Brent #79:
I'm trying to make sense of statistics like those for 1938 - 51 runs scored in 72 times on base. Even with him reaching scoring position on his own power 28 times, that is still a pretty amazing rate of runs scored. You just don't see records like that in the majors. Do you happen to know who would have been batting behind him?

This approach overlooks "reached first on error" rates, so it works only insofar as error rates are similar. Fielders choice rates, too. Is that likely to be important?

Naive theory says that Bell reached first on errors more frequently than his colleagues did, because of his speed. Some latterday study found little or no evidence that speedy players reach first on errors more frequently. Hard to believe.

Dr. Chaleeko #87:
it's possible that in barnstorming the NgL teams could have coasted in games against less difficult opponents or that they may have rotated more players through the lineup in those games to rest star players. I would defer to our experts on that one.

No expert here. I suppose that the barnstorming rosters were small, with great flexibility used to generate unusual usage patterns only for pitchers. (Eg, Satchel Paige three innings in the morning and three at night?) I imagine a 1901 AL team: 14 men including 5 pitchers --5 men who have jobs because they can pitch at that level, although they fill in at other positions.

Gadfly #95
I have no published League-wide data for the Negro Leagues after 1929 and before 1944. Basically from 1929 to 1944 is a statistical wasteland,

Most people (novices to the subject) would guess the opposite about the 20s and 30s. What about
the remorseless progress of human history, at least USAmerican history?

This existence of much more and better data for the 1920s than the 1930s contraposes the fact that we have many more anecdotes about the 1930s than the 1920s, much more anecdotal evidence about the magnitude of 1930s stars. --as they had in the 1970s when the Hall of Fame first considered Negro Leagues players.

Looks like Bell doesn't have much of a peak and as a peak/prime guy I may have him lower than others. His peak isn't even as big as Carey's or Roush's.

However, that career value is almost too much to ignore. He certainly trumps a guy like Beckley in that category. Right now he is not in my Power 22 (the 22 gusy I wish I could put on my ballot) but he could get into that group in time. I tend to be conservative early with NeL players.

How can he possibly trump Beckley, he has an OPS+ of 100 compared to Beckley's 125. Be real!

Does anyone know what Bell's predominant position was for each of his seasons in the Mexican League? Thanks!

How can he possibly trump Beckley, he has an OPS+ of 100 compared to Beckley's 125. Be real!

The same way Max Carey (who I wasn't a huge supporter of, but who I do like better than Beckley) is elected to the HOM and Beckley is decidedly not?

-----------------

Anyway, I got an odd sense of deja vu when looking at his profile, and I just realized what it was. Reminds me of when I saw the WS projections for Lloyd for the first time. Amazingly long career, but no real peak (Lloyd's plateau was certainly much higher than Bell's, of course). Lundy is another whose projections fit in this category. I do like him better than Carey, but Carey would not be on my ballot were he eligible today. Best I can probably do is slightly ahead of Fielder Jones, but behind Dunlap and Monroe, which is in the 20 range.

Anyway, I got an odd sense of deja vu when looking at his profile, and I just realized what it was. Reminds me of when I saw the WS projections for Lloyd for the first time. Amazingly long career, but no real peak (Lloyd's plateau was certainly much higher than Bell's, of course). Lundy is another whose projections fit in this category.

This could be an effect of the conversion system, or it could be an effect of the NeL environment, or both.

Karl,

Bell was an A to A+ CF in the lively-ball era who was the best base stealer of his time. If Chris WS estimates are correct, I have him with 18 seasons with at least 15 WS and over 400 career. I have schedule adjusted Beckley's numbers and he was 14 seasons over 15 WS and he fails to reach 400 career.

Bell had a longer career than Beckley and was at least average longer than Beckley was. I have him with more defensive value and he was a great baserunner. He even has a slightly better peak and prime for the reasons mentioned above. Bell is currently at #24, Beckley is #55.

If it makes you feel better, though, I do have Beckley above Rice, Hooper, and Cross.

Jake Beckley smells bad, and Cool Papa does as well.

Does that help? :-)

Beckley dominated his position more than Bell did for their careers, so that's why Eagle Eye is on my ballot (but not #1!) while Cool Papa is not.

Brent wrote:
I'm trying to make sense of statistics like those for 1938 - 51 runs scored in 72 times on base. Even with him reaching scoring position on his own power 28 times, that is still a pretty amazing rate of runs scored. You just don't see records like that in the majors. Do you happen to know who would have been batting behind him?

Unfortunately, the Mexican League encyclopedia only gives an alphabetical listing of batters and pitchers--no rosters, so it's pretty hard to reconstruct who played for what team in what year.

Bell's teams were:
1938-39 Tampico
1940 Torreon and Veracruz (not sure when he moved)
1941 Monterrey

In 1938 and 1939, one of Cool Papa's Tampico teammates was 1B Angel Castro, who was one of the best (maybe the best) hitter in the league; he led in HR and RBI both years. In '38 he hit .354, slugged .674, and drove in 40 runs in 43 games; in '39 he hit .328, slugged .522, and drove in 50 runs in 58 games.

In 1940 I don't know how long Bell played for Torreon, but Veracruz was the Mexican League version of Murderers' Row. They had, aside from Bell:

Martin Dihigo
Willie Wells
Ray Dandridge
The aforementioned Angel Castro
PLUS Josh Gibson for the last 22 games of the season (he blasted 11 home runs and 38 RBI in those 22 games).

In 1941 Bell moved to Monterrey (Gibson and Dandridge stayed with Veracruz); one of his teammates was Sam Bankhead, who hit .351, slugged .521, and drove in 85 runs in 101 games.

This is just from a rather haphazard trip through the Mexican encyclopedia; there are probably other important guys I'm missing.

what cap should Bell be wear if he makes the HoM?

what cap should Bell be wear if he makes the HoM?

I would say the St. Louis Stars, yest.

In # 40 Chris Cobb wrote:

I’ve projected Bell for 1922 and 1923, but I do not think he would have been in the majors in those seasons, so I have not included them in any of the career totals.

Consistently with this comment, Chris didn’t calculate win shares for these seasons. In # 17, however, Gary A presents Bell’s statistics for 1923, which indicate that he was a successful pitcher, the best on his team. Shouldn’t we figure that performance as the best pitcher on the St. Louis Stars projects as a regular starting pitcher in the majors? I’m inclined to credit Bell with perhaps 15 to 18 additional WS for his 1923 season.

The 1923 St. Louis Stars, if it matters, weren't a very good team--they finished in sixth place, at 31-49-1.

But Bell was the workhorse and best pitcher. The top four starters:

Cool Papa Bell 11-7, 4.53 era, 154 ip
Archie Stewart 5-6, 5.14, 119 ip
John Finner 4-8 6.26, 92 ip
James Gurley 3-8, 4.91, 88 ip

I began this message looking for a clarification. The projected Win Shares I got from Chris Cobb's work was 409.5. But looking at Tiboreau's handy-dandy summary table, I see Bell listed with 419. I'm wondering if one of those numbers is wrong, or if there was an adjustment somewhere that I missed.

