hot topics

1928 Judy Johnson
Hilldale + Colored All Stars

Batting
G-63 (teams 64)
AB-230
H-56
D-9
T-5
HR-1
R-26
W-9
HP-2
SF-2
SH-0
SB-2
AVE-.243 (eastern NeL .282)
OBA-.276 (eNeL .333)
SLG-.339 (eNL .383)

His one home run was a game-winning grand slam. Johnson seems to have had a reputation as a clutch hitter; Rollo Wilson, who liked Beckwith, liked Johnson, too, at least partly because of Johnson's performance in the 1924 World Series.

Fielding-3b
*-led east (ECL disbanded in late May; Hilldale left the league before the season started)
G-62
DI-536*
PO-69*
A-97*
E-8*
DP-5
RF-2.79 (eNeL 3b ave 2.73; Jud Wilson led with 3.04)
FPCT-.954* (eNeL 3b ave .919)

He led in most of the counting stats because he played way more games at 3b than anyone else in the east (Jud Wilson is second with 39 games).

At some point I want to do a real fielding analysis, but I'd like to have multiple consecutive seasons in the database first.

1921 Judy Johnson
NNL associate Hilldale Club

G-25 (team 41)
AB-87
H-22
D-2
T-4
HR-2
R-11
W-5
HP-0
SH-3
SB-1
AVE-.253 (NeL .263)
OBA-.293 (NeL .324)
SLG-.437 (NeL .361)

His rookie season. He started the year with the Madison Stars, a Philadelphia club that served as a sort of informal farm team for Hilldale. The Stars played several major teams in the east, including both Hilldale and the Bacharachs, but I haven't compiled their stats.

Here's Holway's data for Judy Johnson's career


In League Play
1921 .229 for Hilldale; ss
1922 .240 for Hilldale; ss; all-star ( a _very_ dubious selection over Lloyd or Lundy)
1923 .313 for Hilldale; 3b; 21 2b (2nd), 7 sb (2nd); all-star
1924 .324 for Hilldale; 23 2b (1st), 5 3b (2nd); 3b; all-star
1925 .332 for Hilldale; 8 3b (2nd); 3b; all-star
1926 .312 for Hilldale; 16 2b (4th), 6 3b (2nd), 6 sb (3rd); all-star
1927 .220 for Hilldale; 3b
1928 .234 for Hilldale; 3b
1929 .412 for Hilldale; ba 4th in league, 22 2b (3rd), 23 sb (3rd)
1930 .250 for Homestead; 3b
1931 .248 for Phi Stars; 3b
1932 .304 for Hilldale, .235 for Pgh Crawfords; 3 3b (4th); 3b
1933 .281 for Pgh Crawfords; 13 2b (5th); 3b
1934 .232 for Pgh Crawfords; 3b
1935 .263 for Pgh Crawfords; 3b
1936 .239 for Pgh Crawfords; 3b
1937 no data

Postseason play
1921 2-18 in World Series vs. Chi Am Giants
1924 15-44 in World Series vs. Monarchs
1925 6-24 in World Series vs. Monarchs
1929 6-14 vs. Chi Am Giants in World Series
1930 12-43 in playoff vs. New York
1935 1-10 in playoff vs. Cuban Stars

Cuban Play
1926 43-115 in Cuban Play
1927 44-133 in Cuban Play
30-88 in Cuban Play

Vs. Major-league competition
1923 6-20
1929 2-4
1931 3-11 vs. Major-League competition
1932 5-25 vs. major-league competition
1935 3-4 vs. major-league competition

Career
.275 mean average for sixteen seasons
1100-3853, .285 avg. according to Holway
28-105, .267 vs. major-league competition, according to Holway
Black Ink 2, Gray Ink 24


Brief comments. It's pretty clear from this data that Judy Johnson would have been, at best, an average hitter in the majors. He probably was a top-notch defensive third baseman, but that's not enough to make him a serious candidate for the HoM. I don't plan on working up a win-share estimate for him, though if one is greatly desired I could get to it at some point.

Let's not see Judy fall out of sight. Rabbit Maranville is earning votes, and he was even worse offensively than Johnson's presumptive MLEs. Although outstanding SS defense trumps outstanding 3B defense.

I wouldn't mind seeing it Chris. 3B have gotten in with generally fewer WS than at other positions for whatever reason. He could be comparable to someone like Jimmy Collins.

