Chris Cobb (# 92) wrote:

1) adjust Sisler's 1918 and 1919 seasons to 154 games

Chris, it appears that you may have neglected to adjust Chance's 1903 season to 154 games. (The NL played a 140-game schedule that season.)

On league quality & Bancroft vs. Sewell:

Bancroft is hammered in the WARP1 to WARP2 conversion and Sewell is not, but not all that much of the hammering is from league weakness: I think he gets docked maybe 2 or 3 wins more than Sewell does.

Bancroft gets hammered because the shortstops get a huge amount of fielding runs above replacement prior to the lively ball. They probably get 10-15 runs a year more FRAR added to FRAA than they do after the lively ball. Bancroft thus is losing a lot of value in the conversion to "all-time" conditions. Sewell starts at shortstop right after the lively ball, so he doesn't take this penalty. Moreover, shortstop continues to lose a bit in the all-time conversion through the 1920s, but third base, in the late 1920s, doesn't lose anything in the conversion to all time, so Sewell, although he is playing the less valuable defensive position, loses practically no fielding value W1 to W2 after he switches to third base.

Bancroft loses 32.2 wins going from W1 to W2. He loses 71 BRAR, 43 FRAA, and 169 FRAR on top of the FRAA
Sewell loses 5.3 wins going from W1 to W2. He loses 53 BRAR, 27 FRAA, but he _gains_ 32 FRAR on top of losing his 27 FRAA. Yes, his WARP2 FRAR are 5 runs higher than his WARP1 FRAR. He gains 29 FRAR during his years at third base.

So WARP sees Bancroft's NL as about 3.5 wins worse than Sewell's AL over the course of their careers. But Bancroft's earlier start and the NL's slower conversion to the lively-ball game means he loses a lot of his value in the shift to all-time conditions, while Sewell does not. What he loses early in his career he gains back, and then some, during his third base years.

I'm not making an argument here, just trying to describe what is going on in the numbers, since league strength came up in the discussion.

Chris, it appears that you may have neglected to adjust Chance's 1903 season to 154 games. (The NL played a 140-game schedule that season.)

Thanks for the catch. I will be reposting, I hope tomorrow, a broader comparison of Sisler to a number of high-peak hitters, and I will repost Chance with the proper conversions to 154 games. I'll also use his 1902 rather than his 1909 season to get the 7 year set, since he was a better hitter in 1902.

Wow Chris - now I'm much closer to being re-convinced that Bancroft should be ahead, as opposed to just 'even'.

Bancroft's play in the deadball era had value - I don't believe in deflating it because in the future SS wasn't as important. Also helps reconciliate what I thought their reputations were. Very important and I'll keep it mind when comparing pre and post deadball SS's.

This may be enough to get Bancroft ahead of Rizzuto on my ballot actually. I never realized there was that big of an impact.

The death list above reminds me that I need to take another look at Buddy Myer too.

One more on Bancroft and Sewell:

If you apply only the league-quality adjustments,

Bancroft's career WARP is about 100
Sewell's career WARP is about 96

I say "about" because there seem to be some small factors that I can't account for that affect the exact conversion of runs to wins in the WARP system, and I can't judge how they would apply precisely in these cases.

If I were to use these numbers as my basis for ranking Sewell and Bancroft, I think I would here prefer Sewell on peak, but the two are quite close in value by this way of looking.

I haven't submitted a prelim in many years, but I probably haven't revamped my ballot this much in many years either. My first tip-off was that my consensus score had move from Fringe Fest to Tony Bennett, and upon reflection I realized I have gotten away from my previous peak orientation somewhat. So...

Actually the pitchers and the older backlog didn't change much. It was the newly eligibles of the past 15-20 years that moved, mostly.

The previous rating is from 1973 because it is more directly comparable to '75, whereas '74 had a couple of anomolies at the top.

1. Dobie Moore (was #1)
2. Ralph Kiner (6)
3. George Sisler (2)
4. Nellie Fox (10)--go figure, this is where he ends up via revised method
5. Rube Waddell (4)
6. Pete Browning (5)
7. Jose Mendez (9)
8. Minnie Minoso (12)
9. Tommy Bond (8)
(9A. Stan Hack [15A.])
10. Willard Brown (11)

11. Addie Joss (7)
12. Dizzy Dean (23)
13. Charley Jones (15)
(13A. Earl Averill [16B])
(13B. Clark Griffith [17A])
14. Edd Roush (27)
15. Eddie Cicotte (21)

16. Chuck Klein (25)
17. Charlie Keller (22)
18. Don Drysdale (new)
19. Joe Gordon (14)
20. Jim McCormick (37)

21-25. Bresnahan (41), (Doerr [17A]), (Keeler [45A]), (Ruffing [30A]), Stephens (24), Duffy (17), Doyle (18)

Dropped out of top 20: Redding, Gordon, (Doerr), Duffy, Doyle, Williamson.

Doesn't change my ballot tremendously but the PHoM backlog is considerably shaken up. This year I now have Hack, Dean, Averill, Griffith and Roush as my top 5 possibilities instead of Gordon, Hack, Doerr, Averill and Duffy.

This is still permutating, however. The new look also moves Drysdale up from #12 to #7 among pitchers.

Yes, Bancroft was the better fielder. However, Sewell had 159 BRAA, Bancroft 92 BRAA over this stretch. It doesn't come close to covering the batting difference.

Joe, you didn't address my point (repeated above). It's nice that the WS evidence confirms that Bancroft was the better fielder, but I had already conceded that.

If you apply only the league-quality adjustments,

Bancroft's career WARP is about 100
Sewell's career WARP is about 96

Without the league quality adjustments,

Bancroft's career WARP is about 113
Sewell's career WARP is about 105.

Bancroft's career WS is 269.
Sewell's career WS is 277.

So for career, you can pick and choose your system. I rate them about even.

OTOH, by prime or peak, it's no contest. Sewell.

>Sewell had 159 BRAA, Bancroft 92 BRAA over this stretch.

Of course, here is the English language translation of Chris' post #2 above:

It is all just a bunch of gibberish, anyway.

In #72, kthejoker says:

Sandy Koufax was a first-year electee by this group. He may not have been a unanimous selection, but achieving 54% of all possible points in Year 1 seems to me that he was more of a no-brainer than a borderline candidate

Sorry, but 54% in year 1 is far away from no brainer. It is actually fairly poor for a HOM electee. Even Whitey Ford's 79.1% is probably not no brainer, but it destroys Sandy's number. In fact, even Larry Doby (62.1%) and Enos Slaughter (60.7%) were elected in Year 1 with more support than Koufax, as were non no-brainers Pee Wee Reese (69.8%), Hal Newhouser (78.6%), Willie Wells (78.4%), Bullet Rogan (79.9%), Zack Wheat (76.9%), Pete Hill (60.0%), Amos Rusie (69.8%).