But thinking about it, I'm wondering if, when the numbers get that high, it makes all that much of a difference. I mean, if either one is accurate, that's an extraordinarily high WS total, especially for a guy about whom the consensus is "not as good as his reputation". I'm relatively new at this, so I ask: do we have a precedent for rejecting a candidate with such a gaudy number? And if we're discarding Chris' projections for Bell, should we be doing the same for Wilson, Suttles, and Beckwith?

SWW,

Chris also said, however, that his estimates were inflated because he thinks he alloted too much playing time to Bell. In fact, he suggested that he overestimated playing time by 10%. So those 400+ WS are probably more like 360, putting Bell squarley into CF-glut territory since his peak is very low for a HOM candidate.

Just because I happen to be assembling Mexican League batting data right now...

Cool Papa Bell's MxL OPS+s

year  obp+ slg+ ops+
--------------------
1938  139  159  198
1939  132  142  174
1940  137  163  200
1941  111  110  121
====================
total 126  140  166

And translated to an NL environment at the usual .90/.82 conversion rate:

year  obp+ slg+ ops+
--------------------
1938  125  123  148
1939  112  104  117
1940  133  144  177
1941  108   95  104
====================
TOTAL 119  117  136

I would personally argue for the translation rates to be much lower on 1938 and 1939, then higher on 1940 and 1941, but I think this gives a pretty solid picture of what kind of years he was having South of the Border.

BTW: No home park adjustments here.

So will the real Cool Papa please stand up. Does anybody, especially the people who do the MLEs, want to hazard a guess as to whether he is really a 136 OPS+ guy or a 100 OPS+ guy. Seems important to know.

Demonstrates I think that our Mexican league conversion factors are far too high. In reality, the smattering of ML and NEL players down there in its best years did not make it anywhere near the equaivalent of the full NEL. Conversion factor for the MXL should probably be something like .80/.65 on average, maybe a little higher for the peak years.

Also, don't forget a lot of the MxL parks are at high altitude, so there's a Coors effect.

I'm just not sure, Sunnyday. I didn't do full career-length projections for him, just the Mexican stuff. By then he was close to 40 IIRC. If you wanted to pin me down, I'd say look at his 1940-1941 projections rather than this 1938-1939, only because the level of play was better in Mexico those years. Of course, that doesn't help much since there's quite a discrepancy between 177 and 107....

Here's another thing to look at.

      cpb  team  mxlg    cpb  team  mxlg  
YEAR  AVG  AVG   AVG     slg   slg   slg
----------------------------------------
1938  356  262   267     588   384   371
1939  354  278   261     509   380   358
1940  447  304   290     686   426   420
1941  314  258   288     463   395   396
========================================

1938 and 1939 look like his park helped a little bit.

But clearly in 1940, CP's huge 177 OPS+ year, he's either on a great offensive team and/or hitting in a very good hitter's park.

His 1941 park was at Nuevo Laredo which I firmly believe was a pitcher-friendly park.

Overall, I think he's getting some help from context, but I cannot pinpoint the amount.

Karl,

I disagree with you about the conversion rates generally being too high. I think you're right for 1937-1938 when very few non-native players at all were down there. In 1940-1941, this league was pretty damned close to major-league quality, and ditto for 1946-1947. In between it was better than in the 1930s but not as good as 1946-1947. In the 1950s, the Mexican League was classified by organized baseball as a AA-league. I personally believe that it probably slid down to AA level around the same time that the NgLs and the PRWL did (after 1948ish). Until then, however, given that it attracted talent from around the Carribean and the U.S., I think a .90/.82 rate is reasonable for 1939-1948.

Also, the high altitude doesn't seem to have the effect you'd imagine. For instance, Mexico City is not an ungodly Coors type field despite being at 7000 feet. The offensive conditions of the game down there were a tad strange. Averages were generally a smidge higher than in the US, but ISO was lower. Walks, however, were a big part of the Mexican game, much moreso than in the US game at the same time. League ERAs were generally in the mid-4.00s IIRC, so if the altitude were really a big deal, you'd figure that would show up more in the ERAs.

I don't understand how Mexico City doesn't have around 50% more inflationary effect on hitting than Coors, being 50% higher. The Coors effect is not an effect of that particular park, it was predicted by analysts, including I think Bill James, in the 1980s before the park was opened. I suppose today the smog could slow down homers, besides causing base stealers to expire from asphyxiation, but Mexico City wasn't that big a place in 1935-50.

For whatever reason, the conversions appear an order of magnitude less reliable than those for the NgLs. We've now had a number of NgL players whose Mexican conversions are much higher than those from the NgL, which strongly suggests that the factors are wrong.

I just don't believe that the Mexican league was equivalent to the majors in 1940-41, or in 1946-47; the major stars notably did NOT jump to Mexico in 1946, for example, even though one or two MxL players like Sal Maglie later turned out to be really good. The leagues had basically one rich sugar daddy, but otherwise even though Mexico was at a temporary economic peak in the 1940s it was not remotely comparable to the US, so there just wasn't the money there.

This is quick and dirty since I have to leave for the day now. The league totals all include Mexico City, so these aren't exactly correct, but they give a good general idea.

Mexico City
YEAR  TM AVG  LG AVG  AVG+
--------------------------
1940  .308    .290    106
1941  .319    .288    111    
1942  .298    .289    103
1943  .280    .273    103
1944  .274    .284     96    
1946  .287    .281    102
1947  .280    .278    101
1948  .272    .273    100
1949  .266    .276     96
1950  .252    .260     97
1951  .272    .272    100
=========================
mean  .283    .279    101

YEAR  TM SLG  LG SLG  SLG+
--------------------------
1940    .442    .420   105
1941    .453    .396   114
1942    .402    .392   103
1943    .387    .367   105
1944    .375    .387    97
1946    .417    .381   109
1947    .367    .366   100
1948    .408    .378   108
1949    .345    .369    93
1950    .327    .335    98
1951    .400    .367   109
==========================
mean    .393    .378   104

So, it does appear that there is some park effect particuarly early in the 1940s, but that it is less persistent and less dramatic for AVG than for SLG. I wonder if Mexico City got a new park somewhere around 1944?

If you asked me to WAG it, I would speculate that the park may have been a trifle difficult for picking up pitches (bad hitter's backdrop?), but that when a fellow got hold of it, the ball carried well.

However, I also think these numbers do suggest that the effect of the park was no where near to that of Coors (which usually clocks in with a PF of 115-125. At most, it would be like the Ballpark at Arlington (which seems to be 105 to 110 PF in most years).

I think the numbers Dr. Chaleeko has on play in Mexico City show pretty clearly that the expected Coors effect just isn't there. Why not? Who knows.

I wonder what kind of baseballs they were using in the Mexican League.

I wish I had time to give Cool Papa's career another look, but right now I don't.

I take it the figures above are Mexico City vs. MxL average. What was the average elevation of MxL ballparks? I know some were at sea level, but if the average was say 4,000 feet the Coors effect would be half-hidden.

I know some were at sea level, but if the average was say 4,000 feet the Coors effect would be half-hidden.