I wouldn't mind seeing it Chris. 3B have gotten in with generally fewer WS than at other positions for whatever reason. He could be comparable to someone like Jimmy Collins.

Well, I'll get to it when I can, but I can tell you now that he's not going to show up comparable to Jimmy Collins. Collins was an A to A+ fielder with a 113 career OPS+, which came mostly from his good power: he was only slightly above average in batting average and OBP. Dick Lundy is this sort of hitter, actually, though his slugging is not quite as impressive in a 1920s context as Collins's was in an 1890s-1900s context. Judy Johnson was not, for his career, an above-average hitter in a major-league context.

He was, as Ardo points out, a better hitter than Rabbit Maranville, but his defense was not as valuable and his career was much shorter than the Rabbit's.

Charleston and Johnson seem pretty clear N-Bs to me, one in each direction. Bill Foster and Lundy are much more difficult, so perhaps that's where Chris could devote his splendid efforts to get an MLE WS

He was, as Ardo points out, a better hitter than Rabbit Maranville,

I don't see how this is deduced. Maranville hit .258 career. Johnson hit .285, which translates to .248 after applying the 13% NeL discount. Maranville also played @40% of his career in the deadball era. (When did the lively-ball era begin in the NeLs?)

What am I missing?

Did Johnson ahve more power than Rabbit? Did he walk more?

I don't see how this is deduced.

In the first third of Johnson's career, the offensive level in the Negro Leagues was lower than in the majors, so I think that the straight 13% discount drops his average too much. It also appears that Johnson's power was somewhat better than Maranville's: I think he would have had 3 or 4 seasons with a slugging percentage right around .400. But it's possible that closer inspection would show that he is not any better than Maranville at all.

Incidentally, looking at Maranville's record made me wonder if he lost it as a hitter after 1919 or if he was simple unable to take advantage of the lively ball? If he could have sustained the level of hitting he showed in 1917 and 1919 for just a few more years, he'd have a _much_ stronger HoM case . . .

So we are coming to the conclusion that Judy Johnson was no better than Rabbit Maranville? Seems like he is the opposite of John Beckwith. JOhnson is highly regarded by most of the 'experts' yet his performance is much lower than Beckwith's, who isn't generally liked. If this Maranville comparison is correct then I would be surprised to see him in my top 30.

Incidentally, looking at Maranville's record made me wonder if he lost it as a hitter after 1919 or if he was simple unable to take advantage of the lively ball?

Some of it is park effects. Sometimes playing in a pitcher's park (or pitcher's era, or both for Maranville at peak) doesn't diminish a player's raw stats so much as it enhances the real value of what the player can actually do (see Biggio).

Rabbit's "power" was really based on speed - triples and leg doubles. From 1915-20, he played in early Braves Field, a huge park with 402' FOUL lines and over 500' to center. He got grey ink for triples in 1916,17,and 20 (lost 1918 to WWI). After he adapted to Forbes Field (Pit) he got some more ink in '22 and '24 (doubles too). When his speed declined (he's 33 when he left Pit), so did his "power". He probably was unable to take any real advantage of the lively ball. His average went up (like everybody else) but his power not so much.

Some more demographic modeling. (Assumptions and implications.)

Assume that an integrated MLB would have been about 12.5% black. This implies that at each of the 7 full-time positions (exclude catcher and pitcher), an average of two black players would make the majors (lots of variance though). IOW, the Negro League All-Star Game starters are on average above IMLB replacement level, but the runner-up All-Stars are probably not.

Assume that an integrated MLB would have been about 25% black. This bumps the average per position up to 4 (still lots of variance). IOW, the All-Star Game backups are also on average above IMLB replacement level, but those who didn't make the team are probably not (unless the position is very deep).

Assume that an integrated MLB would have been about 37.5% black. This bumps the average per position up to 6 (still lots of variance). IOW, the best players at each position not on the All-Star teams are also on average above IMLB replacement level, though unlikely to be above IMLB average (unless the position is very deep).

Assume that an integrated MLB would have been about 50% black. This bumps the average per position up to 8 (still lots of variance). IOW, there was approximate parity between the Negro Leagues and MLB at the playing level.

jimd,

I tend to imagine that integration would provoke expansion: two eight-team leagues would become two ten-team leagues, with the new players, equal to 1/4 of the former total, being black and/or Latin.

This imagined model posits that the top 4 black/Latin players at each position would become ML regulars, but that replacement level for the major leagues would not change.