Here is my list of no-brainers, culled by taking first-place electees who achieved 90% of all possible points, and second-place electees who achieved 80% of all possible points.



Surprisingly missing are Billy Hamilton (89.1%) and Frank Baker (82.6%).

Biggest no-brainer surprises: Deacon White, Paul Hines, Eddie Plank.

Tris Speaker's percentage is most affected by the other players on his ballot. He would be in the 90% range were Eddie Collins not on his ballot. In fact, he was the reason that I bumped the 2nd place percentage to 80%.

>It is all just a bunch of gibberish, anyway.

I didn't mean Chris' post, BTW, I meant WARP translations into the so-called all-time context.

Surprisingly missing are Billy Hamilton (89.1%) and Frank Baker (82.6%).

Cool list. A few 2nd ballot guys on there (ECollins, SCrawford, Plank), but they had steep competition their first years.

Hamilton was before my time, but it looks like there was still a fair amount of pre-1898 backlog that the electorate was sifting through. Ezra Sutton and Joe Start being the two guys that stole the most votes from him. He is very close (89.1) to Ron's cutoff.

Looking at the ballot thread from 1928 gave some insights as to why people were wary of Baker. 1928 was in the midst of a long candidate drought and its hard to believe -- in retrospect -- that guys like McGinnity & Wallace were taking votes from Frank Baker. Voters seemed wary of Baker's two seasons missed in mid-career that took time away from an already short career.

I think kthejoker's original comment may have referred to the "margin of victory" per se. Koufax's 54% was indeed quite low, but it was a comfortable 21% ahead of 3rd place. It was pretty obvious early in voting week even without tracking the ballots that Koufax was getting inducted.

Oh... 1972 was an elect-three year... so Koufax was a little bit more than 21% out of 4th place.

Well, it is clear that there was great reluctance on the part of a significant portion of voters (though I had him No 2) to vote in Koufax in a weak year.
Perhaps it would be useful to factor in pct of being on the ballot at all, or in top 5, or top 10. Clearly Koufax was NOT considered a no-brainer, whereas a guy who for 2-3-4 years was on the greater majority of top-5 ballots in effect had greater true support.

Paul Wendt - link for you, and anyone else interested in HOMer Jim O'Rourke

O'Rourke's home to be demolished?

Table of top Major-League hitting candidates, seven-year consecutive peak.

To try to assess whether Sisler has a good claim to be the best hitter available over a seven-year consecutive peak (his strongest peak measure), I compiled a variety of views on the seven-year peaks of the top major-league hitters eligible. I included anybody I thought might possibly be close to Sisler: I’d be happy to add any other players. I included McGraw, Elliott, Doyle, and Childs not because I thought they might be as good, but because they are among the best “glove position” hitters eligible, so I thought it might be handy to see how close in value they are to the top hitters at “bat positions.” I threw in Ryan, Duffy, and Van Haltren because I was curious.

No league-quality adjustments. All seasons pro-rated to 154 games.
Players listed in order of EQA. This is an organizing tool, not a claim that this lists the players in order of merit.



*Notes on consecutivity

For Charley Jones, I treated his partial season in 1880 as a full season (prorating from team games played at the time of his suspension to 154) and counted 1880 and 1883 as consecutive.
For Charlie Keller, I treated his 1945 season as a full season, prorating to 154 games. I treated his 1943 and 1945 seasons as consecutive.
For John McGraw, I treated his 1895 and 1897 seasons as consecutive, skipping his 1896 season, which he lost almost entirely to typhoid (?) fever.
For Gavvy Cravath, I counted his 1908 and 1912 seasons as consecutive, skipping his seasons in the American Association

Analysis.

The data here make it clear that Sisler is not obviously the most meritorious hitter available over a seven-year consecutive peak. It’s not obvious who is the best. I would say that Browning, Keller, and Kiner all have arguments for that honor, depending on how you quality-adjust Browning, and how much weight you place on playing time.

Sisler is clearly top 5, either just ahead of or just behind Charley Jones, depending on how you quality-adjust Jones, and he’s not far behind any of the three ahead of him, unless you don’t quality-adjust Browning at all.

How you would rank Sisler’s peak overall against the other players here is another matter, which I won’t try to assess rigorously here, although I will say that I don’t see how Sisler could rank ahead of Keller, who was a very good defensive outfielder. Browning and Kiner lose ground with respect to Sisler on defense, and Charley Jones at best holds his own. Among the more distant trailing players, Klein doesn’t have the defensive value to catch up. If Wilson had been a good defensive centerfielder, he might have passed Sisler, but he was a bad centerfielder. Roush was average in center, but he has playing time issues. Berger was at least average in center, but he is farther behind on offense. McGraw and Chance have huge playing time deficits to make up.

How Sisler’s peak compares to the top peaks of “glove position” candidates and pitching candidates is another issue I won’t try to address here. I only conclude that Sisler looks, by his hitting peak, like a ballot-worthy candidate when he is compared to the other top hitting peaks.

I must say that this study makes me more reluctant to trust win shares for any player who was consistently on very good or very bad teams. Chance’s bws/154 is way out of line with his documented performance, and you can’t just credit his smarts, because McGraw has the same performance and nowhere near the win shares. Also, Keller has a better rate than Browning? No way that isn’t affected by team context . . .

I knew reporter Les Carpenter as he worked his way up in the business, and he is indeed a good man...

Great analysis, Chris.

I'd like to make the case for Cravath. Although he doesn't make your top 5 hitting peaks, he's not far behind. And we need to keep in mind that a) your analysis excludes his age 28-30 seasons in the AA; b) Cravath's MLEs indicate that for 2 or those 3 seasons his hitting was equivalent to his best major league seasons; and c) your other top 5 peak hitters didn't do too much outside of their top 7 seasons (except for possible blacklist credit for Jones), whereas Cravath played 4 more high-quality seasons in the PCL and the AA.

Any thoughts as to what a realistic AA discount would do for Browning. I would think it would have to be a year by year discount, can't really just blanket X% on the AA.

I think I'm going to be bumping Keller (was 63) as war credit effectively turns his into more than a 7 year peak.

This makes me very comfortable with my placement of Cravath, Kiner and Jones, who have more than these 7 years on their resume.

I'm fine with my placement of Sisler at #32 and the relative ranking of Indian Bob at #23, considering he adds 6 more productive years. Sisler's peak just isn't as great as it appears on the surface, and there is basically nothing else.

I'm going to return Hack Wilson to the consideration set. A bad CF is still better than an average LF for one. He'll slot in below Sisler, but not that far below.

I need to bump Childs up some. 49th was too low.

Good stuff Chris, thanks!

