 14 year career.

Chick Gandil put in a year in Sacramento, hitting .282/.392, and leading the league with 16 triples. He stuggled in 1910 with the White Sox (.193/.262) but came back for a .267/.362 career.

Duffy Lewis put up a .279/.372 season for Oakland. Picked up by the Red Sox in 1910, he improved to .283/.407 (128 OPS+).

As well as a handful of others who wouldn't make the majors until years after 1910 (Ivy Olson, George Cutshaw and Ping Bodie are the most well known).
   149. Gadfly Posted: April 30, 2005 at 06:27 PM (#1300249)
DECONSTRUCTING GAVY CRAVATH

If I had any free time last weekend, I intended to dig up and review an old conversion project that I had done long ago for the Mexican League to post on the Dihigo thread. It is an interesting study; but does not really address any of my current thoughts on conversion factors. So, instead, I ended up playing around for that weekend and the entire last week with the quite fascinating and illuminating career of Gavy Cravath.

As anyone who reads my posts probably knows, I have a friendly argument going on with Chris Cobb about conversion rates. Most of the time, this argument pertains to the Negro Leagues, for which I advocate a conversion factor of .95 BA and .90 SA while Chris Cobb supports a lower conversion factor of .90 BA and .80 SA.

However, a full statement or summary of my argument would be this: “The Negro Leagues were of about Triple-A league quality, most probably slightly better, and the conversion factor from the Triple-A quality leagues to the Major Leagues from 1900 to 1950 has been pretty constant at .95 BA and .90 SA.”

[Of course, by Triple-A quality, I mean the American Association, the International League, and the Pacific Coast League, noting that these leagues’ designations have shifted, from A to AA to AAA, over time.]

In other words, my argument does not just encompass the Negro Leagues, it also encompasses conversion factors for the Minor Leagues. I believe that there were, from about 1900 to 1950, pretty consistent conversion factors between the Leagues designated Major, AAA, and AA; and these conversion factors are:

Conversion Rates (1900 to 1950)
BA .95 - SA .90 (AAA to Major Leagues)
BA .90 - SA .81 (AA to Major Leagues)

In other words, I believe that Chris Cobb is reducing Negro League statistics (and also Triple-A statistics by corollary) to Double-A statistics.

Also, though this is not really important for the current argument, I do not think that the current conversion factors between the Major Leagues and Triple-A are still the same as they were from 1900 to 1950. I believe that the gulf between the Major Leagues and Triple-A quality leagues is currently greater than it ever has been; and, of course, there is no shortage of old-time baseball players who have stated just this opinion.

The simple reason for this is economics. Before 1950, it was quite possible for a player to earn almost as much, sometimes even more, in the Minors than the Majors; and Minor League teams had some control, steadily eroding over time, of their players’ careers. On the other hand, now all Minor League players’ careers are completely controlled by the Majors; and Major League players make much more money than Minor League players, to an almost absurd degree.

[I doubt any current minor league player makes more than the Major League minimum, 300 thousand, and the top Major League players make around 20 million. In other words, the top Major League players make about 66 and two-thirds more than the top Minor Leaguers. On the other hand, Gavy Cravath almost surely made an almost comparable salary in the high Minors as he did in the Major Leagues. From 1900 to 1950, it was commonly said that some players in the PCL made, because of the long seasons, more money than Major League players.]

So I decided to deconstruct the career of Gavy Cravath and see if his career would support my assumptions and arguments about conversion rates between the Major Leagues and Triple-A quality leagues.

CONVERSION RATE PROBLEMS

When doing conversion rates from one baseball league to another, there are numerous inherent elements that must be considered. One is quality of competition, two is adjustment factor, three is regression to the mean, four is sample size, and five is everything else. And all of these factors are inter-related. The great difficulty in getting a true conversion factor that only measures the difference in the quality of the competition is that, usually, the measurement also includes a huge helping of these other four elements.

The first element, Quality of Competition (QC), is pretty obvious and should be the only real factor being measured in a conversion rate. If one league is 10 percent tougher than another league, the conversation rates should be 110 percent one-way and 90.9 percent the other way. Of course, the task of eliminating every other factor except QC is quite hard.

The second element, Adjustment Factor (AF) is, in my opinion, the greatest problem with getting a true conversion rate. This problem is made even more difficult because Adjustment Factors vary greatly from player to player. There are quite different adjustment patterns for players with poor and good plate discipline. There are quite different adjustment patterns for players who are great fastball hitters and players who can hit breaking pitches well. This element is so important that a more detailed discussion is warranted below.

The third element, Regression to the Mean (RM) can also throw the conversion rate off. Usually a player is promoted to the Major Leagues after having a good year. In other words, the player has made adjustments and is playing well. However, whenever a player is playing well, the other players (pitchers or batters, as the case may be) will then make counter adjustments and the player will regress to the mean of his talent. By definition, most Minor to Major League player promotions are plagued by this problem since, by definition, a player is usually promoted when he is playing well.

The fourth element, Sample Size (SS) is the counter-balance to RM. Baseball Statistics, in large enough sample sizes, will eventually overcome both adjustment factor (AF) and regression to the mean (RM). However, the difficulty here is to get a sample large enough to be meaningful. One full modern season (over 500 AB) is actually not enough. Two seasons are even better; but three or four or even more seasons are sometimes needed.

The fifth element, Everything Else (EE) or Miscellaneous includes everything from park factors to injuries to opportunity to play to timeline problems to things that I have not considered yet. These are all pretty self-explanatory. The conversion rate can include a player going from a poor hitting park to a good hitting park or vice versa; a change in performance due to injury; lack of opportunity to play consistently; changes in performance due to age, etc, etc.

These five elements are the crux of my argument with Chris Cobb. I believe that his conversion rates are measuring, and contain, not only the difference in quality of competition, but also quite a large helping of all these other elements.
   150. Gadfly Posted: April 30, 2005 at 06:28 PM (#1300253)
ADJUSTMENTS PROBLEMS, PART ONE

Baseball, they say, is a game of adjustments. This is quite literally true, especially for hitters. The fast-twitch muscle response of the human body peaks when the body stops growing at the end of puberty. In other words, a baseball hitter spends his whole career slowly losing his core talent and its attendant component, speed. However, the hitter makes up for this loss with two things, adjustments and strength.

A player can compensate for his slowing reflexes by adjusting and acquiring the knowledge and plate discipline and physical power to actually become a better hitter. Of course, eventually the slow dissipation of his core talent brings his career back down. This gives the baseball hitter’s career an interesting and predictable shape. A great player, who starts his career early and does not get injured or out of shape, will play for 20 years, starting in his early 20s, peaking in his late 20s and even possibly early 30s, with a long slow fade to around 40 if the player desires to do the work to maintain his talent.

In other words, a hitter’s core talent and speed slowly (sometimes quickly) and steadily fades throughout his career while being compensated for by a hitter’s growing talent to adjust and his strength. This is a very important point since it means that an older player’s ability to adjust is greater than a younger player’s ability to adjust. Of course, in the end, every player follows a different course dictated by his talents and, often, luck.

There are many examples of players who could hammer a fastball debuting in the Major Leagues and flaring like brief novas until breaking pitches regressed their talent back to its mean (Kevin Maas, Shane Spencer, and Dave Hostetler come to mind). There are also many examples of players who come to the Majors with poor plate discipline and never really get any better (Benito Santiago comes to mind). However, the most interesting players, to my mind, are those players who have or acquire good plate discipline and who adjust throughout their careers (Joe Morgan and Ozzie Smith come to mind), making the most of their talents.

However, players with good plate discipline also make deducing conversion rates very difficult. By using such a player, the conversion rate will almost surely also include the player’s ability to adjust. Such a player will become better at adjusting the longer he plays in a league; and, upon switching to a superior league, will take two steps back, one in talent and one in ability to adjust. A conversion rate measurement from such a player will almost surely include both steps.

The career of Gavy Cravath, a player with very good plate discipline, provides a pretty good illustration of basically all the problems that have been outlined above. But first, before I use the career of Cravath to discuss conversion rates, I’ll outline his career and provide some statistics for the framework of the discussion.

SHORT BIO OF CRAVATH

Clifford (Gavy) Cravath (nee Gravatt) was born on March 23 of 1881 in California. After graduating High School, Cravath played semi-pro ball in San Diego and worked odd jobs. In 1902, Cravath played semi-pro ball in Orange County and reportedly hit .554, leading to his signing by the Los Angeles Looloos [hard to believe, but that was the team’s nickname before it was changed to the Angels during the 1907 season] of the newly formed Pacific Coast League [actually a reorganization of the California League which had existed from 1885].

Cravath played for Los Angeles for five years, from 1903 to 1907, before being purchased by Boston of the American League in 1908. Despite hitting with great power for Boston, Cravath did not stick and eventually landed with Washington. Joe Cantillon, the manager of Washington, also owned the Minneapolis Millers, an American Association team and farmed Cravath out to his own club. This was an interesting transaction since Cantillon, who would not be rehired by Washington for the 1910 season, ended up owning Cravath.

Gavy Cravath played for Minneapolis, and Joe Cantillon, from 1909 to 1911. The Minneapolis park, Nicollet Park, had a short right field porch; and the right-handed hitting Cravath altered his swing to become an opposite field power hitter. Cravath might have played out the rest of his long career in Minneapolis, but a clerical error made him again eligible to be drafted by a Major League club after the 1911 season. In other words, the Major League club could simply buy Cravath for the much lower draft price rather than the high cost of purchasing him directly from Minneapolis.

The Philadelphia Phillies of the National League drafted Cravath. The Phillies, just like Gavy Cravath’s previous team, played in a park with an extremely short right field porch that suited Cravath’s opposite field hitting style. Cravath played For Philadelphia from 1912 to 1920. In mid-1919, Cravath became the team’s field manager, a post he also held throughout 1920. In 1920, Cravath played himself very sparingly, choosing mostly to simply pinch-hit.

After the 1920 season, Cravath was released as a manager and player by Philadelphia. Cravath returned to the Minor Leagues, playing and managing for Salt Lake City of the Pacific Coast League in 1921; and returning back to Minneapolis for his final season at the age of 41 in 1922. Cravath retired, returned to California, and became a respected judge.

To summarize his career, Gavy Cravath played from 1903, at the age of 22, until 1922, at the age of 41 in the Major Leagues or in two leagues, the American Association and the Pacific Coast League, directly below the Major Leagues. Knowing this, and before analyzing his career, one would make the assumption that Cravath played increasingly well from 1903 to around 1907, peaked from 1908 to probably 1911 or so, and then declined slowly from 1912 to 1922.