Applying this model to third base during the 1920s, Judy Johnson and Oliver Marcelle were second-team all-stars in the NeL: they would be typical starters in the majors, but not HoMers. John Beckwith and Jud Wilson, the first-team all-stars, would be higher than that -- how high it is our business to determine.

The National Agreement at that time prevented MLB from changing from its existing configuration of cities; the minor league clubs had territorial rights, also. Landis would likely have overruled any attempt to change along those lines (by moving into minor league territories - the Browns and Braves were frozen into their predicaments.) I'm sure it's no coincidence that no attempt was made during his lifetime. I'm also sure each owner would have strongly resisted any attempt to add another major league team in his city.

Of course, Landis was also opposed to an integrated MLB. So if there was no Landis then maybe both become possible, though the cost/benefit of buying territorial rights to Baltimore/Milwaukee/Kansas City/Buffalo is different in the 20's/30's as opposed to the 50's.

Warning: unreliable

My understanding is that through principal ownership or interlocking directorship, some mlb clubs had control or great influence over their minor league affiliates.

Boston N over Milwaukee and NY N over Minneapolis in 1950?

I think that the All-Johnson team is the best of the all-one-last-name teams. (Front-line pitching overcoming that advantage the Williams team has in the outfield.) But who's the third baseman? Are we better off with good-glove weak-bat Judy? Or should we go the other way and use Howard? (We need Grant at SS so he's not available for 3B.)

How does Johnson compare to Marcelle, who, as I recall, was another questionable hit, great fielding NL 3B.

Conventional wisdom is that Johnson was a better hitter, Marcelle a better fielder, and that in overall value they were very comparable, except that Johnson had a longer career (16 seasons in top competition to Marcelle's 12).

I'm not sure the evidence supports saying that Johnson was clearly a better hitter than Marcelle. See the stats on the Marcelle thread. Also, what evidence I've seen (and it's still pretty limited) indicates that Marcelle spent most of his career in a fairly neutral park, while Johnson spent much of his in a hitters' park.

OK, I've been working on an alternative way of calculating MLEs from the one I'd previously used. I think it ends up being the same thing in the end, but what the hell. (Chris, sorry in advance if I'm ending up doing exactly what you do!) Anyway, I used the Willie Davis comment in the NHBA as the basis, and using runs created (tech-3 version sans RISP) as the basis rather than league AVG/SLG totals. In the end since those items together form a net conversion rate that is very similar to using a runs converter, I'm not sure there's a hill of beans difference. I worked up Judy Johnson as my test case, and so I'm offering it up here with my methodology for everyone to have at.

METHOD
1) Assessing his actual record.
I took the Shades of Glory stats for Johnson and figured his tech-3 RC (as elaborated by James in Win Shares, the book). I included an estimation of Johnson's CS in this calculation, figuring him to be competent at 70%, which seemed like a reasonable compromise. I adjusted this RC figure where possible with park factors that had been previously used by Chris Cobb, mostly for the two Crawford years. Gary A's also got a 1-year ECL pf for a Hilldale season, which I regressed 75% toward the mean. I took all other seasons as 1.00 pf without any other info to the contrary.

2) Figuring the league's record.
To translated Johnson, I needed to come up with a league-average player comparison, which meant piecing together some data here and there. The HOF committe's .pdf contains combined NgL league average and slugging for 1920-1948. In addition, I did a little research this weekend on walk rates, using the biggest sample I could create. That study indicated that the NgLs were, on the whole, walking a little more than 90% as often as MLB. So I took MLB walk rates and decreased them by 10% to get a rough NgL walk rate. I also had done similar research that suggested the NgLs SHed 35% more often than MLB, so I incorporated that too. Finally, I gave the NgLs a 5% bump upward from the MLB SB rate in a nod to the conventional wisdom of their offensive style of play. Finally, I used the tech-3 formula to figure out what a league average player would have done in Johnson's playing time.

3) Translating to a neutral MLB enivronment.
A few steps down the line, I knew it would be important to have a context-neutral translation of all Johnsons's seasons. So I compared Johnsons's RC to his league's, then applied the same to a netural 4.50 R/G league. Now comes the conversion to MLB. I converted his RC at a rate of .85. Chris has used a .90/.81 previously. I chose to try something different to see where it would take me, and this should not be read as a diss of his approach at all...after all I'm cribbing aspects of it left and right! Anyway, once I had my converted RC, I used the James technique described in the Willie Davis comment to generate a neutral, 4.5 r/g translation of Johnson's exact lines, no projecting for playing time. I made one small change to the routine; I added 10% to Johnson's walk rate to account for the difference in the respective leagues' walk rates. Given the next sentence, I chose not to make a similar adjustment for his SACs since I think he'd have been doing a lot of it. The resulting translation showed him hitting a desultory 255/288/346, and walking about 5 times for every 100 ABs for his career. His best year 304/345/463.