That's quite possible, though if it is half-hidden in that way, we wouldn't have to worry about it as part of the park factor :-) .

I don't recall the team data as suggesting that the sea-level parks were consistently below leagure average in scoring levels, though.

Btw, I agree that the 1938-39 MxL would need to be converted at a much lower rate, and 1940 at probably a slightly lower rate, than the NeL. Still Bell's 1940 season was almost certainly the best of his career, and it would be nice to have a more accurate view of just how good it was.

Just throwing something out of my arse here . . . . I remember in the 00/01 off season there were questions about how well Ichiro would hit for the Mariners. The projections said he'd be a good hitter - over .300, but no fantastic batting average. Midway through the year, when Ichiro was hitting about 50 points over his projections, there were some questions about what went wrong with the math, and the general theory I remember being articulated was that the projections didn't/couldn't account for the role his speed played in his getting on base. Pitching is better here, but the third basemen don't throw over to first any faster.

Could this also be true of Cool Papa?

Chris J,

And IIRC, the opposite happened with Matsui. He hit for the right average, but not with the expected power. I think Davenport or Shandler has proposed that AVG translates almost directly from Japan, but that power is sapped a bit. Of course, this makes sense given what we know about the square root thing that Chris has used.

Anyway, what I think is really peculiar about Bell, however, is the BIG power spike he saw in Mexico. Sure slugging 686 is easier on a 447 average, but that's still a 219 ISO. He's got to be legging a lot of triples and doubles to do that. That's certainly possible given how fast he was supposed to be, and especially since we don't know much about the ballparks he played in, but let me posit one last postulate: scoring was lenient.

If scoring was lenient, some of those leg singles and extra bases on tough hops, or slight outfield bobbles are going to show up as more hits and xBH.

I'm not saying it's the right theory, nor that Bell didn't just have a superb year (at an advanced age...), but it's the kind of thing that would happen to play so perfectly to his skill set that it might make sense.

Or maybe it's all of these things at once:
-lenient scoring
-park factors
-late-career power spike
-his blazing speed and agressive style.

Chris J,

And IIRC, the opposite happened with Matsui. He hit for the right average, but not with the expected power. I think Davenport or Shandler has proposed that AVG translates almost directly from Japan, but that power is sapped a bit. Of course, this makes sense given what we know about the square root thing that Chris has used.

Anyway, what I think is really peculiar about Bell, however, is the BIG power spike he saw in Mexico. Sure slugging 686 is easier on a 447 average, but that's still a 219 ISO. He's got to be legging a lot of triples and doubles to do that. That's certainly possible given how fast he was supposed to be, and especially since we don't know much about the ballparks he played in, but let me posit one last postulate: scoring was lenient.

If scoring was lenient, some of those leg singles and extra bases on tough hops, or slight outfield bobbles are going to show up as more hits and xBH.

I'm not saying it's the right theory, nor that Bell didn't just have a superb year (at an advanced age...), but it's the kind of thing that would happen to play so perfectly to his skill set that it might make sense.

Or maybe it's all of these things at once:
-lenient scoring
-park factors
-late-career power spike
-his blazing speed and agressive style.

In reviewing Ashburn's credentials, I decided to go through this Bell thread again to see Ashburn compares with Bell. In the comparison, Bell has several advantages:
- speed,
- probably a better fielder,
- long career--played well until he was 43; Ashburn's last season was at 35;
- isolated power.
Ashburn's ahead in some other areas:
- probably a slightly higher relative batting average
- drew more walks
- didn't have Bell's 4-year dip from ages 24-27 after taking up switch hitting.

In re-reading the thread, however, the entry that was most striking was Dr. Chaleeko's translation of Bell's Mexican League seasons in #117, showing MLE OPS+ ranging from 104 to 177 and averaging 136. These figures are much higher than the 105-120 than Chris Cobb (in # 92) said he assumed for those years.

When Dr. C posted his translations I remember treating them with skepticism, thinking that a Coors-like park effect must be in play. But I see that subsequent posts have dealt with that issue.

If Dr. C is right about Bell's Mexican League seasons, it's hard for me to see putting Ashburn ahead of Bell. On the other hand, our other expert on Negro League translations, Chris Cobb, seems to be quite skeptical about Bell's credentials; I notice he ranked him # 29 on his last ballot. What does Chris think of Dr. Chaleeko's Mexican League translations for Bell? Have we learned anything in the last 5 months that would cause us to re-think them?

Based on re-reading this thread, I am considering moving Bell up substantially. He appears to have had a whole lot of value that is being masked by a few bad seasons. I think Ashburn is HoM-worthy, but I sort of doubt that he was really any better than Bell (at least leaving out Bell's 1927-30 seasons). Comments?

I've been having some thoughts on moving Bell up as well. I'd like to hear if anyone (especially someone that's a friend of Beckley, a career guy) to see if it's reasonable to have Bell at 21 when I have Beckley at 2.

How long would Bell have played in the 16 team MLB environment (post-integration, no less) that Ashburn played in?

And apropos of our other discussion this week, as we think about Ashburn, yes, we don't yet know about Kirby Puckett and Ken Griffey and... but how long would Ashburn have played in a 30 team MLB environment.

Shouldn't we care about equalizing the opportunity?

It seems that Ashburn had the better peak so how good was Bell from ages 36-43? This does matter because if he wasn't much better than average he doesn't get much credit for those years in my system.

Was Bell really a better fielder than Ashburn? I know that Whitey wasn't as good as his Range factors say, but at the same time Bell would have to be mighty good to be better than he.

Speed isn't something that really matters here is it? I guess by speed you mean baserunning, but do we really know that? Wasn't Roger Cedeno an awful baserunner? Arent' there plaenty of really fast guys who aren't good baserunners? Speed is great when prjecting a player, but not too useful when looking over what he did. It should show up in his stats (3Bs, 2B, inthe parks HRs, SBs, etc.) so extra credit for speed is just double counting a skill that is in actuality very overrated to begin with.

Bell does seem to have had more power, but Ashburn has a great career OBP, which is probably more important.

So I see Whitey as having the better peak while being an equal fielder at least and an OBP machine. Bell did have more speed, more career (though probably wouldnt' ahve played into his 40's in MLB), and a higher ISO (tconnected to his speed). I choose the first.

Good to see some people rethinking Cool Papa.
This type of player loses steam in the early going with tough competition, then doesn't bounce back in the voting as well as I think they should later on.

He's always been on my ballot, at least, and is in the middle now. You could argue having him be higher up.

Without looking back at post 98, I don't recollect off the top of my head whether I used .90 or .87 or .85 for Bell's Mexican seasons. Regardless, even if you use the lowest of the three those seasons end up looking great.

BELL vs. ASHBURN, part one (filling in the gaps)

In comparing two players with different career lengths, like Bell and Ashburn, one strategy is to compare the shorter career with a similar-length subset of the longer career. If the two are comparable, there should be a presumption that the longer career is the more valuable one.