His actual career went like this:

STATISITCS OF CRAVATH

AGE-YR BA-SA-ISA (LBA-SA-ISA) BAC-SAC-ISAC (AB)
22-1903 274-396-122 (262-330-068) 1.046-1.200-1.794 (804)
23-1904 270-397-127 (253-321-068) 1.067-1.237-1.868 (768)
24-1905 259-368-109 (229-287-058) 1.131-1.282-1.879 (697)
25-1906 270-387-117 (240-299-059) 1.125-1.294-1.983 (634)
26-1907 303-441-138 (237-295-058) 1.278-1.495-2.379 (603)
27-1908 256-383-127 (239-304-065) 1.071-1.260-1.954 (277)
28-1909 290-409-119 (237-300-063) 1.224-1.363-1.889 (413)
29-1910 326-505-179 (243-310-067) 1.342-1.629-2.672 (612)
30-1911 363-637-274 (268-357-089) 1.354-1.784-3.079 (608)
31-1912 284-470-186 (272-369-097) 1.044-1.274-1.918 (436)
32-1913 341-568-227 (262-354-092) 1.302-1.605-2.467 (525)
33-1914 299-499-200 (251-334-083) 1.191-1.494-2.410 (499)
34-1915 285-510-225 (248-331-083) 1.149-1.541-2.711 (522)
35-1916 283-440-157 (247-328-081) 1.146-1.341-1.938 (448)
36-1917 280-473-193 (249-328-079) 1.124-1.442-2.443 (503)
37-1918 232-376-144 (254-328-074) 0.913-1.146-1.946 (426)
38-1919 341-540-199 (258-337-079) 1.322-1.602-2.519 (214)
39-1920 289-467-178 (270-357-087) 1.070-1.308-2.046 (045)
40-1921 326-548-222 (288-397-109) 1.132-1.380-2.037 (341)
41-1922 278-444-166 (290-409-119) 0.959-1.086-1.395 (090)

Notes on Statistics:

1) Leagues: PCL 03-07 (AAA), AL 08 (ML), AA 09-11 (AAA), NL 12-20 (ML), PCL 21 (AAA), AA 22 (AAA). Major Leagues (ML): American League (AL), National League (NL). Triple-A Leagues: American Association (AA), Pacific Coast League (PCL).

2) L=League, C=Conversion (Example: BAC=Batting Average Conversion), ISA=Isolated Slugging Average, LISA=League Isolated Slugging Average.

3) Gavy Cravath’s short time (23 games) with Chicago and Washington (both AL) in 1909 is not included in the above chart for obvious reasons of sample size.

4) Isolated SA above comes from simply deducting SA from BA, and not from doing the actual calculation. This could cause an error of plus or minus .001.

5) League conversions are simply calculated by dividing the players BA-SA-ISA by the League BA-SA-ISA. For example, in 1911, Cravath batted .363 and the League average was .268. 363 divided by 268 gives a League Batting conversion of 1.354.

Looking at the career of Gavy Cravath before doing any converting, it appears that he was improving from 1903-1906 (22-25), had his peak years from 1907 to 1915 (26-34) with his absolute peak from 1910 to 1913 (29-32), and faded away from 1916 to 1922 (35-41). In other words, a perfectly normal career pattern for a great and disciplined hitter.
   151. Gadfly Posted: April 30, 2005 at 06:34 PM (#1300273)
ADJUSTMENT PROBLEMS, PART THREE

One of the interesting things about the career of Gavy Cravath is that, using translated BA and SA, his best full season in the Major Leagues should be 1913. In 1913, his translated BA was 1.302 (30.2% better than League) and translated SA was 1.605 (60.5% better than League). The Major League SA translation is the best of his Major League career and the BA translation is the second best of his Major League career, only slightly behind his 1919 BA translation, which was produced in only a half-season.

However, according to the modern measure of batting skill called adjusted OPS (OPS+), Gavy Cravath’s best Major League season was 1915, with an OPS+ of 170, just slightly ahead of his 1913 OPS+ of 169. There are several reasons for 1915 being rated above 1913. One is that the Philadelphia park factor was slightly lower in 1915 (101) than in 1913 (104). Another reason is that Cravath had his greatest translated Major League Isolated Slugging Average (ISA) in 1915 of 2.711 (or 170.1% better than League).

The following table gives Cravath’s age, year played, OPS+, and park factor from 1912 to 1920:

31 1912 118 107
32 1913 169 104
33 1914 157 107
34 1915 170 101
35 1916 146 104
36 1917 151 104
37 1918 105 107
38 1919 207 107, half-season
39 1920 144 105, pinch-hitter

But the most important reason that Cravath’s OPS+ was slightly better in 1915 than in 1913 is, once again, one of adjustment. In 1913, Gavy Cravath was in his second season in the National League. By 1915, he was starting his fourth season in the National League. His knowledge of the pitchers was that much deeper. The reason Cravath has a better OPS+ in 1915 is that his on-base percentage, i.e. ability to get on base, is much greater in 1915 because he is drawing many more bases on balls.

The following table shows Cravath’s plate discipline from 1912 to 1920:

AGE-YR-PA-AB-H-TB-BB-SO-BA-SA-BB% (BB/PA)
31 1912 483 436 124 205 47 77 .284 .470 .097
32 1913 580 525 179 298 55 63 .341 .568 .095
33 1914 582 499 149 249 83 72 .299 .499 .143
34 1915 608 522 149 266 86 77 .283 .510 .141
35 1916 512 448 127 197 64 89 .283 .440 .125
36 1917 573 503 141 238 70 57 .280 .473 .122
37 1918 480 426 099 160 54 46 .232 .376 .113
38 1919 249 214 073 137 35 21 .341 .640 .164
39 1920 054 045 013 021 09 12 .289 .467 .167

This is a very interesting table because it clearly shows the adjustment and counter-adjustment between Cravath and National League pitchers. Cravath comes up in 1912 and plays relatively well against pitchers he mostly does not know. In 1913, Cravath adjusts and has a monster year. In 1914, the pitchers adjust back to Cravath, pitching around him and walking him 50% more than they did in 1912 and 1913. In 1915, Cravath adjusts back, maintaining the discipline from 1914 and recapturing some of his power.

In 1916, age finally begins to catch up to the 35-year-old Cravath. His plate discipline declines steadily from 1915 to 1916 to 1917 to 1918, though he recaptures some power in 1916. After his worst season in 1918, Cravath becomes a part-time player. He has a great partial season in 1919 and a good season in 1920, mostly because his plate discipline reaches an all-time high.

But what if Gavy Cravath had entered the National League in 1904 or 1905, rather than 1912? Obviously, the adjustments and counter-adjustments between Cravath and the league outlined above would have already been made. Throughout the entire 1912 to 1920 time period, his actual hitting skills are in decline. An interesting exercise is to equalize his 1912 base on ball ratio to 1915 (adding 44%) and also to 1913 and 1914. That changes the above table to:

AGE-YR-PA-AB-H-TB-BB-SO-BA-SA-BB% (BB/PA)
31 1912 483 415 124 205 068 77 .299 .494 .141
32 1913 580 500 179 298 080 63 .358 .596 .139
33 1914 582 473 149 249 109 72 .315 .526 .187
34 1915 608 522 149 266 086 77 .283 .510 .141
35 1916 512 448 127 197 064 89 .283 .440 .125
36 1917 573 503 141 238 070 57 .280 .473 .122
37 1918 480 426 099 160 054 46 .232 .376 .113
38 1919 249 214 073 137 035 21 .341 .640 .164
39 1920 054 045 013 021 009 12 .289 .467 .167

In other words, his offense from 1912 to 1914 is increased by approximately 5 percent (5%). After this adjustment, Gavy Cravath’s statistics make more sense from a timeline perspective. His BA and SA are steadily falling. He draws a large amount of walks after his career year in 1913 and he has a slight comeback year in 1917 although still with declining plate discipline. Finally, he has two partial seasons in 1919 and 1920 where he hits well in selected at bats with great plate discipline.

Although this is simply an example, it is clear that adjustment effects do not wash away within a single season (a large factor biasing the conversion rates of Chris Cobb).

In other words, adjustments have to be made for the initial adjustment year and also for several years after the adjustment year. Of course, this is, once again, highly individualistic. A free-swinging hitter will not need nearly as much time adjusting as a disciplined hitter. A good hitter will have to have a larger adjustment period in mid-career than earlier in his career when he is playing on sheer talent. At the end of a player’s career, when he is adjusting constantly to keep up with his declining skills, the adjustment period once again shortens.
   152. Gadfly Posted: April 30, 2005 at 06:37 PM (#1300283)
BASIC CONVERSION RATES FROM CRAVATH

Gavy Cravath twice advanced from a Triple-A quality league to the Major Leagues during his career. The first time was in 1907-1908 and the second time was in 1911-1912. Cravath also twice returned to a Triple-A quality league from the Major Leagues, in 1908-09 and 1920-21. Taking direct conversions from these time frames gives the following conversion factors:

Advancing from Triple-A
BA: 1907 1.278 AAA, 1908 1.071 ML, Conversion Factor=0.838
BA: 1911 1.354 AAA, 1912 1.044 ML, Conversion Factor=0.771
SA: 1907 1.495 AAA, 1908 1.260 ML, Conversion Factor=0.843
SA: 1911 1.784 AAA, 1912 1.274 ML, Conversion Factor=0.714
ISA: 1907 2.379 AAA, 1908 1.954 ML, Conversion Factor=0.821
ISA: 1911 3.079 AAA, 1912 1.918 ML, Conversion Factor=0.623

Returning to Triple-A
BA: 1908 1.071 ML, 1909 1.224 AAA, Conversion Factor=0.875
BA: 1920 1.070 ML, 1921 1.132 AAA, Conversion Factor=0.945
SA: 1908 1.260 ML, 1909 1.363 AAA, Conversion Factor=0.924
SA: 1920 1.308 ML, 1921 1.380 AAA, Conversion Factor=0.948
ISA: 1908 1.954 ML, 1909 1.889 AAA, Conversion Factor=1.034
ISA: 1920 2.046 ML, 1921 2.037 AAA, Conversion Factor=1.004

Of course, there are some inescapable problems with the above conversions. The most obvious would probably be sample size. In 1908, Cravath had only 277 at bats in the American League. In 1920, he had only 45 at bats in the National League; and, in 1921, only 341 in the American Association. These samples are too small.

But these conversions also show how large the problem of separating the adjustment factor (AF) from the quality of competition (QC) can truly be. In 1907, Cravath had been playing for four previous years in the AAA Pacific Coast League. His conversion factors going to the American League were .838-.843-.821 (BA-SA-ISA). Going from the American League to the Triple-A American Association with no experience in that league in 1909, these same factors were .875-.924-1.034. An enormous difference probably completely due to adjustment factor.

Although this is not the place to discuss it in detail, this difference (that players almost always have much lower conversion rates while advancing to the Majors than when they return back to the Minor Leagues) seems to be a constant. Other studies that I’ve done in the 1920s and 1940s show the exact same effect. In other words, adjustment factor is greater going up the Minor League ladder than it is coming down that same ladder.
   153. Gadfly Posted: April 30, 2005 at 06:39 PM (#1300291)
PEAK PERFORMANCE CONVERSION RATES

Of course, this still leaves open the question of how to accurately arrive at a Minor to Major League conversion rate, measuring only the Quality of Competition and not the Adjustment Factor or anything else. One way to do this is to use Peak Performance.

When discussing the Negro League conversion rate in another thread, I broke down Monte Irvin’s career. In the end, I ended up comparing Monte Irvin’s two peak seasons in the Negro Leagues, 1946 and 1947 with his two peak seasons in the National League, 1951 and 1953, prejudicing the comparison against the Negro Leagues as much as possible.

In the end, this calculation resulted in conversion factors of .93 BA and .87 SA. However, in addition to the in-built bias against the Negro League, there was also a serious timeline problem with these conversion factors. The two Negro League seasons were in Irvin’s prime and the two National League seasons were post-prime. KJOK posted a rudimentary aging conversion chart (originally posted on Tangotiger.net) that ended up reportedly adjusting the .93-.87 conversions to .96-.90 conversions.