4) Filling out the seasons.
I have resisted regression in the past, and I guess this isn't strictly regression since I'm not using regression formulae, but in essence I think I'm doing the same thing. To fill out Johnsons's seasons I'm
a) making a subjective guesstimate about his MLB playing time
b) using his career, neutralized stat line to fill in the difference between his playing time and his estimated MLB playing time.

The questions with this approach are well known to all of us (lowering peak seasons), and I think Chris uses a regression of the player's three or four previous years. On the other hand, I've come to realize for myself that just flipping it up to 162 is good for some things and less so for others, so I'm trying something else. Here's an example of how regression can have a big dampening effect. Remember Judy's best year noted above? 304/345/463 in a neutral context. Johnson did that in 1925 in 250 ABs and 181 outs. I'm estimating him for a full season of play at 400 outs and about 530 ABs. That means there's 219 outs and 287 ABs left to fill in. As I mentioned early, Johnson has a Kikoesque career line of 255/288/346. So I plug that into his remaining outs and here's what happens:

Transed: 304/345/463 in 250 AB, 181 outs
Regress: 255/288/346 in 287 AB, 219 outs
TOTAL: 280/316/402 in 536 AB, 400 outs

Granted he doesn't come out like Al Pedrique, but it's obviously a very different season with "regression" operating. As I said, I'm not literally employing regression, but I think what I'm doing is essentially equivalent. If not, please let me know!!!!

OK, so once I do this, I have a neutral translation with playing time filled in by regression, and Johnson is, of course, still a 255/288/346 career hitter. His best season is now 1924, 280/316/405, just a smidge better than the 1925 season above.

5) Final translation into MLB setting.
Here's a place where I'm parting with Chris. I prefer the consistency of seeing a player in one league versus the ups and downs that alternating leagues sometimees yields. Again, this is a personal preference, not a critique, and so I just chose that direction instead. I chose to put Johnson in the NL of 1921-1936. To translate him to this league, I used the same process as earlier, comparing him to the league average player and then applying that to the average NL player based on the league's R/G. This step reintroduces increased variation in Johnson's batting line and improves it a bit. He's now a career 260/292/352 hitter. For his career, he created 826 translated runs. That's 4.07 per game. The league in his time was much closer to five than four.

6) WS analysis.
I've decided to try something other than short-form WS. I've decided that I'm unsure about its universal application to players in more extreme run environments. So I looked in WS the book for help, and it turns out that I think there's almost as short way to do it that may improve accuracy a bit for MLE purposes. Here's how it goes:

a) Multiplying lg R/G by scheduled games (154 in this case)
b) Multiply a by .52 to get the submarginal runs
c) Subtract b from a to get the marginal runs
d) Figure the available BWS by multiplying half the scheduled games (77) times 3 and multiply that by .48. For a 154 season, it's 111, for a 162 it's 116.
e) Figure the league's runs per outs
f) Multiply the player's outs times e and multiply by .52. That's the player's submarginal RC. Now subtract that from his total RC to get his marginal outs.
f) Divide e by c and multiply by d to get the player's BWS.

I might have goofed that, but I think that's it.

Then use the SF method's defensive allocation for FWS. For Johnson, his defensive reputation is excellent, so I upped him to 5.5/1000 innings (or 118 games). I estimated his games by figuring his PA divided by 4.1 and then applied that to the SF estimate.

OK, that's "it." I hope that makes sense to everyone. Here's the estimation of Johnson's career.



In a word, youch. Makes Ray Dandridge look like an All-Star.