Before we do that with Bell, however, I first need to fill in some missing pieces of his MLEs. Chris calculated his MLEs (post # 40) without detailed information on Bell’s seasons in Santo Domingo (1937) or Mexico (1938-41). Consequently, Chris didn’t compile detailed MLEs for the 1937-41 seasons. Rather, he compiled MLEs for the rest of his career, then at the end he added on another set of career totals including the missing 1937-41. We can back out Chris’s estimates for 37-41 by subtraction:

__Yrs __G __PA _BB Hits _TB __BA _OBP _SA
37-41 725 3045 349 864 1008 .320 .398 .374

He writes, “I’ve projected Bell in 37-41 as overall, the same level of play as 1936 & 1942,” and indeed the numbers look like they are just equal to 5 times the average of the MLEs for ’36 and ’42. Later, when Chris posted win share estimates (# 97), he’d seen Gary A’s posting of the Mexican League data (# 78), so he has them varying from year to year, but it looks like they’re still generally consistent with the totals posted in # 40. Since I wanted to work with annual data, I’ve attempted to allocate Chris’s estimated batting totals for 1937-41 across the individual seasons following a pattern that would be consistent with the year-to-year pattern of win shares:

Year __Team _G __PA _BB Hits _TB __BA OBP _SA BA+ OBP+ SA+ OPS+
1937 S Dom 145 0609 065 174 0197 .319 .393 .362 114 115 091 106
1938 Mexico 145 0609 071 171 0219 .317 .397 .406 115 117 104 121
1939 Mexico 145 0609 069 174 0199 .323 .399 .369 115 115 092 107
1940 Mexico 145 0609 064 199 0228 .366 .433 .419 134 129 107 136
1941 Mexico 145 0609 080 146 0165 .276 .371 .313 104 110 083 094
Total______ 725 3045 349 864 1008 .320 .398 .374 117 117 096 113

To be clear, these are my own estimates, though I’ve tried to make them consistent with the information Chris has posted in # 40 and # 97. I’ll note that for calculating OPS+ and other relative rates, I’ve used the following league averages (excluding pitchers) from the NL:

Year *lgBA *lgOBP *lgSA
1937 .280 .342 .397
1938 .275 .339 .391
1939 .280 .346 .401
1940 .272 .336 .391
1941 .266 .336 .375

One result of these MLEs is that it finally provides us with an estimate of Bell’s career OPS+, which is 103. (An estimate of 100 has frequently been cited—see David Foss # 90—but it excludes 1937-41.) Here are Bell’s career MLE batting average, OBP, and SLG, followed (in parentheses) by the league averages over his career, followed by his relative rates:

.302/.373/.380 (.283/.346/.400) 107/108/95

A couple of notes on these estimates:

1) The actual Mexican League data show a surge in Bell’s power relative to his record in the Negro Leagues. I’ll repost the actual data from # 78 by Gary A:

YR--G---AB--H---D--T-HR--R--RBI-W--K-HP-SH-SB-AVE-SLG
38-040-160-056-07-06-06-051-17-20-12--1--1-09-356-588
39-058-226-080-16-02-05-055-24-26-19--1--6-12-354-509
40-089-382-167-29-15-12-119-79-45-24--0--3-28-437-686
41-100-421-132-21-15-04-085-48-65-26--1--4-14-314-463

(Note – I’ve made one correction to Gary’s post, changing Bell’s hits for 1938 from 51 to 56, to make the batting line for that year consistent with the listed average and slugging percentage.)

As Gadfly noted in # 24, Bell averaged 14 home runs, 20 triples, and 39 doubles per 154 games in his actual Mexican statistics. The MLEs based on Chris’s estimates do not exhibit this surge in power, and indeed show Bell as lacking in power during these seasons.

2) Chris’s win shares (and the MLEs that I’ve created to be consistent with them) do not fully reflect the huge year that Bell had in 1940, when Bell led the Mexican League in average and in home runs.

A quite different look at those seasons comes from Dr. Chaleeko in post # 117. Again, I needed to fill in some blanks (assuming the same playing time as in my previous estimates based on Chris’s numbers), but the following MLEs are consistent with the OPS+ and SA+ rates posted by Dr. C:

Year __Team _G __PA _BB Hits _TB __BA OBP _SA BA+ OBP+ SA+ OPS+
1938 Mexico 145 0609 078 180 255 .339 .424 .481 123 125 123 148
1939 Mexico 145 0609 066 170 226 .313 .388 .417 112 112 104 116
1940 Mexico 145 0609 067 205 305 .378 .447 .563 139 133 144 177
1941 Mexico 145 0609 078 143 189 .270 .363 .356 101 108 095 103
38-41_____ 580 2436 289 698 975 .325 .405 .454 119 119 117 136

Comparing the two sets of estimates (here are the totals based on Chris’s data for just the 4 Mexican seasons),

_G __PA _BB Hits _TB __BA OBP _SA BA+ OBP+ SA+ OPS+
580 2436 284 690 811 .321 .400 .377 117 118 097 115


we see that the estimates of average and OBP are similar, but Dr. C has come up with much higher estimates of slugging (117 compared to 97).

With Dr. C’s estimates, Bell’s lifetime OPS+ goes up to 106, with component rates of:

.303/.374/.394 (.283/.346/.400) 107/108/98

BELL vs. ASHBURN, part two (the comparison)

All of this is a long way to come to where we can compare Ashburn with a subset of Bell’s career. Ashburn played 15 seasons, from ages 21 to 35. Ordinarily we would try to match ages, but Bell went through an unusual slump from ages 24 to 28 after taking up switch hitting. On the other hand, he aged unusually well, with his big 1940 season coming at the age of 37. So I’ve taken Bell’s 15-year subset as 1924-26 and ‘32-43 (ages 21-23 and 29-40). I note that both Ashburn and Bell had unusual within-season durability and played (or are estimated to have played) at least 100 games each of these years, and 150 games many years. During these periods, each player had three “off” seasons with OPS+ below 100. I think these seasons make for a fair comparison.

Here is a comparison of Ashburn’s lifetime statistics from bbref compared with 15-year MLEs from Chris and from Dr. C:

Batting Statistics, Ashburn career (1948-62), Bell MLE 15 years (1924-26, ‘32-43)

___Player _G __PA _BB Hits _TB __BA OBP _SA BA+ OBP+ SA+ OPS+
_Ashburn 2189 9563 1198 2574 3196 .308 .396 .382 115 118 093 111
Bell-Chris 2179 9189 0966 2562 3197 .312 .384 .389 111 112 098 111
Bell-Dr. C 2179 9189 0971 2570 3361 .313 .385 .409 112 113 104 116

[Note – for comparability I’ve calculated Ashburn’s PA as AB+BB, but for OBP, OBP+, and OPS+ I used the data shown on bbref, which include HBP and SF (when available). Ashburn’s OBP calculated just from H, BB, and AB is .394.]