In reality, I still think these conversion rates may be slightly low.

Back to the actual subject, the career of Gavy Cravath is made to order for a peak-performance conversion-factor evaluation. The two best years of Gavy Cravth’s career are 1911 (age 30) in AAA and 1913 (age 32) in the Major Leagues; and, using these years, the conversions are:

BA: 1911 1.354 AAA, 1913 1.302 ML, Conversion Rate=.962
SA: 1911 1.784 AAA, 1913 1.605 ML, Conversion Rate=.916
ISA: 1911 3.079 AAA, 1913 2.467 ML, Conversion Rate=.801

Of course, these conversion rates are far different from the ones deduced from the very basic conversion rates calculated above. However, a basic problem with these conversion rates is the obvious one: sample size. These were Cravath’s two peak seasons, but one season may have been greater than the other and equalizing the two may be inappropriate.

Fortunately, Gavy Cravath’s career provides a chance to examine this problem also. From 1909 to 1911, ages 28 to 30, Gavy Cravath played in the American Association. In 1909, he had an adjustment year; in 1910, he had a great year; and, in 1911, he had a career year. From 1912 to 1914, ages 31 to 33, Cravath played in the National League. In 1912, he had an adjustment year; in 1913, he had a career year; and, in 1914, he had a great year. These two time frames, 1909 to 1911 and 1912 to 1914, are virtually made to order for a conversion rate comparison and large enough to reduce most problems with sample size.

In fact, the two time frames are almost freakishly similar with only the timing of the peak season being different and the obvious timeline problem that Cravath’s American Association seasons are directly in what should be his prime and Cravath’s Major League seasons are post-prime.

This comparison of these two three-year periods in Cravath’s career produces the following results:

Three-Year Average (AA to Major Leagues)
BA: 1909-11 AAA 1.307, 1912-14 ML 1.179, Conversion Rate=.902
SA: 1909-11 AAA 1.592, 1912-14 ML 1.458, Conversion Rate=.916
ISA: 1909-11 AAA 2.547, 1912-14 ML 2.265, Conversion Rate=.889

These are very interesting results. As compared to the single peak season evaluation, the BA conversion has gone down from .962 to .902, the SA conversion has stayed exactly the same at .916, and the Isolated Slugging Average has gone up from .801 to .889.

Of course, if these results are following a baseball player’s usual aging pattern, these results are exactly what should be expected. From ages 28 to 30, during a player’s prime, the BA should be peaking. From ages 31 to 33, during a player’s immediate post-prime, BA should be receding, but the player should be retaining his power, his SA. In other words, the conversion rate above probably represents the actual quality of competition (QC) factor, altered by a timeline factor.

Using the Baseball-Aging pattern formulas mentioned above (posted on Tango tiger.net), the true Triple-A to Major League conversion factors should be, if Cravath’s career followed a normal aging pattern, about .95 for BA and .96 for SA. On the other hand, aging pattern formulas are all well and good, but each individual player ages at his own rate.

Of course, there is one large variable here that could be warping these results, and that variable would be park factor. However, Gavy Cravath played in very similar parks throughout the 1909 to 1914 period. Both Minneapolis’ Nicollet Stadium and Philadelphia’s Baker Bowl were great hitter’s parks with a short right field porch very conducive to Gaavy Cravath’s opposite-field power-hitting approach [Cravath would fit right in today].

Another variable that could be a problem is, once again, adjustment factor. In 1909, Cravath, who always hit for great power, adopted an opposite field power stroke to take advantage of his home field. It is quite possible that, because of this adjustment, his power peaked slightly late (or post-prime). In actuality, I think this is exactly what happened. I think Gavy Cravath had his peak season for BA in 1911, but his peak seasons for power in 1913 and 1915. But, all things considered, 1911 was still probably Gavy Cravath’s true peak season.
   154. Gadfly Posted: April 30, 2005 at 06:41 PM (#1300297)
RECONSTRUCTING GAVY CRAVATH, PART ONE

To reconstruct the baseball career of Gavy Cravath, the first necessity is to come up with a true conversion rate to turn his Minor League seasons into their Major League Equivalents. So, the question now becomes: ‘Which is the correct conversion rate to adopt?’ Is it the single season Peak Performance conversion rate of .962 BA and .916 SA? Or is it the multiple season Peak Performance conversion rate of .902 BA and .916 SA? Or is it the timeline adjusted multiple season Peak Performance rate of approximately .95 BA and .96 SA?

Being slightly conservative, I have decided to adopt a .96 BA and .92 SA conversion rate to translate Cravath’s Triple-A quality Minor League seasons into MLEs.

Of course, this is slightly higher than my previous studies have shown the conversion rate to be between Triple-A and the Major Leagues; but, in retrospect, it makes perfect sense. Cravath played in the 1900s and 1910s. In my opinion, the gap in talent between the Triple-A quality leagues and the Major Leagues has been slowly and steadily increasing over time. The rates from the 1900s and 1910s should, if that is true, be slightly higher than those rates deduced from the 1920s and 1940s.

With a basic conversion rate for a Triple-A to Major League conversion, it should then be possible to reconstruct Gavy Cravath’s entire career as if he had played almost all of it in the Major Leagues. Unfortunately, there is still one added complication. During his career, Gavy Cravath was forced several times to adjust to new teams and new leagues. If he had played his entire career in the Majors, this would not be true.

In other words, adjustments must be made for Gavy Cravath’s adjustments.

ADJUSTMENT PROBLEMS, PART TWO

Several times in his career, Gavy Cravath entered a new league. Of course, the first time was his first two seasons in the Pacific Coast League, 1903 and 1904, at the ages of 22 and 23. From 1903 to 1904, Cravath improved at the following rate:

BA PCL 1903 1.046, 1904 1.067, Adjustment Rate=1.020
SA PCL 1903 1.200, 1904 1.237, Adjustment Rate=1.031
ISA PCL 1903 1.794, 1904 1.868, Adjustment Rate=1.041

Interestingly, the young Cravath did not improve much from 1902 to 1903. However, this is perhaps to be expected. Cravath is playing at this point on simply sheer talent. He has not yet learned to make the constant adjustments needed to improve his game.

[Interestingly, Cravath had very similar years in 1903 and 1904, ages 22 and 23, improving slightly in the second year. Cravath made a substantial improvement between 1904 and 1905 and then had very similar years in 1905 and 1906, ages 24 and 25, improving slightly in the second year; but them improving substantially again in 1907. This pattern, a player improving and then maintaining a plateau for two or three years before improving again, is very common; much more common than steady improvement.]

The second time Cravath entered a new league for more than one season was in 1909 and 1910, his first two seasons in the American Association, at the ages of 28 and 29. Cravath was now in his prime and improved at the following rates:

BA AA 1909 1.224, 1910 1.342, Adjustment Rate=1.096
SA AA 1909 1.363, 1910 1.629, Adjustment Rate=1.195
ISA AA 1909 1.889, 1910 2.672, Adjustment Rate=1.415

Of course, these adjustment factors are much greater than the adjustment factors in 1903 and 1904. This makes perfect sense. In the prime of his career, Cravath’s batting skill now relies less on sheer physical talent and more on knowledge of the league and its pitchers and ability to adjust to those pitchers and their pitching patterns and adjustments to Cravath. One thing to note is the large improvement in Cravath’s power from 1909 to 1910, possibly attributable to his mastering the opposite field power batting style.

Gavy Cravath switched leagues once again in 1912 and had his first two seasons in the National League in 1912 and 1913, at the ages of 31 and 32. Cravath was now past his prime as a hitter but he improved at the following even higher rates:

BA NL 1912 1.044, 1913 1.302, Adjustment Factor=1.247
SA NL 1912 1.274, 1913 1.605, Adjustment Factor=1.260
ISA NL 1912 1.918, 1913 2.467, Adjustment Factor=1.286

This is a large across the board improvement, as one would expect, greater than the improvement in 1909-1910 in both BA and SA, as Cravath continued to age as a hitter and compensate for his declining abilities with more adjustments.

So, in order to truly reconstruct the Major League career of Gavy Cravath, adjustments have to be made for his adjustments. In other words, a downward adjustment must be made for when he enters the Major Leagues and upward adjustments have to be made for his statistics for each year during his career that he entered a new league (1908, 1909, and 1912). However, this raises an additional question.

That question is: ‘Do the adjustment effects of moving from one league to another wash out after a single year?’ Unfortunately the answer to this question is probably no.
   155. Gadfly Posted: April 30, 2005 at 06:42 PM (#1300306)
ADJUSTMENT PROBLEMS, PART THREE

One of the interesting things about the career of Gavy Cravath is that, using translated BA and SA, his best full season in the Major Leagues should be 1913. In 1913, his translated BA was 1.302 (30.2% better than League) and translated SA was 1.605 (60.5% better than League). The Major League SA translation is the best of his Major League career and the BA translation is the second best of his Major League career, only slightly behind his 1919 BA translation, which was produced in only a half-season.

However, according to the modern measure of batting skill called adjusted OPS (OPS+), Gavy Cravath’s best Major League season was 1915, with an OPS+ of 170, just slightly ahead of his 1913 OPS+ of 169. There are several reasons for 1915 being rated above 1913. One is that the Philadelphia park factor was slightly lower in 1915 (101) than in 1913 (104). Another reason is that Cravath had his greatest translated Major League Isolated Slugging Average (ISA) in 1915 of 2.711 (or 170.1% better than League).

The following table gives Cravath’s age, year played, OPS+, and park factor from 1912 to 1920:

31 1912 118 107
32 1913 169 104
33 1914 157 107
34 1915 170 101
35 1916 146 104
36 1917 151 104
37 1918 105 107
38 1919 207 107, half-season
39 1920 144 105, pinch-hitter

But the most important reason that Cravath’s OPS+ was slightly better in 1915 than in 1913 is, once again, one of adjustment. In 1913, Gavy Cravath was in his second season in the National League. By 1915, he was starting his fourth season in the National League. His knowledge of the pitchers was that much deeper. The reason Cravath has a better OPS+ in 1915 is that his on-base percentage, i.e. ability to get on base, is much greater in 1915 because he is drawing many more bases on balls.

The following table shows Cravath’s plate discipline from 1912 to 1920:

AGE-YR-PA-AB-H-TB-BB-SO-BA-SA-BB% (BB/PA)
31 1912 483 436 124 205 47 77 .284 .470 .097
32 1913 580 525 179 298 55 63 .341 .568 .095
33 1914 582 499 149 249 83 72 .299 .499 .143
34 1915 608 522 149 266 86 77 .283 .510 .141
35 1916 512 448 127 197 64 89 .283 .440 .125
36 1917 573 503 141 238 70 57 .280 .473 .122
37 1918 480 426 099 160 54 46 .232 .376 .113
38 1919 249 214 073 137 35 21 .341 .640 .164
39 1920 054 045 013 021 09 12 .289 .467 .167

This is a very interesting table because it clearly shows the adjustment and counter-adjustment between Cravath and National League pitchers. Cravath comes up in 1912 and plays relatively well against pitchers he mostly does not know. In 1913, Cravath adjusts and has a monster year. In 1914, the pitchers adjust back to Cravath, pitching around him and walking him 50% more than they did in 1912 and 1913. In 1915, Cravath adjusts back, maintaining the discipline from 1914 and recapturing some of his power.