I think I've got a re-worked prelim . . .

1. Beckley (3)
2. Cravath (4)
3. Easter (5)
4. Drysdale (n/e)
5. Pierce (6)
6. Kiner (7)
7. C.Jones (8)
8. Gordon (15)
9. Van Haltren (11)
10. Rizzuto (9)
11. Howard (12)
12. Fox (10)
13. Bancroft (-)
14. Trucks (13)
15. Leach (14)

16-20. Minoso, Schang, Bridges, Sewell, Boyer
21-25. Stephens, Leonard, W.Brown, Walters, B.Johnson
26-30. Moore, Monroe, Lombardi, Trout, Mackey
31-35. Taylor, Ryan, Keller, Bartell, Waddell

Dave Bancroft, Dick Bartell, Buddy Myer, Ed Cicotte, Dolf Luque return to the consideration set, which now numbers 81.

Addie Joss drops from #72 to out. I have him down in the Chesbro, Marquard, Haines group now. He was better than them when on the mound, but pitched far fewer innings relative to his peers. Because he wasn't terribly durable his peak isn't great, despite his gaudy ERA+. And quality wise he was equivalent Waddell or Coveleski, not Dean, Walsh or Koufax.

Why does everyone make a big deal about the unearned runs behind Waddell? Joss had the same %, about 2/3 of his runs were earned.

Iron Man McGinnity had nearly 30% of his runs unearned.

Have we compared Waddell to other great pitchers from his era, or did we just kind of run with what seemed like a high number because of his reputation?

I was just looking at Waddell's 1901 season (he gave up a serious amount of unearned runs that year and finished 14-16.
He spent the majority of the season with the Cubs. It seems that the 1901 Cubs had very poor defense, buth in MIF and in the outfield - much worse than league average. Waddell, clearly the Cubs ace that season, whether giving up grounders or flyballs, batters had a decent chance of the fielder flubbing.
Imagine Waddell wasn't such a K king?
Interesting to note that the Cubs 2B for half the '01 season was none other than Cupid Childs - looks like he was trying to make the Rube look bad. Steal some votes from him....

Brent,

I agree with you (and with Joe) on Cravath. I think that with proper MLE credit, he has more outside his peak than most of the other players I've listed. He's been mid-ballot for me for a long time.

The top hitters on Chris' list are mostly players I've voted for and have in my PHoM. I'd be happy to see any of them get elected. Having said that, and assuming the consensus would only agree on one or two, how does Sisler stack up?

Just for the record I have Kiner, Sisler and Browning as the tops among this group. Here are what I see as the pros and cons.

Browning--I discount AA year by year, from a hefty .35 discount in the beginning to parity in the middle and about .25 at the end, by which time Pete has moved to the PL and NL anyway. Even with the discount his numbers are too big to ignore. It takes a timeline not to support Pete, IMO.

Keller--not enough outside the peak, though I agree his peak is pretty awesome. I have him around #20-25 with Cravath but behind Kiner, Sisler, Browning, Jones and Roush.

Jones--needs credit for two blacklist years plus a non-timeline approach to make it, but I do have him in my PHoM and he is still on my ballot some years.

Kiner--way too awesome not to elect, IMO. Not much more than Keller outside of the peak, but it's all in 150 game chunks. 3 chunks of 50 games (Keller) don't have the same pennant potential.

Wilson--not enough outside of the peak.

Sisler--has a lot of value outside the peak. I mean basically when he is not George Sisler he is Jake Beckley.

Klein--more outside the peak than Keller, Kiner and Wilson. Klein's standing depends basically on rejecting Win Shares as penalizing him for playing on horsebleep teams. I do reject the WS
analysis.

Chance
McGraw--not only are these two short careers, there's not enough durability within seasons.

Cravath--a special case to be sure. It all depends on whether you give him MLEs for his MiL years. Even with those MLEs you also have to be a peak voter to support him. I do support him somewhat, meaning he is stuck in my second 15.

Roush-- not just a hitter, probably has more defensive value than any of the other OF on the list. A solid candidate IMO.

Berger--I blow hot and cold. The hardest on this list to get comfortable with.

Bottom line--Sisler has 14 years of 10+ WS and a peak that stands up among if not over the rest. Roush had 13 such years but a lower peak, Jones had 12 such seasons only if you give him the 2 blacklist years, Browning 11 such years, Kiner and Berger and Chance and Klein all 10, Keller only 9.

Of course I'm a peak voter but Sisler's 7 years >25 WS is not matched by anybody else on this list, nor are his + 7 years >10. Look at his 3-5 year peak and he is not that special (though in rarified company), look at prime and he already moves ahead, and then try and find somebody who combined that prime with a 14 year career of >10 WS. You can't find that combination elsewhere, I don't think.

Interesting to note that the Cubs 2B for half the '01 season was none other than Cupid Childs - looks like he was trying to make the Rube look bad. Steal some votes from him....

I have read in 1900 that he will retire at the end of the season. I think I haven't read that he is as good as ever, which seems to be said sometime of almost everyone who is over the hill. Rookie 3Bman Bill Bradley has complained that Childs is too fat to cover second.

Browning--I discount AA year by year, from a hefty .35 discount in the beginning to parity in the middle and about .25 at the end, by which time Pete has moved to the PL and NL anyway. Even with the discount his numbers are too big to ignore. It takes a timeline not to support Pete, IMO.

Well, the other way not to support Browning would be to accept the win-share evaluation of him, apply an AA discount, and then place significant weight on career value. That's what I have always done, and I have never supported Browning. I am reconsidering my reliance on win shares in his case, but I still have problems with his lack of career value and his playing time issues. I think Charley Jones is the most meritorious pre-1893 bat candidate available.

Joe Dimino Posted: April 22, 2006 at 03:10 AM (#1981580)
I think I've got a re-worked prelim . . .

ahem, Roush not in the top 35?

Obviously you are still in rework mode :)

I would agree that they are close. I now rate Browning over Jones but Jones went into my PHoM first. Browning's OPS+ of 164 AA discounts to about 148, while Jones' 150 is about half AA though in his case it is in a slightly stronger AA (no 1882) but still probably discounts to about 145-146.
Browning's WS adj to 154 games AND AA discounted:

Browning 42-37-32-31-28-28-23-23-21-18-16/11 years +10/313 total
C. Jones 46-36-36-36-28-26-26-26*-26*-25-23-14/12 years +10*/348

*This is WITH 2 blacklist years at 26 each, which I have always regarded as problematic. C. Jones is already in my PHoM WITHOUT blacklist credit as I am a peak voter and 46-36-36-36 is not ambiguous. But neither is 42-37-32-31 and 313 for the era. And that is WITHOUT any playing time adjustments for Pete, not that I advocate any.