Here are some observations:

1. Even though Bell projects to almost the same number of games as Ashburn, his projection shows almost 4 percent fewer plate appearances. Ashburn spent most of his career as a leadoff hitter and was used as a principle leadoff hitter in 11 seasons and for a total of 1,410 games (see <a >article </a> by Herm Krabbenhoft in November 2003 issue of SABR's By the Numbers.) For Bell, of course, we do not have nearly as much information, but Gary A (in # 78) reports, “I know for sure that Bell batted leadoff as late as 1934. I'm pretty certain he did for years after that, and his Mexican League record seems to indicate that he was still leading off then.” Bell’s projected 1924-43 NL environment had a slightly higher OBP than Ashburn’s (.342 to .337), so if the two players were used in the same manner, we would expect Bell to have had a few more plate appearances. So one question is whether the difference in PA reflects a real difference, or is it just an artifact of the assumptions used for the MLEs?

2. Comparing Ashburn with Bell based on Chris’s MLEs, they have the same OPS, but Ashburn’s is more OBP intensive. Since we know that OBP should receive a larger relative weight, I think we can conclude that with more appropriate weights Ashburn would be slightly ahead in the comparison.

3. If we look at win shares, Ashburn has a total of 329, which compares to Chris’s projection for Bell of 306.7 over these 15 seasons. The difference is not due to fielding—Ashburn receives 73.8 fielding win shares (or 3.95 per 1000 innings), compared to 75.6 fWS for Bell. (Chris assumed a rate of 4.03/1000 over Bell’s career, so the similarity was expected.) The difference is that Ashburn has 10 percent more bWS than Bell. We saw that about 4 percentage points can be explained by the difference in PAs, and another 2-3 percentage points may be due to the difference in OBP-intensity, but it seems that the remaining difference of 3-4 percentage points is unexplained. Could Chris’s methods for deriving bWS be giving results that are a bit too conservative in this case?

4. Under Dr. C’s version, Bell’s advantage over Ashburn in OPS+ (116 versus 111) surely exceeds any advantage Ashburn may have had in PA or OBP intensity.

5. Which version should we accept? Dr. C (#117) said he used conversion rates of .90/.82. As described above, Chris didn’t do a direct conversion of these seasons, but based the estimates on the bounding seasons (1936 and ’42). In post # 97 Chris suggests that appropriate conversion factors for Mexico would be between 85/.72 and .87/.75. However, as part of my generation of season-by-season statistics that are consistent with his WS estimates, I had to come up with implied conversion factors for each season. For batting average, they were in the range Chris indicates, but for slugging, only one season (1941) had an implied conversion factor as high as .72-.75; the implied factors for the other three seasons (including his big 1940 season) were about .65. These factors are clearly too low and explain why the surge in slugging that is apparent in Bell’s Mexican League data does not show up in the MLEs. While Dr. C’s version (1940 OPS+ = 177) may be an overstatement, the actual statistics are huge and I think it’s hard to see this season converting to a value lower 155. A conservative position may be to take the midpoint between the two sets of MLEs.

Turning to some considerations outside the scope of this comparison:

6. The 8 remaining seasons in Bell’s career include several with significant value, including seasons with OPS+ estimates of 118, 113, 97 and 92. The other 4 seasons (including 3 immediately after his adoption of switch hitting, 1927-29) have OPS+ estimates in the 70s, which I think would be close to replacement level for an excellent defensive center fielder. I think many voters’ evaluations of Bell have been negatively influenced by career averages that include these seasons, leading them to overlook the fact that he was a very good offensive player over most of his long career. We’ve seen that during Bell’s prime he was quite comparable to Ashburn, and he does have additional value outside his 15-year prime.

7. Bell is projected to an environment of the NL from 1926-46, which due to segregation was not as competitive as the integrated 1950s environment in which Ashburn played. As others have described it, during the late 1940s baseball essentially contracted from 3 major “leagues” to 2 (keeping in mind that there were actually 2 Negro leagues, though neither was of fully major league quality). This is an important factor in Ashburn’s favor.

8. On the other hand, it’s been noted that a similar contraction took place within the Negro Leagues from 1933-36 when there was only one league operating. I don’t believe Chris’s conversion rates have ever been adjusted to take account of this, so it’s likely that Bell’s MLEs for those seasons understate his true performance.

9. It’s sometimes argued that Bell’s career doesn’t have a comparable peak to Ashburn’s (for example, see jschmeagol’s # 133). However, I think this perception is merely an artifact of the methods used in constructing MLEs. First, with the exception of Bell’s 1940 season, both players had fairly flat career profiles (which is also true of several other players in this class, such as Carey, Dom DiMaggio, and Leach). Second, as mentioned above, IMO Chris’s WS estimate for Bell’s monster 1940 season in Mexico is much too low. More generally, several assumptions in the conversion process (including, but not limited to, regression) tend to smooth out the MLEs, so I strongly caution against paying much attention to the year-by-year values of MLEs or other Negro League position players. The MLEs are designed to reflect trends over periods of several years, not individual seasons.

Bottom line – comparing a 15-year interval for Bell against Ashburn’s full career raises some questions about various details of the MLEs, but the general impression is that Bell was either slightly below or (more likely) slightly above Ashburn. Ashburn has an advantage of having played in a more competitive era due to integration, but I find it hard to see that factor offsetting Bell’s additional 8 years (including 4 that were well above replacement level). Ashburn will be high on my ballot, but Bell will move ahead of him.

Reposting link to article by Herm Krabbenhoft in November 2003 issue of SABR's By the Numbers.

Great analysis from Brent!

Using the better information about the Mexican League compiled by Dr. Chaleeko after Bell became eligible, I have started working on new MLEs for him. I'm very busy with work right now, so it'll probably be a couple of days before I have everything ready. Preliminary results suggest that Brent's view of Bell as very close to "Ashburn + 8 other seasons" is probably reasonable.

One advance note: It should be kept in mind in judging Bell's outstanding 1940 season that the MxL was a very high offense context that year: converting to the NL norm knocks 8% off Bell's BA and SLG each, before the competition factor. With that offense adjustment, a conversion factor of .87/.76 for the season, and regression, it looks like Bell's 1940 season will come out with an OPS+ of around 140. The regression formula probably lowers it a bit more than it truly should, but I would counter Brent's assessment that the season should be no lower than 155 OPS+ with one that claims that the season should be no higher than 155.

It's right for us to be reassessing Cool Papa Bell in comparison to Ashburn. I would add, however, that there are two other centerfielders who bear careful comparison to him also: Alejandro Oms and George Van Haltren. Oms especially is still being ranked, I think, lower than his statistics warrant. I hope that the renewed consideration of Bell will extend to these other center fielders also.

While the MLE's may knock a little off of aplayer's peak, it usually does add to a player's career. I dont' think that Bell plays ages 41-43 in MLB or if he does he is a part time player who isn't really adding anything to his HOM resume. Add thsoe years and his time as a switch hitter and you end up with a player just under ashburn Ashburn and a lot of seasons that aren't adding anything to his HOM resume in my system.

I'm happy to reassess Cool Papa, but the fact is that we are doing so because of his reputation and his reputation alone. If we were basing a reassessment on the numbers, we would be reassessing another 10-12 guys who have better numbers. And reassess or not, I'm not that interested in electing NeLers or anybody else based on their reputation. I am satisfied that we have enough numbers and enough context to elect them based on the numbers.