In 1916, age finally begins to catch up to the 35-year-old Cravath. His plate discipline declines steadily from 1915 to 1916 to 1917 to 1918, though he recaptures some power in 1916. After his worst season in 1918, Cravath becomes a part-time player. He has a great partial season in 1919 and a good season in 1920, mostly because his plate discipline reaches an all-time high.

But what if Gavy Cravath had entered the National League in 1904 or 1905, rather than 1912? Obviously, the adjustments and counter-adjustments between Cravath and the league outlined above would have already been made. Throughout the entire 1912 to 1920 time period, his actual hitting skills are in decline. An interesting exercise is to equalize his 1912 base on ball ratio to 1915 (adding 44%) and also to 1913 and 1914. That changes the above table to:

AGE-YR-PA-AB-H-TB-BB-SO-BA-SA-BB% (BB/PA)
31 1912 483 415 124 205 068 77 .299 .494 .141
32 1913 580 500 179 298 080 63 .358 .596 .139
33 1914 582 473 149 249 109 72 .315 .526 .187
34 1915 608 522 149 266 086 77 .283 .510 .141
35 1916 512 448 127 197 064 89 .283 .440 .125
36 1917 573 503 141 238 070 57 .280 .473 .122
37 1918 480 426 099 160 054 46 .232 .376 .113
38 1919 249 214 073 137 035 21 .341 .640 .164
39 1920 054 045 013 021 009 12 .289 .467 .167

In other words, his offense from 1912 to 1914 is increased by approximately 5 percent (5%). After this adjustment, Gavy Cravath’s statistics make more sense from a timeline perspective. His BA and SA are steadily falling. He draws a large amount of walks after his career year in 1913 and he has a slight comeback year in 1917 although still with declining plate discipline. Finally, he has two partial seasons in 1919 and 1920 where he hits well in selected at bats with great plate discipline.

Although this is simply an example, it is clear that adjustment effects do not wash away within a single season (a large factor biasing the conversion rates of Chris Cobb).

In other words, adjustments have to be made for the initial adjustment year and also for several years after the adjustment year. Of course, this is, once again, highly individualistic. A free-swinging hitter will not need nearly as much time adjusting as a disciplined hitter. A good hitter will have to have a larger adjustment period in mid-career than earlier in his career when he is playing on sheer talent. At the end of a player’s career, when he is adjusting constantly to keep up with his declining skills, the adjustment period once again shortens.
   156. Gadfly Posted: April 30, 2005 at 06:44 PM (#1300311)
RECONSTRUCTING GAVY CRAVATH, PART TWO

Now, to reconstruct the lost Major League career of Gavy Cravath, the basic ground rules should be figured out. The first ground rule, as already discussed above, is that the basic conversion rate for all Triple-A quality League BA and SA statistics should be about .96 BA and .92 SA.

Next, which year Gavy Cravath would have debuted in the Major Leagues has to be deciphered. By the conversion factors above, Cravath was a league-average Major League hitter for BA and a 10 percent above league-average hitter for SA in his first year, 1903, in the PCL. In a perfect world, a 22-year-old baseball player with those credentials would surely be given a chance by some Major League team, in hopes that he could improve into something special. So, once again in a perfect world, 1904 should be considered Cravath’s first year as a Major League player.

[Of course, this can be debated; but the general theory is to give a player credit whenever he becomes an average Major League player. Because of the far end of a Bell Curve distribution of baseball talent, an average player is more valuable than the designation makes it sound.]

The second ground rule to be considered is what will be the adjustment factor for Cravath’s first year, 1904, in the Major Leagues? In his original 1903 to 1904 adjustment period, Cravath’s BA and SA adjustments were .980 BA and .970 SA, downwards from 1904 to 1903. In his 1907 to 1908 adjustment period, Cravath’s adjustments were .838 BA and .843 SA downwards, though admittedly with a small sample size. His 1909 to 1910 adjustment rates downward were .912 BA and .837 SA.

Of course, in 1904, Cravath was simply getting by on sheer talent, not his ability to adjust, so the 1902 to 1903 rates are the most relevant. However, if Gavy Cravath had advanced to the Majors in 1904, he surely would have faced the same dislocations he faced in 1908, moving away from California to the East Coast. For these reasons, Cravath’s BA and SA reductions for 1904 will be set at .95 BA and .90 SA. However, no further reduction should probably be made for 1905, considering that Cravath is in the dawn of his career and getting by on talent, not his ability to make adjustments.

A second consideration for 1904 has to be playing time. Playing for Los Angeles of the Pacific Coast League in 1904, Cravath played 210 games for a team that played 216 games to a decision. However, in 1908, Gavy Cravath played 94 games in his first year in the American League; and in 1912, he played 130 games in his first year in the National League. Taking all three years into consideration, Cravath’s games played in 1904 should probably be set at about 125.

Then from 1905 to 1918, Cravath’s games played should be set at his ratio of games played to games played to a decision by his team (example: in 1906, Cravath played 177 games and his team played 184 to decision, thus his games played is set at 148) with the following exceptions: 1908, 1909, and 1912. Since, in each of these three years, Cravath lost playing time due to his changing of leagues but not actual talent, each year will be calculated by the closest full years, with full years one year away counting twice and full years two years away counting once.

[Of course, this is far from perfect and I am not sure whether Cravath actually lost time in any of these years to injury; but Cravath was an unusually durable player.]

The next ground rule concerns his adjustment periods from 1908 to 1912. What is the proper adjustment pattern for these years? In 1910, his second year in the American Association, his BA and SA increased by 1.096 and 1.195, respectively. In 1913, his second year in the National League, his BA and SA increased by 1.247 and 1.260, respectively. I think that the second of these two measurements is slightly fluky and will adopt the first as his true adaptation rate.

A related question to this is how quickly Cravath became totally adapted to his new Leagues? As discussed above in Adjustment Rates, Part Three, it probably took Cravath four or possibly even five years to become totally acclimated to a new League. For these reasons, Cravath’s adaptation factors for a new league should be set at about:

Year 1: BA 1.10, SA 1.21
Year 2: BA 1.04, SA 1.08
Year 3: BA 1.02, SA 1.04
Year 4: BA 1.01, SA 1.02

[These rates also follow a formula not discussed here. Most analysis of conversion rates show that BA and SA generally act as the square or square root of each other. In other words, if BA is increased by 10 percent (110%), SA increases by the square of ten percent (121%); and, if BA is decreased by 10 percent (90%), SA decreases by the square of ten percent (81%) and vice versa. The reason for this is simple. As BA goes up, some singles become extra base hits, increasing SA. As SA goes down, some extra base hits become singles, increasing BA.]

An additional problem is that, in 1908, Gavy Cravath only played for half a season. In his first season in the Majors, he actually played quite well, posting Major League conversion factors of 1.071 and 1.260 for BA and SA. However, if he had been allowed to play the full season, it is obvious that these rates would be different. In 1909, Cravath posted adjusted conversion rates of 1.175 and 1.254 in the American Association (1.224 and 1.363 multiplied, respectively, by .96 and .92). Using these rates for the second half of his 1908 season brings Cravath’s 1908 rates up to 1.123 BA and down to 1.257 SA before the further timeline adjustment.

The final considerations in reconstructing Gavy Cravath’s career are his final seasons from 1919 to 1922. In 1918, Cravath, still playing full-time, played poorly. In 1919 and 1920, he became the field manager of Philadelphia and played very well, but hardly full-time. In 1921, Cravath returned to the high Minors, as a manager, and played far more than in the previous two years. In 1922, he played very sparingly in his final year.

How would Cravath have done if he had not become a manager and still had to rely on his bat to earn his keep?

For purposes of this study, his last two seasons (1921 and 1922) will be combined and yield an adjusted MLE of 1.052 BA and 1.213 SA in 431 ABs. His 1920 conversion rates will remain the same, being slightly better than 1921 at 1.070 BA and 1.308 SA, but his ABs will be increased by the 431 ABs credited in 1921 for a total of 478 ABs. In 1920, his ABs will be doubled to 428 and his conversion rates reduced by using the 1920 conversion rates of 1.070 and 1.308 for these extra ABs, resulting in 1919 BA and SA rates of 1.196 and 1.455, respectively.

In addition, the games played for 1919 to 1921 will be calculated by using Cravath’s 1918 rate of 426 ABs in 121 games played.

Of course, this is hardly a perfect scenario but probably reflects pretty well what would have happened if Cravath had not become a manager. He would have had to fight for playing time in 1919 after his poor 1918 season; but posted a good comeback season. He would have played a little more in 1920 because of his good 1919, and then, still being a good Major League hitter, played slightly less in his final season of 1921 before retiring.

Finally, using all of the conversions spelled out above, Cravath’s career is reconstructed as:

GAVY CRAVATH’S LOST MAJOR LEAGUE CAREER
AGE-YEAR-G-BA-SA-WS-BA-SA
23 1904 125 (0.973-1.024) 16 .242 .330
24 1905 147 (1.086-1.179) 23 .277 .391
25 1906 148 (1.080-1.190) 24 .264 .369
26 1907 148 (1.227-1.375) 31 .298 .425
27 1908 149 (1.235-1.521) 35 .295 .465
28 1909 150 (1.293-1.517) 37 .315 .476
29 1910 150 (1.340-1.619) 41 .343 .547
30 1911 154 (1.326-1.707) 44 .345 .608
31 1912 151 (1.148-1.542) 33 .312 .569
32 1913 150 (1.354-1.733) 44 .355 .613
33 1914 149 (1.215-1.554) 32 .305 .519
34 1915 152 (1.160-1.572) 36 .288 .520
35 1916 138 (1.146-1.341) 26 .283 .440
36 1917 142 (1.124-1.442) 26 .280 .473
37 1918 151 (0.913-1.146) 14 .232 .376
38 1919 122 (1.196-1.455) 23 .309 .490
39 1920 136 (1.070-1.308) 19 .289 .467
40 1921 122 (1.052-1.213) 17 .304 .482

Projected Career Win Shares: 521
Yearly BA and SA for the National League.
Batting Champion: 1910-1911, 1913 (3 times)
Slugging Champion: 1910-1911, 1913-1915, 1919 (6 times)

GAVY CRAVATH IN MODERN TIMES

In 2004, the National League had a 263 BA and a 423 SA and the American League had a BA of .270 and a SA of .433. Just for fun, how would Cravath’s reconstituted statistics look today in a league with a BA of about .267 and SA of about .430?

AGE-BA-SA
23 .260 .440
24 .290 .507
25 .288 .512
26 .328 .591
27 .330 .654
28 .345 .652
29 .358 .696
30 .354 .734
31 .307 .663
32 .362 .745
33 .324 .668
34 .310 .676
35 .306 .577
36 .300 .620
37 .244 .493
38 .319 .626
39 .286 .562
40 .281 .522

He started a little slow and then grew into Albert Pujols.
Not Bad.
   157. Gadfly Posted: April 30, 2005 at 06:48 PM (#1300329)
Before anyone points it out to get the warm glow of smug satisfaction, post 51 is out of order and unnecessary, being repeated in post 55.
   158. Brent Posted: May 01, 2005 at 02:32 AM (#1301226)
PhillyBooster:

Finishing up the "aughts" with 1909. Did anyone else find this helpful?