Still I agree that they are close. I now rate Browning over Jones but Jones went into my PHoM first. Browning's OPS+ of 164 AA discounts to about 148, while Jones' 150 is about half AA though in his case it is in a slightly stronger AA (no 1882 and no 1889) but still probably discounts to about 145.

For that matter, here are the OPS+ head to head in the AA beginning in 1883:

Browning (age 22-28) 183-76-90-51-78-64-(98)
C. Jones (age 33-38) 146-66-56-30-7-(36)

Granting the age difference this doesn't prove Pete was better but it doesn't prove the opposite either.

adding OWP to part of Chris Cobb's informative table:

Player ...... yrs . EQA BWS OWP
..............................per154
Browning 82-88 .348 29.92 780
Keller..... 39-46 .332 29.98 764
CJones... 76-84 .331 26.53 733
Kiner..... 47-53 .331 25.70 728
Wilson... 26-32 .323 25.31 710
Sisler.... 16-22 .322 27.17 737
Klein..... 29-35 .321 23.98 710
Chance. 02-08 .320 32.04 763
McGraw. 93-00 .320 26.73 740
Cravath. 08-17 .317 25.11 718
Roush.... 17-23 .315 26.91 699

I must say that this study makes me more reluctant to trust win shares for any player who was consistently on very good or very bad teams. Chance’s bws/154 is way out of line with his documented performance, and you can’t just credit his smarts, because McGraw has the same performance and nowhere near the win shares. Also, Keller has a better rate than Browning? No way that isn’t affected by team context

I thnk much more of it is the difference bewteen OWP and EqA, if you compare the two above. I would trust OWP, which has been adjsuted for pre-1920 ball, more than EqA, which tires to use the same measure in 1885, 1905, and 1960. OTOH, even by OWP, Chance's BWS/154 are still a little high. But he is definitely underrated by OPS+ and most typical measures. Ditto McGraw.

If Dave Bancroft is going to get another look in comparison to Joe Sewell, why not Rabbit Maranville?

He has the highest WARP1 total of any eligible player (I think). His career offensive rates are hurt by his long period of weak hitting as a starter for terrible Braves teams while he was in his late 30s and early 40s, but if you give him appropriate war-credit for 1918 (+120 g, +8.4 WARP1), his career from 1912-1927 is a very near match for Bancroft's.

Maranville, 1930 games, 111.3 WARP1
Bancroft, 1913 games, 112.6 WARP 1

And Maranville goes on to add another 37 WARP1 in his long decline phase as a light-hitting defense specialist. He was a slightly lighter hitter than Bancroft, but even better in the field, 159 FRAA, as opposed to Bancroft's 118.

I think the electorate has largely dismissed from consideration the long-career, high-defense players like Maranville, Herman Long, and Lave Cross, and I'm not sure why. Sure they are not great peak candidates, but neither is Jake Beckley or George Van Haltren, and they continue to attract support, and I would argue that Long and Maranville have better peaks.

Food for thought, I hope!

That should be "another 27 WARP1" above.

Revised prelim:

1. Mendez (was 3)
2. Gordon (4)
3. Trouppe (5)
4. Drysdale (new)
5. Oms (7)
6. Redding (8)
7. Pierce (6) - a closer look, in the context of evaluating Drysdale, hurt him ever so slightly. He's still a clear HoMer.
8. Boyer (new) debuts here, on the strength of his consecutive peak.
9. Schang (9)
10. Sisler (11) - Singles advance runners; walks seldom do.
11. Fox (10) - 2B is not as difficult as SS.
12. Kiner (12)
13. Minoso (14)
14. Mackey (15)
15. Sewell (13) - never had to play against his strong NeL contemporaries.

16-20: Cravath, Beckley, Rizzuto, Leach, W. Brown.
21-25: Howard, Roush, Bridges, Bancroft, Willis.

I don't know much about Willard Brown, so I repeat last year's plea: Will a fan of Brown's make a strong case for his inclusion?

Even more deserving of re-inspection is Adolfo "Dolf" Luque. I haven't ranked him for lack of data. But what was he doing before he got a fair shot in MLB at age 27? Was he better than Mendez?!

There was a thread with lots of information about Luque.
Look for 'Luque' in the title.

I'm not finding the thread about Luque. Paul, could you help me?

http://www.baseballthinkfactory.org/files/hall_of_merit/discussion/burleigh_grimes_and_dolf_luque/

did that work for you, Ardo?

Okay here is a WS comparison between my favorite peak hitter, Charlie Keller, and George Sisler, the electorate's favorite peak hitter.

Here are all of their WS seasons above 15 WS in descending order, fully adjusted for schedule (up to 162 games) and war discounts (10% off 1943 for Keller). I am giving Keller credit for 31 WS in 1945 (22 missed due to the war) and 28 in 1944, they will be in parentheses.

Keller: 36,34,33,32,25,23 (31,28) (8)
Sisler: 35,30,30,28,26,25,23,20,17 (9)

With my war credit, which I don't think is too generous, Keller has five seaons above 30 WS (MVP level according to James) and one that is really close. Sisler has three at or above 30 WS and one that is close. Both are within the margin of error. Sisler has one more seasons above 15 WS (which I use as my average), Keller has 8. they both have 8 seasons above 20 WS (which James calls All-Star level).

As peak voter, Keller looks decidedly better, not A LOT better but there is enough of a different between them to place Keller higher, even after adjusting for Keller's playing on a good team his whole career. Sisler has more outside of those nine years, but they aren't very good seasons, merely filler. Career voters will appreciate Sisler's filler, but I am very confused how we can have these two 30+ spots apart. I would hope that most Sisler voters are voting for him based on his peak and prime and both what I have here and what Chris Cobb has above should serve to prove that Keller's peak/prime is just as impressive if not more so.

So why is Keller so much lower than Sisler?

By the way, Sisler is on my ballot and will make my PHOM someday, it isnt' that I am not a fan of his, just that i dont' see how he is much better than Keller, Kiner, or even Browning.

revised prelim

Drysdale and Browning are still PHOM.

Gordon, Wynn, Rixey, Doerr, Boyer are the next five in my PHOM backlog.

1. Childs
2. Drysdale
3. Keller - I have moved him up after our talks about peak hitters
4. Duffy - Moved down a bit (from 2 to 4) after our recent discussion about him, GVH, and Ryan.
5. Redding
6. Moore
7. Kiner
8. Walters
9. Browning
10. Gordon
11. Boyer
12. Trouppe
13. Dean
14. Sisler (moved down after talking about him)
15. Howard

There could be some more movement, I am not all that comfortable with Boyer or Trouppe that high, yet at the same time I am not sure I want anyone else that high either. Just wanted to let everyone know how our discussions have affected me.

for all those who think Charlie Keller's 7 year peak was better then Sisler's I think you better re think your WWII discount

Yest,

His best year was 1943, he earned 38 WS. I took 10% off of that, which is 4WS and gave him 34. Do you really think that I should take off more than 10%? Does anyone else take off substantially more? How large is your WWII discount yest?