Well, with the exception of pre-1920. How about a reassess of Jose Mendez and Cannonball Adderly, er, I mean Redding.

Well, we're also reassessing him because his Mexican League numbers play a significant role in his case, and we are in a position to interpret those numbers with much more accuracy than we were when Bell's MLEs were first run. In the reassess, I'm also going to be redoing my handling of Bell's walk rates, which I based on purely on the model of Max Carey, rather than on Bell's actual data. The MxL stats provide an important body of data on this that also need to be taken into consideration.

We have a similar expansion of data (and an improvement of methodology) in the case of Jose Mendez, so he really should receive a reevaluation as well. Unfortunately, MLEs for pitchers are much more labor-intensive and less certain than MLEs for hitters, so, speaking for myself as reevaluator, I just haven't been able to do it.

Dick Redding and Dobie Moore are also deserving of fresh looks. Although we don't have new data, improved methodologies could be applied in both cases. Again, speaking for myself, the limiting factor is time.

Alejandro Oms simply deserves having everyone look at his numbers again.

In 1928, in games in St. Louis, the two teams playing scored an average of round 14 runs per game (7 per team).

St. Louis played 59% of their games that year at home.

It's only one year, but the other limited info I have indicates this is not an abberation, but that the St. Louis Stars Park was heavily hitter friendly.

Based on that, Bell's St. Louis seasons do not appear to be "HOM" quality to me, once I adjust for the park context.

The question arises as to whether Bell's St. Louis seasons need to be "HoM' quality for Bell to be a strong candidate, given the rest of his career.

Yes, that's the question.

He played 10 years with the Stars, at the start of his career and thru the ages that would normally be considered a players "prime" years, so for me, so far, the answer is yes, they need to be a little better offensively to get on my ballot...

Chris Cobb wrote:

Using the better information about the Mexican League compiled by Dr. Chaleeko after Bell became eligible, I have started working on new MLEs for him.

Great! I look forward to seeing them.

One advance note: It should be kept in mind in judging Bell's outstanding 1940 season that the MxL was a very high offense context that year: converting to the NL norm knocks 8% off Bell's BA and SLG each, before the competition factor. With that offense adjustment, a conversion factor of .87/.76 for the season, and regression, it looks like Bell's 1940 season will come out with an OPS+ of around 140.

But I thought Dr. Chaleeko's (Eric's) estimate of OPS+ of 177 took account of offensive context. That's a huge difference! My recollection is that Eric hasn't used regression in his MLEs. If it's not to much trouble, would you mind posting the results with and without regression. Also any other information that may help us understand the difference between 140 and 177. I think so far most of us have been assuming that MLEs from you and Eric are essentially comparable, but this difference calls that assumption into question.

jschmeagol wrote:

While the MLE's may knock a little off of aplayer's peak, it usually does add to a player's career. I dont' think that Bell plays ages 41-43 in MLB or if he does he is a part time player who isn't really adding anything to his HOM resume.

Why would you think that? Players generally keep their jobs as long as they keep playing well. The evidence on the thread shows that Bell continued to play well at least through age 41, contributing to a pennant-winning Homestead Grays team. Furthermore, his age 39-42 seasons come during WWII when draft-exempt players were especially in demand. The only season late in his career that seems like a question mark to me is age 43 in 1946, when he was coming off a poor season and the soldiers were coming home.

There are a couple of issues that need to be addressed to produce better MLEs for Bell: 1) the contraction of the Negro Leagues circa 1933-36, resulting in the tougher competition for those years we've seen with several hitters (Wells, Suttles, etc.); and 2) the quality of the Mexican League, especially for the 1940-41 seasons. Eric argues above (post 21) that it was near-ML quality--I seem to recall that he lays out the argument somewhere, listing all the NeL players down there in those seasons (in a 6-team league). I don't remember where this is, though.

Chris C.,

Just to sound the usual note of caution, as you reasses Bell's walk rate via his Mexican walk totals, please be sure to look at his walks versus the league walk rate. Mexico walked much more often than the Big Leagues. I think the stat pack I sent you should have walk rates built into it.

Brent and Gary A.,

I think the Mexico quality question is literally a year-by-year proposition. 1940-1941 and 1946-1947 being much better years than the surrounding ones. I've always used a single discount rate, but only because there seems to be little agreement about how much each year should differ.

Dr. C. : I have taken the league walk rate into account, using the data you provided.

Brent, I'll present the unregressed along with the regressed data for 1940 when I have it. As I started delving further into the walks issue, I discovered that I made a change in the formula that I used for dealing with walks since I did my MLEs for Bell, and the values that I used as inputs for the new walk estimates were calibrated according to the new formula, not the old formula that was built into Bell's spreadsheet and so were too low for the old formula. I expect that when I switch Bell to the new formula, his OPS+ for 1940 will be higher than I estimated above, probably close to 150. The difference from Dr. C's MLE will probably be then found to derive from a) regression, b) my use of a .87/.76 competition adjustment for the MxL in 1940 rather than .90/.82, and slight differences in the ways in which Dr. C and I handle walk estimates, as those are the only differences in our methods that I know of.

This has become an exceedingly difficult week at work, unfortunately, so I doubt I'll be able to get much more done on the new MLEs until Friday.

Thanks. There's no rush here; please take care of your real work first.

http://www.umsl.edu/~whmc/guides/nleague.htm

BELL: I broke the record from second base. The record in the major leagues is 13.3 and I did it in 12. A guy named Swanson had the record, with Cincinnati, I believe, and Archie Deacon, a St. Louis boy broke his record.

SHAFFER: Is that swinging the bat and then running.

BELL: No, that's just starting from the plate.

On defense, Bell had the same liability as Ashburn: a weak arm. It's hard for me to think of Bell as being in the top tier under those circumstances. I would describe him as a good outfielder, but not a great one.

Ashburn was probably as close as any centerfielder is going to get to A+ defense with that liability. Before Mays, and long before anybody had used the words "range" and "factor" in the same sentence, Ashburn was clearly the model for the position. So, in terms of assessment of defense, I would definitely put Ashburn above Bell.

Any chance I could get an updated MLE for Bell by the voting deadline?

If necessary, I would not have an issue with extending the deadline a day or two if that would make a difference, although finishing it up tonight would be better. But as always, more information is better.

I tried to send Chris an email, but his linked email is no longer valid.

I should have asked earlier; I didn't realize that the existing MLE's for Bell were so sketchy.

My ballot is complete save for Bell and Oms, but I still wouldn't mind extending if people feel a new MLE from Chris would help, assuming he could get one in sometime soon.

Also, a question about Bell's defense. Clearly, the guy was an all-time, historically great baserunner--I have no problem crediting him with the same 10-11 career baserunning wins I give to Raines, Henderson, Carey, Willie Wilson etc. But was his defense merely excellent or also otherworldly? Is he considered to be hands-down the best fielder in Negro League history, or merely a perennial Gold Glover? Whether he's close to Speaker/Mays territory (+200 and up) or merely an excellent +100 will have a big impact on where he is placed on my ballot. No matter what, Carey is clearly the relevant comp here.