Yes, thank you! You've put together a lot of interesting and useful information that we will have to try to digest. Though I have to say that after reading Gadfly's recent posts, analyzing these data appears to be a lot more complicated than I had initially though.

Gadfly:

Wow, impressive! It will take me some time to think about this thoughtful research. Here are a few quick, first reactions, though.

- The section "Conversion Rate Problems" is the best summary I've read of the statistical issues that arise in converting performance from one league to another. Thank you.

- Just as a general research strategy, the detailed deconstruction of a single player (I remember you did something similar with Monte Irvin) is likely to be less convicing than a study of a group of players. (That's why I encouraged PhillyBooster to put together data on players who transitioned between the PCL and the majors during the aughts.) The advantage of looking in detail at a single player is that you can consider factors that are unique to the player and might be missed in a broader study. But the disadvantage is that every players' statistics are subject to random variation, and without looking at a group of players it is really difficult to know what may be systematic and what may be random. What's really needed are standard errors (though surprisingly few sabermetric studies report them). You can't calculate standard errors from a sample of one.

- If you are right about the conversion factors for the top minor leagues being much closer to the majors than we have thought, it would imply that we should be giving more serious HoM consideration to outstanding players who spent the bulk of their careers in the minors. Truck Eagan? Frank Shellenback? Jack Ogden? Ike Boone? Johnny Bassler? (just to name a few)

I appreciate your sharing this study with us. Now we need to give it some serious thought.
   159. OCF Posted: May 01, 2005 at 05:25 AM (#1301468)
Finishing up the "aughts" with 1909. Did anyone else find this helpful?

I'm interested in this - and the next several years, the early teens, might be even more interesting - because of the light it might shed on a different issue: the relative strength of the NL versus the AL. The data problem we have is that there was very little direct movement of players between the two major leagues. But players who were teammates or opponents in the PCL moved freely - no, not the right word, not freely, but at least they moved - into both major leagues. This data (and similar data for the AA and the IL) might be the best data we have.
   160. Brent Posted: May 01, 2005 at 05:58 PM (#1302006)
Gadfly -

After some reflection on this fascinating study, here are a few more comments:

- Your general strategy is imagining a counterfactual world in which Cravath enters the majors in 1904 and stays there until 1921 without ever becoming a manager. On the 1950 ballot discussion thread, # 125, I've argued for a different strategy, of not trying to construct a counterfactual but simply evaluate the player's statistics in the leagues in which they played and then make the necessary adjustments for quality and context in comparing players across leagues.

There are two differences in this approach:

1) My approach constrains the evaluation more closely to the player's actual statistics. In your evaluation, because you assume that Cravath makes the adjustment to a different environment only once, you create a career record with a quite different shape and contour. Other voters using the counterfactual approach have done quite different things like assuming away Dihigo's pitching or moving Beckwith to first base. It seems to me that the counterfactual approach, although interesting to think about, simply introduces too many degrees of freedom into the conversion. I like conversions that are more closely tied to the player's actual playing experience (recognizing, of course, that for some events such as military service there is no alternative but to extrapolate).

2) Rhetorically, I also dislike the fact that these conversions always project the Negro Leagues or minor league player to a major league environment. This device can leave the impression that the Negro Leagues or minor leagues did not, of themselves, matter. I have been a strong advocate of the idea that the PCL was just as important to the residents of San Francisco, and the Negro Leagues were just as important to the African American community, as the National and American Leagues were to their fans.

I have to say though, that my position on this does not seem to be in the majority. David C. Jones is the only other voter I recall posting similar comments to mine regarding interpreting these conversions.

A couple of other things make me uncomfortable about your approach.

1) Most players go through some periods of adjustment at various points in their careers. Sometimes it is being traded to a team with very different playing conditions (a pitcher being traded to Colorado), or general circumstances change (the move from the dead ball to the live ball, or the introduction of the DH in the AL) or a pitcher loses velocity on his fastball, or a player is asked to move to another position in the field. Generally, we treat these as simply part of the game, and in evaluating a player no modifications are made to account for these adjustments. It would seem strange to me to single out the adjustments required in moving from minor leagues to major leagues for special consideration.

2) You also assumed as part of your counterfactual that Cravath remained a full-time player rather than managing late in his career. This factor of course affects many other candidates in addition to Cravath (notably, Frank Chance and John McGraw, but many others as well). I think this group has reached a consensus that no special consideration is warranted for players who become managers, and they should be evaluated on their playing record alone.

Another concern regarding Cravath is to what extent his skills may have been specific to parks that had, as you said, a short right field porch. I've seen home/road splits for Chuck Klein, for example, that indicate his success was largely due to an unusual ability to take advantage of the Baker Bowl, and that he was a rather ordinary hitter on the road. I haven't seen similar splits for Cravath, but I've always wondered if he may have experience a similar benefit. If he did, I wouldn't take it as detracting from his performance while playing for the Millers and the Phillies -- I believe in evaluating players in the actual context in which they player -- but I would have a problem extrapolating it to the seasons before 1909 when he played in quite different home parks.

By the way, I don't believe you discussed park factors. I've got a method for coming up with minor league pseudo-park factors (actually run environment factors), see post # 62 in this thread for Cravath's case. Since Cravath's minor league statistics, similarly to his major league statistics, were greatly influenced by his home ballparks, I'd be interested in any comments or suggestions.

Finally, although I am not convinced that MLEs should include allowance for the player's adjustment to new leagues along the lines that you've presented, I do think this is a fascinating study that demonstrates that those adjustments are an important consideration influencing the development of many players' careers.
   161. Chris Cobb Posted: May 01, 2005 at 06:36 PM (#1302150)
I have little to add to Brent's comments, which I think raise all the major questions I would raise about gadfly's highly informative study of Gavvy Cravath.

But since my NeL conversion factors are part of the discussion here, let me say on my own behalf that I share Brent's concerns about sample size and about the idea of including not just one season but potentially 3-4 seasons in an adjustment period.

The former issue could be addressed by more analysis, though it would obviously be time-consuming!

The latter issue is more theoretical; I'd like to see more disucssion of it.

Let me also add as a note to gadfly: since I can see that you have been busy with this study for a while, you may not have noticed that the Dick Lundy MLEs have provided what I regard as a smoking gun on the need for applying different conversion factors to different periods of NeL play. I hope that as more MLEs get worked out, seeing many career paths in parallel, and seeing the MLEs generated for league-leading performances season by season, may provide evidence for making conversion factors more accurate to the varying playing conditions in the NeL from 1920-1948.
   162. TomH Posted: May 01, 2005 at 09:15 PM (#1302737)
A lot of work, Gadfly, and you wrestled with some tough issues. Many thanks.

However....if I read your ##s right, you basically say that Gavy's 'adjustment' cost him 50 to 100 pts of career slg pct, among other things. That seems like a huge thing to swallow.

BP's translated stats show Cravath with a SLG of .559, 4 seasons with 49 to 55 HRs; all very fine ##S. And an EqA of a mere .003 pts above Frank Chance, who gets no votes from anyone. To imagine that Cravath would have put up a >.600 slg (how many in history have done that?), given the career he actually had, is possible. But to make that my best estimate? I dunno...
   163. Gadfly Posted: May 01, 2005 at 10:26 PM (#1302824)
58. Brent-

I completely agree with you that you can't make any generalizations from a sample of one. The whole above study started one week ago when I realized that I could figure out the PCL's seasonal BA and SA from 1903 to 1907 and use it to analyze Cravath's career, which was my main concern.

All the rest of it came in because it was related to my analysis of Cravath and I figured that others could take it and run with it (hopefully).

Of course, conversion analysis is ridiculously complicated, something I was trying to convey. Cravath is very unique . There are very few other examples (if any) of Major League superstar quality players who played several consecutive years of their prime or near-prime (thus reducing the timeline problem) in both the Majors and the Minors.

The odd thing is that, despite having him listed on my Hall of Merit ballot already, I had not realized how unique Cravath was until recently.

60. Brent-

I agree that there are benefits to using a completely factual approach: the main one being, of course, that you do not stray too far from reality.

I'm not sure I like the term counter-factual for my approach, it seems too much like fake or phony, but I understand what you mean.

But I am sure that you will admit that a player, given four or five years previous experience in a league before a season, will have quite different statistics than if that same season was his first season in that same league. That was what I was trying to get at.

In other words, what I did above is more of a possible reconstruction of what might have been. Of course, many people hate what ifs. Also, I freely admit that my 'reconstruction' is just a best case scenario. For all I know, Cravath, if he started his Major League career in 1904, would have peaked in 1912, reconstructed above as an off-season.

Of course, my reconstruction is not a complete what if scenario. Cravath played ball from 1903 to 1922. I just tried to analyze what would have happened in he had played the majority of his career in the Majors.

But I don't believe in Pete Reiser's lost career from 1940 to 1960 or George Sisler doesn't have sinus problems type reconstructions. There has to be some basis in reality.

One thing that I had not realized until I did this reconstruction was how durable Cravath was. As far as I can tell, Cravath would have played a full slate of his team's games from 1903 to 1918.

As for your time frame argument, that again is a matter of personal preference or comfort level.

A good example of this is something that seems to have bothered you: giving Cravath credit while he was managing. For this, I must admit that I have a sort of timeline bias.

In Cravath's time, a player who was employed as a field manager was paid more, sometimes much more, than if he just kept on playing. This is, of course, no longer true. In essence, a player now who became a manager would most likely be getting a pay cut.

This fact alters the careers of players right into the 1950s and, when I analyze players, I try to adjust for this fact. I admit that this is a function of my timeline bias. In essence, I am trying to evaluate players under today's financial considerations.

Once again, this is simply a personal prefernce.

For Cravath, there seemed ample enough evidence of what type of hitter he still was from 1919 to 1922. Something like this would be much harder, if not impossible to do with someone like McGraw, who has compensating factors that 1) he stopped playing, 2) he had a serious knee injury, and 3) he got fat too.

On interesting thing is that I also like to give catchers extra credit, especially if they were good hitters and far back in baseball history. Sometimes these two biases, catchers and managers, combine to really make my analysis difficult.

Two guys who I have a hell of a hard time really evaluating, because of this, are Roger Bresnahan and Frank Chance.

But I am pretty sure that, if neither man caught or managed, their baseball careers would have been considerably different and much more valuable in a retrospective Win Share kind of way.

Finally, I completely agree with you on Park Factors. They are the great unknown in my study, though I do mention them in passing. A full analysis of Cravath's career, with all the appropriate park factor adjustments might tell a different story.

However, it seemed to me that the park factors for the 1909 to 1914 period, the crux of the analysis, are very probably the same or very near the same, canceling one another out.

However, I could be wrong.

61. Chris Cobb-

Of course, the great problem with conversions is sample size and the great problem with sample size is that conversions with large enough sample sizes are actually quite rare.

Cravath, with his matching three season PCL and Major League stints around his prime, is virtually unique. There are some PCL to Major League players in the 1920s with multiple seasons (Waner and Averill both come to mind), but they all have timeline problems since they are young and improving.