Keller's peak is better because he hit for power, drew a bunch of walks, was more valuable defensively than Sisler. Sisler hit for a very high average and was thus very valuable, but Keller's other advantages push him slightly ahead.

In another thread someone mentioned that Waddell's contemporaries also had large numbers of UER. Can someone run some sort of chart on this? As I redo things, this may make a big difference for Waddell as he is a ballot canddiate for me.

jschmeagol - I took a sampling of Waddell's contemporaries focusing on pitchers who, like Waddell, had very good K/9 and WHIP rates.
I don't have the time right now to isolate the Waddell focus years, but take this for what you will:

Some pitchers whose careers were centered (at least half) in the 1900's
Name years ER's UER's UER%
Waddell 1897-1910 711 352 33.1
Mathewson 1900-1916 1133 483 29.8 - his % was really good in 06-07
Ames 1903-1919 934 377 28.8 - better post 1910
D. White 1901-1913 808 307 27.5
M. Brown 1903-1916 725 319 30.6
C. Young 1890-1911 2147 1020 32.2
Ed Walsh 1904-1917 598 279 31.8
Joss 1902-1910 488 241 33.1

The electorate needs to seriously consider the possibility of minor league credit for Charlie Keller. His minor league record is genuinely impressive and suggests that he was an MLB All Star-caliber player from the moment he entered the pros (full stats are at the bottom of each link):

1937 Newark Bears

"The 1937 Newark Bears’ roster was filled with future major leaguers. Their most sensational player was 20-year-old outfielder Charlie (King Kong) Keller, a first-year pro from the University of Maryland. Keller led the International League in batting (.353-13-88), runs (120) and hits (189) and tied for the lead in triples (14). He was named Minor League Player of the Year by The Sporting News.:

Keller OBP and SLG were .428/.541. The Sporting News 1937 Minor League Player of the Year, having completed this outstanding season at the highest level of the minors, was sent back to Newark for all of 1938.

1938 Newark Bears

Whereupon Keller hit .365/.465/.569 with 22 home runs and 108 RBI. He led the International League in RBI, hits (211), walks (108) and runs scored (149). He finished second in batting average to teammate Buddy Rosar, who hit .387 in 323 at-bats.

Since we have full data for these two seasons, I am going to run MLEs for them in a little bit. The MLEs will be based on Bill James converting seasonal batting lines to different run contexts using Runs Created. Basically, I will figure out Keller's short-form RC for 1937 and 1938, multiply it by .90 and adjust his statistics dowanward propotionally. This method is explained by James in the Willie Davis comment of the Abstract.

An error on Keller's 1938 RBI total:

As the stats are displayed in the link I posted, the BB table is next to the HR table. I confused the BB for RBI. Keller actually had 129 RBI, finishing second in the International League.

I don't see how a 20 year old in the minors is an exception to anything and should be given credit, no matter how well he raked. Only in cases where something seriously beyond a player's control kept him in the minors and he has already shown bonafide major league ability in the pros, can I justify minor league credit. I can't think of a good case for minor league credit since the end of WWI.

What's next? Little League credit? Old-timers allstar softball credit?

I will do a little more research into Waddell's UER %'s today. I will take every season in which he was one of his team's top 2 IP men and compare his % to the other pitcher in the team's top 2. I believe that if the %'s are close, then Waddell should not be punished for his team's poor defenses. If there is still a discrepancy, then there might be room for a negative mark against the Rube.

CHARLIE KELLER INTERNATIONAL LEAGUE MLEs

Methodology
1. I use a conversion rate of .90 for Keller's International League Runs Created. I leave the number of outs the same and then adjust the other numbers (H, TB, BB) downward on a proportional basis using the quadratic equation (explained on pages 740-742 of the Bill James Abstract).

2. For Keller's 1938 season, I used a conversion factor of .81 (.90^2) to take into account the fact he was repeating the league.

3. For defense, I have him as a "C" outfielder: 2.10 dWS/1000 innings. Using a conservative average of 8.50 defensive innings per game, Keller is credited with 2.59 dWS in 1937 and 2.68 in 1938.

4. The short-form Win Shares are regressed.

YEAR LG AGE PO AVG OBP SLG G PA AB H TB BB ops+ sfws
--------------------------------------------------------------------
1937 IL 20 of .338 .412 .511 145 590 524 177 268 66 139 28.2
1938 IL 21 of .334 .431 .519 150 645 551 184 286 94 145 30.1</pre>

Is this a good estimate of Keller's ability? I think so. The OPS+ numbers are in line with what he did in his first two years (144, 141) before taking that small next step to a ~165 OPS+ level. As far as Win Shares are concerned, he posted 22 and 24 in his first two years while playing 111 and 138 games, respectively. Here is how his seasonal Win Shares per 150 games look from 1937-1940:

1937: 29
1938: 30
1939: 30
1940: 26

Keller starts at 218 career Win Shares. With IL credit, he gets up to 276. But what about war credit?

For the 1945 season, Keller earned 11 Win Shares in 44 games. Extrapolating that out to 136 games (his seasonal average to date), is up to 34 Win Shares for 1945. Regressing that to his career average per 136 games, his 1945 season was worth 30 Win Shares.

For Keller's 1944 season, we can take the average of the two seasons before service (1942, 1943) and the two seasons afterward (1945, 1946) to come up with an estimate of his Win Shares and games played for 1944.

1942: 34, 152 G
1943: 36, 141 G
1945: 30, 136 G
1946: 31, 150 G
Average: 33 WS, 145 G

For 1944 and 1945, Keller gets 52 additional Win Shares in 237 games for war credit. Adding that to the 276 Win Shares of actual play and International League credit, Keller is now at 328 career Win Shares.

Here is how his career now looks:
1937: 28
1938: 30
1939: 22
1940: 24
1941: 32
1942: 34
1943: 36
1944: 33
1945: 30
1946: 31
1947: 10
1948: 8
1949: 5
1950: 3
1951: 2
1952: 0
Career: 328 in 1702 G
WS/162: 31.22

For what it's worth, the Keller projections leave six players tightly grouped in my all-time left field rankings. The Top Five stand out: Williams, Bonds, Musial, Rickey, Yaz. But the next six players are all grouped within five percent of each other: Delahanty, Raines, Shoeless Joe, Burkett, Keller, Simmons.