As it happens, I have been working on new MLEs for Bell, and I was almost ready to post them when I got sick and busy a week and a half ago. I've been able to accomplish nothing since then. Now I am much improved, and I could get them posted tonight, I think, for what they're worth, but not in time for a voting deadline tonight.

However, the new MLEs for Bell will show us the limits of the MLE system when dealing with the 1930s and the Mexican League. We can use them, but they are obviously not rendering Bell's career accurately.

As to Bell's defense, I would say that we don't have enough evidence to put him into Speaker/Mays territory. I think projecting him into Max Carey territory can be done with more confidence.

Joe, definitely extend the deadline so Chris can get his Bell MLE's done. They'd be tremendously valuable, and Bell, due to the gap between his reputation and his stats, is, I think, a *very* important player to get as "right" as possible.

I have no issue extending the deadline a day.

Just talked with John also. Will update. Deadline tomorrow night - this will work, right Chris?

Cool Papa Bell MLEs


1929* Also played for Chi Am Giants
1932* Also played for Homestead and Kansas City
1937* Also played for Trujillo All-Stars
1938-39 played for Tampico in MxL
1940 Played for Torreon and Veracruz MxL
1941 Played for Monterrey in MxL
Conversion factors for MxL
1938 .85/.72
1939 .87/.76
1940 .87/.76
1941 .90/.81

Notes
1. The shape of Bell’s career by these MLES, with a huge offensive peak in his late 30s while in Mexico, is obviously implausible. It suggests strongly that the MxL conversion factors are too high, and that the mid-1930s factors are too low. Just as a test, I re-ran the MLEs set at .95/.90 as the conversion factor for 1931-36, and that produced the following OPS+ totals for Bell:

1931 84
1932 131
1933 101
1934 107
1935 106
1936 111

This seems more plausible. I haven’t re-run the MxL numbers with lower conversion factors, but it should probably be done.

2. I chose to start Bell’s MLE career in 1925, because that was the earliest year for which I could see a case. His MLE OPS+ totals for 1922-24 were 79, 57, and 74 respectively, not the sort of performance that gets a call-up for a major-league starting position. I’m not sure that an 85 OPS+ cuts it, either, for a centerfielder in 1924, but it’s not outside the realm of possibility.

3. It is worth remembering that Bell’s offensive growth was stunted in 1927 when he was converted to a switch-hitter by Candy Jim Taylor.

4. 1931 is a very small sample size season: 10 games, in which Bell hit, according to the HoF statistics, for very low average with zero power. The MLEs smooth this out somewhat, but probably not enough.

Note that fielding win share estimates can be found in post 97 above on the Bell thread. I made a few small changes in my playing time estimates for Bell in the new MLEs, but I think the old FWS projections are as accurate as anything I would come up with now, so in the interest of time, I would suggest that Dan R use them for purposes of fielding WAR, if he hasn't already come up with independent numbers.

Would it be possible for you to do new MxL numbers in time for the deadline tomorrow?

I can just treat 1931 as a mostly missing year and fill in 90% of it like I do for war credit if you think that would be appropriate.

440 SB seems damn low for a guy who's supposed to be up there with Rickey--that's not going to get him anywhere close to 10 BRWAA. Are Bell's NgL SB numbers not head and shoulders above those of his contemporaries?

Dan, in 1935 Augie Galan led the NL with 22 SB. Just picking a random year in the middle.

1925-43 league leading SB totals:

AL - 43, 35, 27, 30, 27, 23, 61, 38, 27, 40, 29, 37, 35, 27, 51, 35, 33, 44, 61

NL - 46, 35, 48, 37, 43, 37, 28, 20, 26, 23, 22, 23, 23, 16, 17, 22, 18, 20, 20

Are Bell's NgL SB numbers not head and shoulders above those of his contemporaries?

No.

Now there's been debate about stolen bases being underreported in many cases, so maybe the data is not reliable, but I treated Bell's SB record exactly the same way I treated Oscar Charleston's, and Turkey Stearnes', and these are the results, with one exception.

In that weird 1931 season, Bell is recorded as having 9 stolen bases in ten games, which, by the formula, gives him 96 stolen bases for that season. I cut that total in half and doled the stolen bases I cut from that season out to nearby seasons. I don't use a regression formula for stolen bases, but in this case a manual adjustment seemed appropriate.

If you want to give him baserunning credit based on repuatation, that would not be unreasonable, but the published data don't, so far as I have been able to tell, in any way imply that Bell would have walked away with Cobb/Rickey career SB numbers.

As Joe points out, that's due in part to his playing in the lowest SB context ever. Maybe he would have been stealing 60 bases a year in the majors and just broken the curve every year, but there's no evidence that he was breaking the curve that way in the NeL.

I can plug in some other numbers for the MxL, sure. It's complete guesswork, of course, but we can apply the smell test to the results.

1937* Also played for Trujillo All-Stars

Bell's batting peak in this version begins 1937, one year before Mexico. Is the playing data all Pittsburgh or does it include playing for Trujillo?

2. I chose to start Bell’s MLE career in 1925, because that was the earliest year for which I could see a case. His MLE OPS+ totals for 1922-24 were 79, 57, and 74 respectively, not the sort of performance that gets a call-up for a major-league starting position. I’m not sure that an 85 OPS+ cuts it, either, for a centerfielder in 1924, but it’s not outside the realm of possibility.

There is Roberto Clemente 1955-57, ages 20.8-23.2:
OPS+ 77, 105, 73 (aggregate 85) in 382 games, 1548 PA

3. It is worth remembering that Bell’s offensive growth was stunted in 1927 when he was converted to a switch-hitter by Candy Jim Taylor.

Even if we knock the late thirties down 10%, to OPS+=120, have we ever seen such a late development as a batter in the major leagues?

Clemente is famous partly for that but (a) he put up his first great season in 1961 at age merely old 26; (b) altho he put up his four best OPS+ seasons 1967-70 at ages 32.8-35.2, that peak was only 20+ points above his preceding six seasons 1961-66, from <140 to >160.
(huh, "only" 20+ points in his mid-thirties)