As far as your conversion rates go, my main point above is that there are a lot of things that go into a conversion rate, but only one thing that should actually be present and matter (Quality of Competition). As you already know, I believe your Negro League conversions are way off because of these other factors.

If my post above just makes anyone consider that possiblity, I'll be happy.

I haven't looked at the Lundy thread, but I'll happily give it a once over. This Cravath study pretty much consumed all my free time for baseball for the last week since I figured out that I could actually do it.
   164. Brent Posted: May 01, 2005 at 10:46 PM (#1302864)
Gadfly:

Thank you for the reply. Just to be clear, I didn't intend anything pejorative in using the term "counter-factual." I was just trying to find a term that describes a process of imagining a player's career under different circumstances than actually occurred. I agree with you that we can learn from thinking through the consequences of a player's career taking an alternate path, but for my own evaluations I prefer to stay with MLEs that are more closely tied to the actual statistics.
   165. Devin has a deep burning passion for fuzzy socks Posted: May 02, 2005 at 03:06 AM (#1303441)
One problem I have with giving someone a bonus for being a player-manager, is that this was their own decision. I understand that the extra money was hard to refuse under the economic circumstances of the times, but it's still part of the historical record. They had to know that managing wouldn't help their playing, but they still made the choice. If it affects their statistics, well, tough luck.

Also, putting aside the occasional Harris or Boudreau, when teams named a player-manager, they would tend to go for an older player, so it's natural that their numbers would decline. How do you know which effect you're seeing? Yes, Cobb declined in the early 1920s, but can you be sure why?
   166. Joey Numbaz (Scruff) Posted: May 02, 2005 at 06:52 AM (#1303971)
Great stuff Gadfly. Wow.

"To imagine that Cravath would have put up a >.600 slg (how many in history have done that?), given the career he actually had, is possible. But to make that my best estimate? I dunno... "

I hear you Tom - but . . . Cravath's numbers in his 30s are right up there with the Aaron's of the world. He's probably one of the 10-best hitters ever after age 31. His career SLG, which is mostly from age 31-37, is over 1/3 higher than the league average. Whatever one projects, I would think it has to be 'up there'.
   167. Paul Wendt Posted: May 02, 2005 at 10:53 PM (#1305337)
In Cravath's time, a player who was employed as a field manager was paid more, sometimes much more, than if he just kept on playing.

In the aughts, at least, this was true for some good mlb players who became minor league player-managers. I don't know commonly the minor manager was also President or part owner, which complicates the career move.
   168. Paul Wendt Posted: May 04, 2005 at 08:47 PM (#1311118)
'SLYYYYMMDD' is Sporting Life publ date

SL19080523
Bos 0516. Thoney has "lame arm" and will visit Bonesetter Reese, Youngstown OH

SL080530
Bos 0525 (J.C. Morse). a critic says "the fastest lot of players he has ever seen in one club"

The outfield after release of Barrett (injured, 3 games) and LaPorte(?):
Thoney - "as fast as they make them"
McHale, Sullivan - "speedy"
Cravath, Carlisle - evidently covered by the general statement, hence somewhere between McHale/Sulli and Gessler, not plodders
Gessler - "for a big chap"

SL19101105 (ex post 1910)
"Home Run" Cravath, the Boston Red Sox discard, whom Detroit is trying to secure from Brooklyn, walloped the horsehide for 11 hrs [38-13-11 2B-3B-HR]

Cravath finished second in batting average behind Barrett, who played the first 51 games only.
   169. sunnyday2 Posted: May 04, 2005 at 09:18 PM (#1311223)
Re. counterfactual arguments, I share Brent's caution. But:

1. I've argued in the past in opposition to the counterfactual that Dobie Moore didn't play baseball until he joined the NeL. It took me until now to recognize that it is equally counterfactual to pretend Gavy Cravath didn't play baseball except in the MLs. Gadfly counterbalances that pov.

2. OTOH, of course, anyone who finds Chris' NeL MLEs to be generous would have to find these even more so. I don't think one can question that Cravath played ball 1909-11 etc. or that he played with enough skill that he would have had at least some value in the MLs. But how much?

Since I don't know the answer to the question how much, I am inclined to split the difference between 0 and gadfly. I am not prepared to make a technical argument that that is the right answer, but it is an answer.

This of course is exactly analogous to the question of how much value Dobie Moore had with the Wreckers, and the answer clearly is somewhere between 0 and very very much, with very very much being perhaps the same MLE value he has been assigned for his years in the NeLs. In his case, I do NOT split the difference because, as I have argued elsewhere, he never got a shot. Cravath got a shot and some putatively knowledgeable ML manager or other decided he wasn't that good. I incorporate that opinion in my method. Nobody gave Moore a shot or rendered an opinion that is accessible to us today that he wasn't that good.

Likewise Dick Lundy. How much value did he have in the NeL? Well, we don't really know because we don't know how to translate it. In his case, however, I DO split the difference (between OPS+ 122 and 94) because we have data. For Moore we don't.

Anyway, just some different perspectives on conversions.

Finally, however, no way does Cravath get any credit for the ball he didn't play as manager.

Summary: Cravath is in my strong consideration set (top 25-30) as a result of my acceptance that he 1) played ball 1909-11 and 2) had some sort of MLE value. Right now I expect to have him rated somewhere between 15 (IOW possibly on-ballot) and 20.
   170. Devin has a deep burning passion for fuzzy socks Posted: May 07, 2005 at 03:17 AM (#1319006)
All right, you can take this for whatever it's worth (which probably isn't much). I wanted to try and estimate how Cravath's Translated Win Shares would work out to in WARP. My impression was that WARP generally didn't like Cravath as much as Win Shares did, for whatever reason, but I wanted to try and see the whole picture.

My method was to take 3 sets of Win Shares and WARP ratings in Excel, and find the trendlines. Then plug the MLE Win Shares in, and see what we get. The 3 groups were
A) Cravath's numbers from the rest of his career
B) The comparable players identified by Dr. Chaleeko in Post #121, for those seasons.
C) All position players elected to the Hall of Merit playing between 1906-1911, for the full or mostly-full seasons they played in that period.

Here's the results I got (the 1909 numbers here only reflect the partial season MLE)

Group A-WS-WARP1-WARP3
1906-16.7-4.8-4.6
1907-21.3-6.2-5.8
1909-17.3-5.1-4.8
1910-30.9-9.0-8.2
1911-34.1-9.9-9.0

Group B-WS-WARP1-WARP3
1906-16.7-5.8-3.8
1907-21.3-7.1-4.9
1909-17.3-5.9-3.9
1910-30.9-9.6-7.2
1911-34.1-10.4-8.0

Group C-WS-WARP1-WARP3
1906-16.7-5.8-4.0
1907-21.3-7.4-5.3
1909-17.3-6.1-4.2
1910-30.9-10.5-7.9
1911-34.1-11.6-8.8

Average-WS-WARP1-WARP3
1906-16.7-5.5-4.1
1907-21.3-6.9-5.3
1909-17.3-5.7-4.3
1910-30.9-9.7-7.8
1911-34.1-10.6-8.6

One player I personally see as similar to Cravath is Chuck Klein. They both played for the Phillies, took advantage of the Baker Bowl, were power hitters, corner OFs, and are mainly peak arguments. So, here's their WARP records, from best to worst season. I'm leaving out 1906 for Cravath, which seems to be the general practice. First, WARP3, where Klein should have an advantage (my translated seasons for Cravath are starred):

Klein: 11.4  10.7  10  7.6  7.1  5.9  4.9  4.9  3.9  2.9  2.6 1.8  0.6  -0.1 
-0.3  -0.4  -0.6
Cravath: 9.5  8.6*  7.8*  7.1  6.4  5.9  5.7  5.3*  5*  3.9  3.8  2.5  1.4  0.4


Cravath catches up by the 6th season, and when it's all added up, Cravath's ahead 73.3-72.9, which is a trivial difference.

Now, WARP 1
Klein: 12.8  12.2  11.5  9.1  8.5  6.4  5.8  5.4  4.5  3.2  3.2  2.1  0.9  0 
-0.3  -0.4  -0.6

Cravath: 10.6  10.6*  9.7*  8.6  7.9  7.3  7.2  6.9*  6.6*  5.3  4.6  3.7  2.5  0.4


Now Cravath's clearly ahead by season #6, and has a significant overall advantage, 91.9-84.3.

So, by WARP, Klein's ahead on peak, Cravath's probably ahead on career. In Win Shares, of course, Cravath is miles ahead of Klein. (305-235, again starting with 1907.)

Now, when I look at everything, taking into account the assumptions implied in the translations, and the problems with both Win Shares and WARP, I do agree that Cravath is better than Klein. But he's not really that far ahead of him to me, and nobody really thinks Chuck Klein is that impressive a candidate (he's in my 30s). Cravath was at 29 on my last ballot, and he'll move up some more this time, but not enough to make my top 15.

Like I said, take it for whatever you think it's worth.
   171. Mark Shirk (jsch) Posted: May 08, 2005 at 01:47 AM (#1320392)
I finally got around to running Cravath's MLE's through my system. Right now he is off ballot, I have him ranked at #25, which is ten spots higher than I had him before. I was giving him some minor league credit but it wasn't anything specific, more a nod toward a good minor league career.

The closest outfielders to him in my rankings are Browning (23), Bell (24), Veach (28), Berger (30). My top group of OFers are all Cfers (Duffy, Averill, GVH, Roush). Cravath has a very nice peak in my system, one very even with Berger and Browning. The MLE's also make Cravath's extended prime look better than Berger's. Bell has some great career value and I am a bigger Veach fan than most.

I didn't give Cravath credit for 1907, but in my system it woudldt' really change things, he would still be behind Browning in the corner OF queue.
   172. Jeff M Posted: May 15, 2005 at 03:10 PM (#1338177)
I ran a quick search to see which players in MLB history were within 3 WS on either side of Cravath at ages 27, 32 and 35 (Cravath doesn't have MLB totals for ages 29-30). I initially tried to do this for every age/season, but no one qualified.

At those ages, Cravath had 12, 29 and 26 WS, respectively, so I looked for 9-15, 26-32 and 23-29.

Two other players qualified, Dolph Camilli and Luis Gonzalez. I'll list them here from age 23-39, without any comments on what conclusions might be reached regarding whether Cravath is in 500 WS territory and better than Foxx:

Age.......Cravath......Camilli.....L.Gonzalez
23........--...............--............15
24........--...............--............11
25........--...............--............20
26........--...............1.............13
27........12...............10............15
28........2................14............17
29........--...............22............12
30........--...............25............12
31........15...............25............26
32........29...............28............27
33........28...............25............37
34........35...............29............26
35........26...............28............24
36........26...............13............12
37........11...............--............active*
38........16...............4.............
39........2................--............

*Gonzalez has an estimated 6 WS (or so) through the 2005 season so far. If he has the same number of PAs as last year (when he got hurt), then at this rate he'll have about 18 WS for the season.

Cravath had 202 career WS, a 3-year (non-consec) peak of 92, a 5-year consecutive peak of 144 and 26.82 WS/162.

Camilli had 224 career WS, a 3-year (non-consec) peak of 85, a 5-year consecutive peak of 135 and 24.35 WS/162.

Luis Gonzalez has 267 career WS, a 3-year (non-consec) peak of 90, a 5-year consecutive peak of 140 and 21.16 WS/162.