In every season between 1902 and 1909, Rube Waddell finshed the season in the top two of his team in IP.
The following is a comparison of the top two pitchers in IP in each of those seasons on Waddell's teams.
Between 1902-1905 and 1907, pitcher #2 is Eddie Plank. In 1906, it is Chief Bender. Harry Howell in 1908 and Jack Powell in 1909.

Year Rube (RA ER URA%) Pitcher 2 (RA ER URA%) % Difference (Rube's perspective)
1902 90 63 30 140 110 21.4 +8.6%
1903 109 88 19.3 128 89 30.5 -11.2%
1904 109 69 36.7 111 86 22.5 +14.2%
1905 86 54 37.2 113 87 23 +14.2%
1906 89 67 24.7 98 67 31.6 -6.9%
1907 115 68 40.9 115 84 27 +13.9%
1908 93 60 35.5 103 68 33.4 +2.1%
1909 78 58 25.6 83 56 32.5 -6.9%

Totals 769 527 31.5 891 647 27.4 +4.1%

Not such a huge difference between him and his immediate contemporaries, is it?

The next step. I'll add 4.1 percent of his unearned runs back to his earned run total, using 4.1% as a base for his whole career. (I know, it's a little bit sloppy, but probably not that much)
So, as is, 66.9% of all runs allowed by Rube were earned. Add 4.1% and we have 71%. We'll leave the numbers allowed as is.
So Waddell still pitches 2961.3 innings. He allows 1063 runs, 755 (rounded up) earned. ERA: 2.29 (real life 216). His translated ERA+, career, is now 127 (real life 134). Not as impressive, but still impressive enough for my tastes.

I've said it before and people either thought that I was joking or they thought what I said was a joke, one or the other.

Apropos of #44 above...what about MLE credit for George Sisler's record at the University of Michigan. The very next year after he completed his eligibility as a Wolverine he was in the MLs and contributing.

How would this be substantially different from giving Charley Keller MLE credit for his years in Newark?

And is it substantially different to give Keller credit for his play in Newark, as opposed to, say, giving Gavvy Cravath credit for his PCL years and/or Hank Sauer (a future MVP) for pretty much a decade as a Triple Crown threat in AAA? You bet it is. Keller was not "held back" any more than 1,000 players of the pre-expansion era. Cravath was, Sauer was. Sisler, of course, was not.

OTOH I cannot argue that one should give MLE for MiL play to Earl Averill and not Keller.

But anyone who seriously entertains MLE credit for MiL for Averill and Keller certainly ought to be looking at Dobie Moore.

I don't think that much credit is justified. It gives Keller bonuses for full seasons when he was only playing partial ones and I've already explained my view on his minor league numbers. Once the minors became the minors as we know them, credit will not be given by me.

I wouldn't credit Sisler either. Not for college. I do give credit to Cravath. I also give credit to Fred Carroll. I am accepting any future applications for minor league credit. Keller's application has been summarily rejected.

Marc,

I see the mission of the Hall of Merit to evaluate the entirety of a player's professional record to judge the meritoriousness of his career. The hypothetical case of college credit for Sisler acts outside the bounds of this mission.

The other major difference between the Sisler hypothetical and crediting Keller for his minor league performance is that we have a complete statistical record for Keller that we can analyze. For his 1937 and 1938 seasons, I analyzed them using methods (conversion and regression) that the electorate has long accepted for Negro Leaguers. If anything, we can be more sure about these projections considering the completeness of Keller's minor-league record and the documentation of his MLB career.

For the record, I did give MiL credit to Earl Averill. I will consider Hank Sauer when he comes up and I do have Gavy Cravath in my top 30. As for Dobie Moore, I am one of his strongest supporters.

Very interesting that our top 2 pitchers are Jose Mendez and Dick Redding. It's been a while since I looked at their MLEs. Just as a refresher.

Jose Mendez

Mendez threw about 8 years of "normal" career, breaking down in the 8th year. After a 2-year break, he became a part-time "Sunday" type pitcher for another 9 years about 3 of which seem to have included what I will figuratively call an ERA-eligible number of IP. So essentially he had 10 years with an ERA-eligible type of workload. After the first 7 years, they were scattered out over a period of about 11 years. (He played a lot of SS after the first breakdown of his pitching arm.)

Win Shares 257/40-36-31-31-28-21-17-11

ERA+ 121/156-43-28-27-26-26-16-9-1

The Win Shares are not very similar to any other candidate. I mean Drysdale leads him 258-257 but had 12 years > 10 WS, while Mendez had just 8 then pulled in a few WS a year for many years. For the peak he is more comparable to Wes Ferrell, each of whom packed a lot of WS into about 8-9 years (Ferrell 213 in 9 years, Mendez 215 in 8). But again, the rest of Mendez were spread out, the rest of Ferrell's concentrated. Cicotte, Welch, Walters and Mays are all within 10 career WS but all have 10-13 seasons of ≥ 10 WS. So Mendez is actually quite unique among the pitchers who are still eligible.

But of course he is. The WS that Chris C. calculated are for pitching only and apparently he was a creditable SS during that 2nd half of his career. If as a pitcher he is comparable to Drysdale and Ferrell, sorta, and if you count his total value as a SS and hitter, he is clearly ballot-worthy (and IMO top-of-ballot-worthy). If you discount his years as a SS is non-MLE, well, he is still comparable to Drysdale and Ferrell, unless you subscribe to the uncertainty theory (discount his record due to the speculative part of the method).

Dick Redding

Compared to Mendez, Redding is "the NL career candidate." He just kept on ticking for 18 years, though unlike Mendez he scattered a lot of good and bad years throughout. More like an Early Wynn or Red Ruffing except that his best years came early where their worst years mostly came early on. But still an overall up and down pattern. And of course his workload declined over the years.

Win Shares 267/40-33-27-27-21-19-19-14-13-11
Don't have ERA+

He has that one 40 WS season like Mendez, 5 20 WS seasons vs. Mendez with 6, but 5 more 10-19 WS seasons vs. Mendez with just 2. But for his 18 apparently ERA-eligible seasons vs. Mendez with only 11, he only beats Mendez 267-257 on career WS.

Redding ends up most comparable to Carl Mays among current vote getters but with a much higher best year (40-35). For peak he is also similar to Bucky Walters (Redding 40-33-27-27 vs. Walters 38-32-29*-27), then both fall off to 21 and 20 respectively. (Walters' 29 is with WWII discount.) Redding beats Walters 267-252 on career, despite Walters having 12 seasons ≥ 10 WS vs. Redding's 10 years. Redding has 8 additional seasons of <10 WS.

The lack of MLE ERA+ makes comping Redding harder than Mendez. But while Redding is obviously the career guy and Mendez the peak guy, the two have in common the fact that both had numerous years < 10 WS. This is something that MLers didn't generally have the luxury of doing, with an occasional exception like Teddy Lyons. Take all of that away and Mendez is more of a 225 WS guy (rather than the actual projection of 257) and Redding is more of a 240-245 guy (rather than the actual projection of 267). For this reason I can see why somebody might in fact discount the total somewhat.