I think when we mention Kelly, Marquard, McCarthy and Haines, we need to add Johnson.

we need to add Johnson

Despite the fact that my own MLE suggests Johnson's an awful HOF pick, I'm going to tentatively disagree with you on this John. I don't think the HOF NgL comission of the 1970s had particularly good numbers to work with...if any. I don't know this for a fact, but there's lots of reasons to believe that was true. (I hope someone will corroborate my speculation or countermand it.) Even with player stats, they surely had little sense of the league's batting norms, or of anything except the oral history and their own recollections. Johnson was a very strong candidate via the oral tradition since he exceled at the hustle skills/stats: fielding, running, bunting, contact, and gamesmanship/clutchiness/leadership. And he played on numerous strong or championship caliber clubs. With nothing more than oral tradition to guide them, I can't really blame that committee for selecting him. To use a modern example, Garrett Anderson or Alfonso Soriano would look great if all you knew was their AVG, HR, RBI, SB, and nothing more about their leagues or their other stats. Especially if you didn't have a deep sense of every single candidate's numbers too due to the spottiness of the record.

Anyway, yes, Johnson is retrospectively less defensible a selection, but in the moment that committee operated in, he I can't imagine he could have been an obvious error.

In a word, youch. Makes Ray Dandridge look like an All-Star.

I think when we mention Kelly, Marquard, McCarthy and Haines, we need to add Johnson.

I agree with both of these statements. First, I think that we actually are underestimating Ray Dandridge. He was a very good player for a very long time. I'm not arguing that he should be in anybody's top 15 (he's not in mine) but he should probably be in a few more top 100's than he is.

But, even as big a proponent of the Negro Leagues as I am, I just can't make the case for Judy Johnson. He was a flashy glove man who used his speed to turn some singles into doubles and doubles into triples. But he was a flashy glove man at third rather than short, and he had pretty much no patience of power. I was going to compare him to Orlando Hudson but even that's giving Johnson too much credit.

I agree Judy Johnson is among the weakest members of the HoF.

I am not sure, though, that Eric's approach to regression isn't underrating his peak somewhat. I never did a full workup of Judy Johnson, so I don't have data to offer in comparison, but my system regresses players to their average play over a five-year period. If a player was significantly above his career averages for a period of several years, the effect of regression will probably be less than would result from filling in partial seasons with career averages. It would be interesting to see how the results of Eric's system compare with mine.

I think when we mention Kelly, Marquard, McCarthy and Haines, we need to add Johnson.

[]bDespite the fact that my own MLE suggests Johnson's an awful HOF pick, I'm going to tentatively disagree with you on this John. I don't think the HOF NgL comission of the 1970s had particularly good numbers to work with...if any.

How's that differ for Tommy McCarthy? Heck, the Old Timers committee in the 1940s was further removed from McCarthy's era than the Negro Leagues committee was from Johnson's day.

Johnson was a very strong candidate via the oral tradition since he exceled at the hustle skills/stats: fielding, running, bunting, contact, and gamesmanship/clutchiness/leadership. And he played on numerous strong or championship caliber clubs.

Yep. Sounds exactly like Tommy McCarthy. Should we cease to mention McCarthy among the worst HoF picks?

Chris, you are correct that I was underrating Johnson. I'm working on Charleston too (since I don't think either of us ever worked him up), and he was coming out really low. After going to a party with free-flowing drinks, loud music, and yummy appetizers, I realized that deep in the bowels of my spreadsheet I was using Johnson's outs to determine the lg RC/G instead of using the est. lg outs to determine it. Which skews the results away from strong individual performances. In changing this, it raises Johnson us a bit toward the league average raising his OPS+ from 69 to 77. Here's version 1.1 of his MLEs.



I think that's a little more like it, particularly since he now has three seasons in his prime where he's a good All-Star candidate. This MLE makes him much closer to Dandridge than the last one did, giving them roughly similar career MLE WS totals.

He's still not very good outside of those three seasons, however, and even they aren't great years.

Eric, you have an error in your line for 1929. 13.8 bws and 7.1 fws add up to 20.9, not 2.9.

And there's a similar error in 1933.

Thanks, Chris F! (Blasted conversion to word....) The sum total is correct, so I'll just put the correct seasonal totals below for clarity's sake:


So per my previous comment, four A-S or near A-S type seasons here, but no MVP caliber years by any stretch of the imagination.

Round Two. After hashing some things out over on the Charleston thread, I've taken the same method for the second version MLEs for Oscar and applied them to Johnson. It actually helps Judy just a little. If anyone's got any ideas on playing time, I'm all ears. I put him into a pretty bellish shaped arc, but it's not at all definitive.

Here is Judy Johnson at the SABR Bioproject.

Judy Johnson's Real Stats

Please click on through to the other side for my latest Judy Johnson MLEs.

We are missing some info on Johnson, so MLEs will be updated as new info comes along.

Currently showing 29.4 WAR.