The totals are not season-length adjusted for the search (I don't have season-length adjusted data for every player in history), so Cravath and Camilli are slightly underrepresented (7% or so).
   173. Mark Shirk (jsch) Posted: May 15, 2005 at 05:00 PM (#1338286)
The only way that I can see Cravath as a serious canddiate RIGHT NOW and not a backlog guy who may end up on the fringes of the HOM is if you give him A LOT of extra credit. I think you would have to do what Gad does and give him extra WS, hits, HR's, etc. on some of his 'down' MLB years on the basisusfrom the PCL or AA to MLB. I can't really do this, I can give him credit for the years he should have been in MLB, but nothing extra and as thus he isn't more than a guy near the top of the backlog (#25 for me).
   174. Paul Wendt Posted: December 03, 2005 at 10:24 PM (#1758435)
In 1910, each class A club might lose one player, no more, in the major league draft. The major clubs made 133 selections of 67 players (on average, a selected class A player was selected twice) from 30 of the 38 minor clubs (so 30 selections were approved and 8 clubs lost no one).

Six Minneapolis players were selected 16 times, including Cravath 3 times (by the Boston, Detroit, and New York Americans). Teammate Dave Altizer was selected 6 times and the National Commission "decided that Altizer was subject to draft."

National Commission Secretary John E. Bruce wrote in preface to the listing by class A club, "The system of drafting by lot was used in all of these cases." Multiple selections of one player were normally resolved by lot during that era, covering 1900-1910 and more, iirc. I suspect that Altizer among the six teammates was "decided" by deliberation or formal rule with a uniform lottery deciding only among the clubs who selected Altizer. Does anyone here know?

One good candidate formal rule does not fit the facts:
the player selected most times is subject to draft.
That was almost true in 1910. The formal criterion, if any, was closely correlated with desirability of the player.

Given the basic system (multiple simultaneous selection by major clubs, one from each A club to be approved), the selection conflicts might be resolved entirely by lot, which might be structured in various ways. Today, it would be on TV in two days, one to decide the player drafted from each A club, if any (six ping pong balls for Altizer, three for Cravath, etc); another day to decide the major club that gets each of those players.
   175. Paul Wendt Posted: December 03, 2005 at 10:38 PM (#1758453)
P.S.
I believe the system of approving drafts handled each minor club independently and each major club anonymously. Detroit's record in cases already decided and its record on the playing field were irrelevant.

Sporting Life 1910-09-10 is my source for the particular facts and my general understanding is based mainly on SL coverage.
   176. Brent Posted: May 20, 2006 at 04:42 AM (#2028512)
Looking over this thread, our work on Gavy Cravath’s MLEs went through several iterations, so some of our newer voters may find it difficult to sort through all of the old information. Also, the table formatting of old posts is no longer visible. So I thought I’d re-post the “final” MLEs so that it will be clear what numbers we ended up with. (Note that Gadfly later did another set of MLEs that most of us found interesting, but not entirely convincing.)

I’ve decided to give Cravath major league credit starting with 1906, Cravath’s fourth season in the PCL. For 1904 and 1905 we’ve estimated his MLE OPS+ as 99 and 101. With those numbers he wouldn’t have been the worst right fielder in baseball, but they also don’t suggest ML teams would have been missing much by passing him up. By 1906, however, his MLE OPS+ had risen to 109, which was above the median OPS+ of MLB regular right fielders that season. So it makes sense to me to start giving him credit with 1906. However, if you would prefer to start Cravath’s clock ticking with an earlier or later season, the complete MLE data are available in post # 124.

Cravath’s minor league records for each season were translated to the major league average environment of that season. For the league quality adjustments, the American Association was treated as Triple-AAA quality, while the early PCL was assumed to be between Double-AA and Triple-AAA.

Regarding win shares, those were calculated by Dr. Chaleeko based on my MLEs (see post # 121). However, Dr. C’s WS estimates were based on an earlier iteration of my MLEs that assumed a few more plate appearances. To make them consistent, I’ve rescaled the batting win shares to the latest estimates of plate appearances.

Cravath’s MLEs:

Year Lg  Age   G  AB  R   H 2B 3B HR BB  AVG  OBA  SLG OPS+   WS
----+---+---+---+---+--+---+--+--+--+--+----+----+----+----+----
1906 PCL  25 133 473 60 114 26  5  4 62 .241 .329 .342  109 16.0
1907 PCL  26 138 440 61 119 29  2  6 59 .270 .357 .386  135 21.3
1909 AA   28 115 394 53 109 21  5  4 59 .277 .371 .386  136 17.3
1910 AA   29 150 517 70 151 32  7  9 89 .292 .396 .433  150 28.5
1911 AA   30 153 534 97 171 42  7 19 98 .320 .426 .532  166 32.8 


Combining Cravath’s minor league MLEs with his major league record, here are Cravath’s career statistics

Year  Lg  Age    G   AB   R    H  2B  3B  HR  BB  AVG  OBA  SLG OPS+   WS
----+----+---+----+----+---+----+---+---+---+---+----+----+----+----+----
1906 PCL   25  133  473  60  114  26   5   4  62 .241 .329 .342 109  16.0
1907 PCL   26  138  440  61  119  29   2   6  59 .270 .357 .386 135  21.3
1908  AL   27   94  277  43   71  10  11   1  38 .256 .354 .383 136  12.4
1909 AL
/AA 28  138  450  60  118  21   5   5  79 .262 .372 .364 131  19.8
1910  AA   29  150  517  70  151  32   7   9  89 .292 .396 .433 150  28.5
1911  AA   30  153  534  97  171  42   7  19  98 .320 .426 .532 166  32.8
1912  NL   31  130  436  63  124  30   9  11  47 .284 .358 .470 119  15.4
1913  NL   32  147  525  78  179  34  14  19  55 .341 .407 .568 172  28.5
1914  NL   33  149  499  76  149  27   8  19  83 .299 .402 .499 160  27.7
1915  NL   34  150  522  89  149  31   7  24  86 .285 .393 .510 171  34.6
1916  NL   35  137  448  70  127  21   8  11  64 .283 .379 .440 147  25.8
1917  NL   36  140  503  70  141  29  16  12  70 .280 .369 .473 153  26.2
1918  NL   37  121  426  43   99  27   5   8  54 .232 .320 .376 106  11.5
1919  NL   38   83  214  34   73  18   5  12  35 .341 .438 .640 213  15.5
1920  NL   39   46   45   2   13   5   0   1   9 .289 .407 .467 145   2.0

Total         1909 6309 916 1798 382 109 161 928 .285 .389 .457 147 318.0 


Finally, I’ll recommend that you take a look at the last two posts by Paul Wendt (# 174 and 175). People sometimes ask, “if Cravath was so good, why didn’t the major league teams want him?” As Paul points out, the major league teams did want him. The rules, however, protected the independent minor league teams of the time from having to give up most of their best players by limiting the draft to one player per team. An excellent team, such as the Minneapolis Millers, typically had several players that the majors wanted to draft. Therefore, by losing only one player per year to the draft, the Millers were able to retain Cravath for 3 seasons, including two seasons (1910-11) when he was playing at nearly an MVP level.
   177. Brent Posted: May 20, 2006 at 05:03 AM (#2028551)
Now for my question. Last election Sisler appeared on 33 ballots and placed 2nd, while Cravath appeared on 12 ballots and placed 27th. Why?

Sisler's peak season may have been a little bit better than Cravath's (OPS+ of 181 vs. 172). Sisler's career was also a bit longer career, at least under the assumptions I made in the last post (2055 games vs. 1909), though to be fair to Cravath, Sisler had several major league seasons that weren't nearly as good as Cravath's 1904-05 that I decided not to give him credit for. But do those small advantages for Sisler really make up for the huge difference in career OPS+ (147 vs. 124)? Or Cravath's advantage of 26 win shares (318 vs. 292) in 146 fewer games? Sisler had 7 seasons with 20+ win shares, while Cravath had 9.

Maybe a FOGS can enlighten me, because I sure don't get it.
   178. Brent Posted: May 20, 2006 at 05:16 AM (#2028562)
Correction - Sisler appeared on 33 ballots and placed 3rd.
   179. rawagman Posted: May 20, 2006 at 06:40 AM (#2028578)
I have Cravath ahead of Sisler. Woo-Hoo!
   180. andrew siegel Posted: May 20, 2006 at 12:10 PM (#2028606)
Did we ever do full MLE's for Fournier?
   181. sunnyday2 Posted: May 20, 2006 at 12:15 PM (#2028609)
Well, there is the fact that Sisler's record is real and Cravath's (post #76) is MLE, and one of the more hypothetical MLE this side of Buzz Arlett.

Not that there isn't merit to the MLE record posited here. But it's still hypothetical.
   182. Howie Menckel Posted: May 20, 2006 at 12:56 PM (#2028622)
I vote for both Sisler AND Cravath...
   183. Brent Posted: May 20, 2006 at 01:29 PM (#2028633)
Well, there is the fact that Sisler's record is real and Cravath's (post #76) is MLE.

One of the important contributions we made on this thread was to assemble Cravath's actual minor league record, which as far as I know was not available from any single source. It is posted on # 62 of the previous page. Cravath played 1,603 “real” games in the PCL and the AA and led his Los Angeles and Minneapolis teams to five non-hypothetical league championships.

Cravath's actual minor league totals (all PCL or AA)
   G   AB   R    H  2B 3B  HR  AVG  SLG
1603 5589 893 1643 348 73 114 .294 .444
10 seasons (1903-07, 1909-11, 1921-22) 
   184. Chris Cobb Posted: May 20, 2006 at 01:41 PM (#2028637)
Not that there isn't merit to the MLE record posited here. But it's still hypothetical

It's a lot less hypothetical than, say, the Negro-League MLEs. There's a complete statistical record of a number of games that equals (or in the case of the PCL, exceeds) a major league season. There is a surrounding set of numerous major league seasons that help to verify the level of play attributed to Cravath for his minor-league seasons.

What basis do we have for not accepting the conclusions of these MLEs about Cravath's quality, other than the fact that he was playing for Minneapolis rather than for a major-league team?
   185. Brent Posted: May 20, 2006 at 02:00 PM (#2028645)
Did we ever do full MLE's for Fournier?

See the Jack Fournier thread, # 23 and 24.
   186. sunnyday2 Posted: May 20, 2006 at 05:44 PM (#2028819)
Well, I would just ask the question. How many voters consider Cravath to be a bona fide 318 WS player, same as Jake Beckley? One more than Bob Gibson? 2 more than Stan Hack? With a hell of a peak to boot.

Clearly in the aggregate HoM voters do not regard him as such.

Meanwhile, Willard Brown is a hypothetical 320-380 WS player, depending on the version of MLE WS one chooses to acknowledge. How many regard him as such?

Forgetting the numbers, the idea that Brown is a HoM-caliber player is clearly regarded as less hypothetical than the notion that Cravath is a HoM-caliber player.

Chris, where do you have Cravath on your ballot?
   187. sunnyday2 Posted: May 20, 2006 at 06:18 PM (#2029038)
The other difference between Cravath and W. Brown (or any NeLer), now that it occurs to me, is this:

Nobody that I'm aware of ever said that W. Brown couldn't play ML ball. What they said is, he's black.