I don't discount the NeL values, however. This is the value they had in the NeL and I prefer to take their NeL accomplishments at more of face value (including Mendez' years as a SS, though I can only guess at how much value that really had, there are no projections). As a peak voter, I am downgrading Redding a bit, however. All else being equal I would prefer some of the guys who had 3 or 4 30 WS seasons and who don't rely so much on those < 20 WS seasons for so much of their career value. Mendez' peak holds up to another look, OTOH, and will remain high on my ballot with Rube Waddell.

A fresh look at all of the pitchers also gives a boost to Clark Griffith who now looks like Jose Mendez + to me. Other pitchers who look better today than they did before include Walters, Welch, Grimes, Mays, Jim McCormick and Willis, though none of these will be on my ballot.

Sunnyday on Sisler - "I mean basically when he is not George Sisler he is Jake Beckley."

What?????

George Sisler when he wasn't George Sisler didn't have a single season OPS+ that was within 15 of Jake Beckley's career average.

George Sisler, when he wasn't George Sisler was Travis Lee. 3 years of 100-110 OPS+, and the rest dreck.

James, the only way that I can see fit to credit Keller for his minor league play at age 20-21 would be:

1) He had a big enough year at age 19 that he would have been promoted in most organizations.

2) He was cleary and obviously blocked by being in the loaded Yankee organization.

Do those apply?

Good point on Maranville - I will look at him also.

I've looked at Marty Marion, and as rawagaman says, I've summarily rejected his application.

As far as Waddell, based on everything I've seen here, I'm reducing his penalty for unearned runs to nothing. Those numbers fall well within the realm of random chance.

By the way, getting back to the Sisler/Beckley comment - I hope the general consensus on Beckley isn't that he was as bad as Sisler post-peak and just played forever, if so, that's a really bad misconception by the group.

He was a very, very good player for a very, very long time.

Mike I've got Roush around 50 IIRC - right below Mike Griffin and Hugh Duffy. Don't have access to my rankings right now.

I've given him another look already, but I'll double it check it as a courtesy to you later in the week :-)

Joe - even if he was blocked at his own positiion (clearly and obviously), the Yanks could have dealt him to a team that could have used him sooner for a piece the Yanks didn't have. It is very possible, and even likely that Keller was ready at age 20-21.
To credit him with a handful of win shares may be fair. To credit him as being an All-Star at MLB level is absurd, discount or not.

I agree with your sentiments on Beckley. His career was all prime. A plateau, but a very high plateau.

Just reread #42 James, and I see that Keller was a first year pro in 1937. I'm going to have a hard time giving him credit for that year - unless it can be shown that comparable amatuers in that time frame were going straight to the big leagues, and Keller didn't get that opportunity because extreme big league talent (or entrenched 'proven veterans') were blocking him.

I could be convinced to give him credit for 1938 though.

Rawagman, I see what you are saying, but the Yankees didn't deal guys like Keller for missing pieces. They never have. Except for in the early 1980s :-)

They deal guys like Brad Halsey. They keep guys like Keller and Rizzuto down on the farm for an extra year or two. They stick Mariano Rivera in the bullpen for a year when they sign Dwight Gooden, they don't trade him. Things like that.

They have generally (again, 1980s excepted) realized that it's not worth it to trade a guy like Charlie Keller's career away for stopgap to fill a hole somewhere.

I'm sure you are familiar with the term 4A. The can't miss prospect who just doesn't make it?
Tearing up the mionors does not make a major league All-Star. It doesn't even always make a amjor league bench warmer.

Thanks James and rawagman!

James,

I cannot give credit to Keller for 1937 as I do believe that you need a good season in the minors for your team to notice you. It is the standard I have used this whole time, which has lopped off years for the likes of Minoso, Cravath, and Averill. However, I will be giving him credit for 1938. To me, being stuck in the Yankees' system is something that a player cannot control, if he had been signed by the Browns he may even have been playing in mid-1937.

Sunny,

I think that University of Michigan credit for Sisler is different from MiL credit because going to the University of Michigan instead of playing pro ball is a choice that was made by Sisler, whereas playing in the minors when he was clearly ready was not a choice made by Keller. It is a fine line I admit but the line must be drawn somewhere.

Overall, assuming that Keller was not MLB ready in 1936, which we must do unless we have evidence to the contrary, I think that he only deserves credit for 1938. Still, that is another season that could be called MVP level and it was certainly an Al-Star level performance.

Rawagman,

Thanks for the UER data on Waddell. If appears to me that Waddell did allow a higher percentage of UER to ER but the difference isn't so large as we have assumed to this point. Odd how Waddell got the bad rap while Addie Joss hasn't seeing as how they are near exact contemproraries with the same percentage of UER. This may be enough to push Rube onto my ballot this year.

RE: Mendez and Redding

I have Redding in my top 5 and my PHOM, so I don't need any convincing there. However, I have a tough time putting Mendez over Waddell, they pitched a comparable number of innings and Waddell has the much higher ERA+ with a comparable peak. Since rawagman has shown that Waddell DOES NOT deserve a huge discount because of his UER, the only thing that I can see that would push Mendez ahead of Waddell is his time at SS. However, I find it highly unlikely that he would have been doing this had he been playing in MLB or the MiLs. Is there any evidence that Mendez ability as a SS is what kept his IP totals so low for those years? That is about the only way that I can see giving him credit for that.

While I realize that we should in theory be giving credit to NeL players for what they did and not what they might have done in MLB (the old Joe Rogan and Martin Dihigo debates), I think in this case, giving Mendez SS credit would be giving him an advantage because he played in the leagues that he did. How many white pitchers may also have been decent SS's had they been latino or black?

Things is, Charlie Keller did make it and I think it is pretty clear that he would have been a All-Star level performer had he been signed by anyone but the Yankees and maybe a few other teams (Cardinals?). I don't think it is right to punish Keller because he was signed by the Yankees and not the Browns or Phillies.

Right on Keller jschmeagol - Keller obviously wasn't 4A. He was a big-league All-Star.

I've got to really wade through the old threads on Mendez and Redding this week too.

And guys that tear up the minors as young as Keller rarely end up 4A anyway. Those are the guys that tear up the international league at age 25.

Retrosheet got updated this weekend. New seasons include 1921-22 (NL only), 1957-58 (both leagues) and 1959 (now both leagues instead of just AL).

All the new eligibles have 2-3 new seasons there.

Top returnings with *new* retrosheet seasons are Pierce (#16), Fox (#20), Roush (#27), Grimes (#32), Traynor (#39), Maranville (#61), WCooper (#72) and Bancroft (#78).