Cool Papa Bell – alternate MLEs

Age Year  Team G    PA   Hits TB    BB  SB   BA     OBP    SA    OPS+
22  1925  Stl 150   645  167  236   47  20  0.279  0.331  0.395   85
23  1926  Stl 152   684  177  252   62   8  0.285  0.350  0.405  103
24  1927  Stl 154   693  175  226   57  20  0.275  0.335  0.355   85
25  1928  Stl 154   693  175  241   62  21  0.277  0.342  0.381   89
26  1929  Stl 145   653  152  206   68  29  0.259  0.336  0.352   71
27  1930  Stl 148   666  191  275   58  13  0.314  0.374  0.452   98
28  1931  Stl 149   626  146  195   64  48  0.260  0.336  0.347   84
29  1932  Det 145   609  180  259   59  34  0.328  0.393  0.471  131
30  1933  Pgh 154   647  166  212   57  32  0.282  0.345  0.359  101
31  1934  Pgh 151   634  176  227   63  35  0.308  0.376  0.397  107
32  1935  Pgh 154   647  175  233   63  29  0.299  0.367  0.399  106
33  1936  Pgh 150   630  174  232   60   5  0.305  0.371  0.407  109
34  1937  Pgh 144   605  166  245   68   8  0.309  0.387  0.457  128
35  1938  MxL 150   630  160  246   74  24  0.288  0.372  0.442  123
36  1939  MxL 139   584  161  233   60  21  0.308  0.379  0.445  120
37  1940  MxL 140   588  177  257   47  32  0.327  0.382  0.476  135
38  1941  MxL 144   605  140  203   62  15  0.258  0.334  0.374   99
39  1942  Chi 131   550  142  183   54   8  0.286  0.355  0.369  112
40  1943  Hom 120   480  123  152   45  18  0.282  0.349  0.349  102
41  1944  Hom 118   472  119  147   47   9  0.279  0.350  0.345   96
42  1945  Hom  92   323   88  100   41   0  0.311  0.397  0.355  110
43  1946  Hom  82   283   85   97   26   9  0.328  0.389  0.379  118
career       3066 12946 3415 4658 1241 440  0.292  0.360  0.398  104

These MLEs raise the NeL conversion factor to .95/.90 for 1931-36.
These MLes drop the MxL conversion factors to .80/.64 for 1938, .82/.67 for 1939-40,
and .87/.76 for 1941.

One check on possible under-reporting of stolen bases is to compare his Negro League rate (144 SB in 3444 AB, or 4.2%) with his rates in the Cuban League (9.0%, see post #18) and the Mexican League (5.3%, or 6.4% if you leave off his last year; see post # 78).

Let's try that again, with square brackets around the <pre> tags . . .

Cool Papa Bell – alternate MLEs



These MLEs raise the NeL conversion factor to .95/.90 for 1931-36.
These MLes drop the MxL conversion factors to .80/.64 for 1938, .82/.67 for 1939-40,
and .87/.76 for 1941.

Thanks Brent!

What would be more helpful would be to check his CWL and MXL rates against other base-stealers in those leagues, if league-wide totals and/or leaderboards are available.

It doesn't surprise me that his rate in Cuban was much higher. Torriente's was much higher there as well. We can't know if Torriente's lower totals are also a result of underreporting, but we can see if Bell outdistanced other players in a league whose stats are fully documented. He was in his late 30s in Mexico, so I wouldn't expect him to be a league-leading base stealer, but if he was in Cuba earlier in his career, he ought to stand out there, if his deeds are to match his reputation.

For 1928-29, the Cuban League SB rate was 3.0% (118 SB in 3949 AB). Bell's 17 SB led the league by a good margin. Second place was shared by Isidro Fabré, Frank Duncan, and Chaney White, each with 8 SB.

1930 was a short special season. Charlie Dressen led the league with 5 SB in 49 AB.

For 1940-41, the league rate was 2.8% (189 SB in 6710 AB). Bell's 11 SB was 2nd behind Alejandro Crespo with 15 (in 198 AB). Two players had 10 SB (Helio Mirabal and Pedro Pagés).

For Mexico, Cisneros isn't organized by season, so it would take quite a bit of work to put together seasonal stats. He does list the league leaders for each season. For 1938-41, the leaders were:
1938 Agustín Bejerano 23
1939 Agustín Bejerano 14
1940 Sam Bankhead 32
1941 Burnis Wright 26

Chris--is Bell playing CF every year? Can I get a 2B/3B/HR breakdown? Should I add in HBP in extra PA at the league average rate?

Yes, all CF.

Estimated Bell hit breakdown:

2618 singles
477 doubles
194 triples
126 home runs

Here are Chris's latest MLE's for Bell in my WARP.



3-year peak: 15.2
7-year prime: 31.8
Career: 60.4
Salary: $140,578,275, definitely not a HoM'er. This is down by Roush, Bobby Veach, and Tim Salmon.

Comments: Ouch. Chris is giving Bell not just a very long 22-year career, but one with remarkable durability: a career SFrac of 20.07 would represent the sixth-longest career of anyone who debuted after 1893 (not counting any extra credit)--only Rose, Yastrzemski, Aaron, Cobb, and Rickey accumulated more in the bigs. Despite that, he can only manage 60 WARP, which is plenty for a guy with a peak but not nearly enough for someone whose career is this long and this flat. 11-12 fielding wins certainly seems right on to me, given what I've heard about Bell's reputation. Five baserunning wins is rather low, given that the max is about 10-11; I'd subjectively split the difference and credit him with an extra 2.5 baserunning wins. But those aren't going to get him up to HoM territory, especially since I doubt he would have accumulated this many games played in the big leagues. I'd be inclined to give Bell a nice bump based on reputation, but these MLE's are so poor that even substantial additional reputational credit couldn't make him more than a borderliner. I think the verdict is clear: Cool Papa Bell was a rich man's Lou Brock, famed for his speed and with impressive longevity, but ultimately just not valuable enough to compete at this level.

Thanks guys!

Chris is giving Bell not just a very long 22-year career, but one with remarkable durability: a career SFrac of 20.07 would represent the sixth-longest career of anyone who debuted after 1893 (not counting any extra credit)--only Rose, Yastrzemski, Aaron, Cobb, and Rickey accumulated more in the bigs.

On my measure, 20.07 full seasons would be sixth-longest in the majors 1871-2006 and fourth in DanR's timespan. Ahead of Ty Cobb and Rickey Henderson, behind Anson & O'Rourke from the early days.

I give equal credit to a leadoff batter with five PA, the ninth batter with three PA, and the latter's 8th-inning pinch-hitter with one PA.

While we can't really measure the quality of the 1930s Negro leagues or Mexican League relative to the major leagues, we should be able to accurately assess their quality relative to each other, based on the fact that quite a few players played in both leagues. Did players other than Bell show such large increases in performance when they moved to Mexico? If not, then maybe Bell's improvement was a real late-career surge.

Did players other than Bell show such large increases in performance when they moved to Mexico? If not, then maybe Bell's improvement was a real late-career surge.

Yes, I think so. Monte Irvin had by far his best season there. Josh Gibson destroyed the Mexican League even more thoroughly than he did the Negro Leagues.

With the caveat that I really ought to do new MLEs for Gibson based on the HoF data (I was relying on Holway for some of Gibson's late career numbers, and I am pretty sure they are distorted by Holway's desire for tragic hero-building), I think that both Irvin and Gibson's MLEs, using current projections, would show them as having their best seasons in Mexico.

Even with the lowered conversion factors, Bell still shows very well in Mexico, so I don't doubt that he did have something of a late-career surge. But was it of historic proportions?

The NeL-MxL comparison is not as straightforward as it may seem, though, because the number of NeL players there varied significantly from year to year. When Bell went to Mexico in 1938, he was nearly the only North American black in the league. In 1939 he was joined by some others, and then in 1940 and 1941 many NeL stars, perhaps even the majority in 1941, came to Mexico. By 1942, the numbers start dropping again. So competition levels in Mexico and the NeL probably varied a lot from season to season during this period, because of these player movements.

Cool Papa Bell's Real Stats