There were those, however, who saw Cravath play and who were baseball men who said that Cravath couldn't play ML ball.

Clearly there are voters who take that contemporary opinion as valid to a degree. Otherwise as a 318 WS player with a very high peak he would be a HoMer already.

I'm not saying who's right or wrong, just that there clearly are voters who believe that Cravath's case is hypothetical--it is based on a hypothesis about which there is conflicting evidence.

Certainly there is conflicting evidence about W. Brown, too, or else he would have been elected already. But it is not in a sense hypothetical. Or to put it another way, the hypothesis is that he is borderline, and it is a hypothesis that most voters agree with. We know what kind of player Brown was, we just disagree on whether he is above or below the line.

There is less agreement, I would guess, on Cravath--the hypothesis is that he coulda/shoulda/woulda been a star in the MLs for several more years if given the chance. I would say that is more hypothetical than the case for Brown. Specifically, players with his skill set often do have short careers; they master and then lose those skills with unusual rapidity.
   188. Chris Cobb Posted: May 20, 2006 at 06:30 PM (#2029103)
Chris, where do you have Cravath on your ballot?

I had him at #6 in 1976; I'm expecting he'll be at #5 in 1977. I rate him pretty much where the MLEs put him in my system; down a touch, maybe, because I think outfield defense is overrated slightly in win shares for this era. Here's his position on my 1976 ballot relative to some comparable players who have been under discussion and a couple of other close comps:

1. Willard Brown
6. Gavvy Cravath
7. Ralph Kiner
11. George Sisler

As to the hypotheticals issue: everybody was giving Brown major-league credit because that's what the Constitution requires, whether or not they accepted one or another or none of the MLE versions of Brown as calculated. The Constitution does not require that credit for outstanding minor-league play to be given; I expect that some voters simply aren't giving any. That's a valid choice, but I think it's the wrong choice in this case.

As to Cravath's candidacy as a whole: I don't believe the electorate has erred in not electing Cravath so far, but I think he is being underrated now. He's a candidate who ought to be in the high backlog, with a reasonable likelihood of induction before we finish. Unless something changes in the electorate's treatment of Cravath, there is no likelihood that he will be elected.
   189. OCF Posted: May 20, 2006 at 07:36 PM (#2029438)
Re posts #187 and 188:

Certainly there is conflicting evidence about W. Brown, too, or else he would have been elected already.

-----

Here's his position on my 1976 ballot relative to some comparable players who have been under discussion and a couple of other close comps:

1. Willard Brown
6. Gavvy Cravath


Uh ... if you'll check the 1975 results, we did just elect Brown.
   190. sunnyday2 Posted: May 20, 2006 at 08:07 PM (#2029607)
The other difference between Cravath and W. Brown (or any NeLer), now that it occurs to me, is this:

Nobody that I'm aware of ever said that W. Brown couldn't play ML ball. What they said is, he's black.

There were those, however, who saw Cravath play and who were baseball men who said that Cravath couldn't play ML ball.

Clearly there are voters who take that contemporary opinion as valid to a degree. Otherwise as a 318 WS player with a very high peak he would be a HoMer already.

I'm not saying who's right or wrong, just that there clearly are voters who believe that Cravath's case is hypothetical--it is based on a hypothesis about which there is conflicting evidence.

Certainly there is conflicting evidence about W. Brown, too, or else he would have been elected already. But it is not in a sense hypothetical. Or to put it another way, the hypothesis is that he is borderline, and it is a hypothesis that most voters agree with. We know what kind of player Brown was, we just disagree on whether he is above or below the line.

There is less agreement, I would guess, on Cravath--the hypothesis is that he coulda/shoulda/woulda been a star in the MLs for several more years if given the chance. I would say that is more hypothetical than the case for Brown. Specifically, players with his skill set often do have short careers; they master and then lose those skills with unusual rapidity.
   191. sunnyday2 Posted: May 20, 2006 at 10:45 PM (#2029835)
>everybody was giving Brown major-league credit because that's what the Constitution requires, whether or not they accepted one or another or none of the MLE versions of Brown as calculated.

But how much MLE credit? Not "everybody" really and truly had him in the 320-380 range or, like I said, he would have been elected more easily.

Voters make judgments whether to accept MLEs or not for every player, NeL as well as other. It's not unconstitutional to say, e.g., that "I think that Willard Brown would only have played 10 years in the MLs. He wasn't ready when he joined the NeL and he would have flamed out 5 years earlier," or something like that.
   192. sunnyday2 Posted: May 20, 2006 at 10:47 PM (#2029836)
I do agree, btw, that if one accepts a Cravath MLE career from 1906-1920 continuous, then he surely is in Kiner-Sisler territory, maybe higher, what with more career than Kiner, a peak that is comparable to either one, and a decline that is vastly less precipitous than Sisler's.

I go hot and cold on Cravath having 15 years as a ML star, however.
   193. Chris Cobb Posted: May 21, 2006 at 01:17 AM (#2030053)
I wrote:

Here's his position on my 1976 ballot relative to some comparable players who have been under discussion and a couple of other close comps:

1. Willard Brown
6. Gavvy Cravath


OCF wrote:

Uh ... if you'll check the 1975 results, we did just elect Brown.

Yes, we did, but we did it 1976: I was reporting the rankings on my most recent ballot because Cravath was being discussed in comparison to Brown.

Brown will not be on my 1977 ballot, being no longer eligible :-) .
   194. Brent Posted: May 21, 2006 at 03:28 AM (#2030157)
There were those, however, who saw Cravath play and who were baseball men who said that Cravath couldn't play ML ball.

Who said he couldn’t play ML ball? I’d be interested in seeing a reference, if indeed one exists.

The fact that Cravath was once traded/sold from Washington to Minneapolis shouldn’t be construed as evidence that he was considered unable to play at the ML level. More likely, it simply indicates that he was considered more valuable to Minneapolis than he was to Washington.

Many modern fans fail to comprehend the relationships that existed between the minor leagues and the major leagues during the first three decades of the twentieth century. During that period the best minor league franchises were often more valuable (in economic terms) than the less successful major league franchises. Compare Minneapolis and Washington in 1909—Washington’s population was only 10 percent larger (331 thousand for Washington vs. 301 thousand for Minneapolis in the 1910 census). Washington had never had a winning record and usually placed 7th or 8th. There were no broadcast or licensing revenues; ticket sales were all there was—and Washington placed last in AL attendance every season from 1906-09. Minneapolis, on the other hand, was putting together a championship team that would win pennants in 1910 and 1911. In an era before television and radio, the AA was just as important to baseball fans in Minneapolis as the AL was to fans in Washington. I haven’t seen AA attendance records, but it seems entirely possible that a championship Minneapolis team could have drawn more fans and earned more revenue than a cellar-dwelling Washington team. If Cravath was worth more to Minneapolis than he was to Washington, it would have been perfectly sensible for Washington to trade or sell him even if they thought he was capable of playing well at the ML level.

In modern baseball, the economic values of major league and minor league baseball are separated by an order of magnitude. In Cravath’s era, however, their values overlapped. One needs to understand this aspect of the history of baseball in order to understand how excellent players—players like Lefty Grove, Paul Waner, Joe DiMaggio, and, yes, Gavy Cravath—could have spent years playing for independent minor league teams after they had established their ML bona fides.
   195. Joey Numbaz (Scruff) Posted: May 21, 2006 at 08:53 AM (#2030266)
Thanks for reviving this thread Brent.

I've been a big Cravath supporter myself, I had him #2 behind Beckley in 1976.

I'd urge the people who have joined the project in the last year or so who may have written Cravath off without further examination to take another look at this thread.

Especially since we are backlogging it up of late, I think it's really important to take a look at Cravath, he's the kind of player this project was hoping to find when we started the whole process.
   196. sunnyday2 Posted: May 21, 2006 at 11:34 AM (#2030274)
The management of the Boston, Chicago and Washington clubs all had the rights to Cravath and they all passed.

I am supposed to infer that they all thought he was a great player?
   197. rawagman Posted: May 21, 2006 at 12:08 PM (#2030282)
No. You are to inferm that they goofed. The passed on a great player so he could play in a level where he was clearly head, shoulders, knees and toes above the competition. Even though he proved that he was at least as useful as what they already had when he was given a brief trial.
When someone finally realized that he'd be pretty good at MLB level, he was better than that, which should show us that he was not merely an early incarnation of Calvin Pickering, or some other AAAA stud.

That is why he deserves at least some MiL credit on top of his monster hitting abilities.
   198. andrew siegel Posted: May 21, 2006 at 01:12 PM (#2030298)
He deserves minor league credit--not extra credit, but simply credit--because he played those seasons. This is not the major league baseball Hall of Merit, but the baseball Hall of Merit. We usually write off minor league seasons for 2 reasons: (1) most minor leaguers are not playing at a level that gets them many points towards the HoM and (2) most stars have a few seasons where they would have been ok major leaguers and one season where they would have been good before they get promoted, so leaving those seasons out of our calculations goes out in the wash. But, for players who either, (1) played for a long time in the minors at a level where they would have been very good major leaguers (usually b/c/ of misjudgments by major league squads but not always) or (2) stars who spent an extra year or two in the minors after their breakout years, minor league credit isn't extra credit, it's just giving them their due.

That having been said, I'm not fully sold on Cravath. While WS thinks he was a major league superstar, WARP sees him as just a good major leaguer (except for 1 season). It all depends on what you make of slow, defensively-challenged OF's who can mash. While I normally like such players, it is worth noting that both WARP and contemporary opinion are more skeptical of the value of such a player during the deadball era.
   199. Brent Posted: May 21, 2006 at 01:27 PM (#2030304)
The management of the Boston, Chicago and Washington clubs all had the rights to Cravath and they all passed.

I am supposed to infer that they all thought he was a great player?


No, I didn't say they thought he was a great player. I was debating your earlier comment that they thought he couldn't play ML ball. Every year ML teams trade away dozens of players, not because they can't play ML ball, but because they don't fit the team's current needs or the team wants to acquire another player.

Cravath unquestionably had an unusually shaped career. At almost every stage, he surprised on the upside. How many players with a career OPS+ of 147 (according to MLEs in # 176) don't have a season with OPS+ of 150 until age 29? How many players (without assistance of steroids) have their best seasons at ages 32 and 34? I can understand how Boston might have passed on him. I really can't understand what Chicago was thinking, but it looks like they were hoping Cravath could play center field and were disappointed when he couldn't. And Washington? I don't know the details of the transaction that sent Cravath from Washington to Minneapolis, but (as I said earlier) I think it was most likely simply a case of a poor franchise cashing in on a good player because he was worth more to a richer team that hoped to win a championship. It happens every year.
   200. sunnyday2 Posted: May 21, 2006 at 01:54 PM (#2030307)
Good discussion.

Just to return to base, what I said earlier was that the evidence on Cravath is mixed. There is reason to think he could have been a 318 WS player, and there is reason to think that 180 (the actual) is more like it. This is an unusually large range of possiblities. The fact that not one but three teams passed on him in '08-'09 seems significant, however.

By comparison, I said the evidence on Willard Brown is not really "mixed" per se. It's just that the evidence on Brown says he is borderline.

These are completely different kinds of uncertainties.
Page 2 of 3 pages  < 1 2 3 > 

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
Adam M
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 1.1953 seconds
49 querie(s) executed