And guys that tear up the minors as young as Keller rarely end up 4A anyway.

Sporting News voted Keller Minor League Player of the Year in 1937. I don't know who won in 1938.

Seems reasonable to give Keller MLE credit for 1938. I wouldn't give him a full 30 WS because it seems odd to project more for him in 1938 than he ended up getting in 1939 or 1940. For 1937, I'd say no. That was his first pro season out of college. I think he'd spend the full year in the minors no matter what system he was in.

Keller didn't get 30 WS in '39 or '40 because he didn't play many games in those seasons. Was this due to injury or because he had to share time with Heinrich and Selkirk. If the latter I can see giving him full credit in 1938 and some other credit in '39 and '40. If the former, I think he should get full credit for 1938 (he played the entire season) and obviously no credit and '39 and '40 as injuries happen.

I guess it depends no how far you want to push the, "He played on a deep team and didnt' get a full shot until later" thing with him. Something tells me that over the course of his career, this is offset by the extra WS or two a season that he earned by not hitting against the Yankees pitching staff and defense.

In case anyone missed it, retrosheet updated. This comes in handy for a new candidate like Roy Face. Want to know how he won 18 games in '59? Well . . .

- 4 times he came in with the Pirates leading, blew the lead, and they came back.
- 9 times he came in with the game tied, and held on until they won.
- 3 times he came in with Pitt losing and they came back.
- 1 time I can't tell - no PBP for a game against the Cubs, and he might have let an inherited runner tie the game.
- 1 time he came in with Pittsburgh leading 14-11, pitched three innings of scoreless ball, the lead never changed hands, and he got the win. Must be one of those screwy "scorer discresion" moves where Face got it because no one else could get anyone out.

I once heard an interview w/ Jerome Holtzman saying that he invented the save because Face's 18-1 record was a product of a bunch of vultured wins. Looking it up, I think Holtzman's observation may have been off. He had some vultured wins, but mostly it was tie games.

Interesting discussion for a "sunny" Sunday afternoon.

Re. Mendez and Waddell, I have Waddell as my #1 pitcher with Mendez in the top 3. Mendez' play at SS has never really factored into that rating but it's a reasonable tie-breaker.

I just gave Charlie Keller an extra year for 1938 and ran him through my new system. I'm trying to get back to my historical peak voter approach, which I had lost somewhat in recent years. I got all starry eyed over the longer careers that are becoming more commong place. Well, anyway, even with an extra year I still see Keller as no better than the 6th best hitter available. By the time you get the arms and gloves in there, he is probably off ballot. Still he may move (guessing now) from around #20-22 to around #16-18.

But if I adopt an approach that lets Charlie Keller rate that highly, Hack Wilson rates even higher. Caveat emptor.

If you google for "Jerome Holtzman Face" the first hit returns the following quote from Holtzman.

The year Elroy Face was 18-1 he blew ten leads. Did you know that? But they had such a good hitting team they came back in the last inning and won the game for him.

There's a couple of factual errors in there.

1 time he came in with Pittsburgh leading 14-11, pitched three innings of scoreless ball, the lead never changed hands, and he got the win. Must be one of those screwy "scorer discresion" moves where Face got it because no one else could get anyone out.

That is a weird game:
Box

I thought the scorer discretion only applied when the starter left with the lead the team never relinquished, but didn't complete five innings. I would have given Bob Smith the 'W' there.

dosn't win shares over rate players from good teams (keller)and under rate them from bad teams(Sisler)
not to mention that first base defense in the dead ball era is under rated
also isn't 31 WS a lot to give some one in a year for war credit

Yes, WIn Shares does not give enough first base defensive credit prior to the live ball. Not sure b Sisler's time if this is a big thing.
As to underrating guys on losing and winning clubs, I think the consensus is that you have to really be a good player on a truly bad team to make much difference. I am much less sure that WS overrates players on great teams. Yes, if a team 'overperforms' or wins extra games because they are 'lucky', WS assigns them extra credit. But I do not recall a study that shows guys on consistent pennant-winners are overrated by WS. Hmm...maybe I'll post this over on the Uber stats thread.

the Browns in 1917 were down right lousy
57 wins 97 losses .370 winning % 43 games out of first
while Sisler hit 353/390/453 in 587 PA in leauges that averaged 250/320/324

Yes, WIn Shares does not give enough first base defensive credit prior to the live ball. Not sure b Sisler's time if this is a big thing.
maybe by the second half of his peak it wasn't but 1915-19 I'm quite sure it did

Sisler's 29 Win Shares in 1917 matches his 29 Win Shares in 1922. I don't Sisler is a great example for this.

31 may be a lot to give a player from a war year if that player doesn't have years aurrounding the war that are even better. The year that I give 31 WS to Keller for he already has 11 in 44 games, which means he is on pace for 38 over 154 games. I dont' think that 31 is too much to give him in that season. I give Keller 28 for 1944.

As far as 1B defense, while WS does underrate it during the dead bal era, it isn't like WS likes Sisler as a defensive 1B anyway. Sisler's rep exceeds most every metric I have seen. In fact I can't remember, off the top of my head, any metric that places him as a great defensive 1B. In other words, it isn't like WS is missing any more than 1 or two WS ina couple of different seasons here.

The more I look at this it becomes clear to me that Keller has the better seven year peak/prime his peak/prime can be extended to 9 seasons (10 if you count 1937, which I don't). The advantage that Sisler has is that he has a better rep, he is a HOFer with a career .340 BA and that he has a number of filler seasons at the end of his career which push him ahead in career metrics whereas Keller didn't play much after that 9 season peak.

Also, were the tactics of 1915-1919 the same as those of 1893-1900? I know that during the 1890's we have a high scoring era with a lot of 'baltimore chops' and bunts where the 1B is heavily involved. Does this change by the WWI era?

Not only that but Sisler's Browns were not that bad, not consistently (Ok 1917). But if I recall we're talking 2-3 WS in an extreme case. This only really makes a difference if a guy is disadvantaged (or advantaged) for several years. I don't think Sisler was.

The issue with Sisler, to me, is his defense. And I don't even care about the deadball thing. The point is James said that Sisler sucks and contemporaries said he was great. There is your discrepancy.

Try Chuck Klein, though, if you're lookin' for a guy who played great for horsebleep teams.

No, scorer's discretion can come anytime.
14-11 games are exactly when you used to see it employed more. One guy gives up a couple of runs but technically would be the winner, but another guy pitches a few scoreless, so he gets the nod.
There was a weird one like this last week, except in a lower-scoring game (Baltimore-Toronto?). One guy was supposed to get the save, but got the win instead because Brower only picked a guy off as his lone out, so the scorer wouldn't give him the win.
Or something similar to that.