Baseball for the Thinking Fan

Login | Register | Feedback

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

Wednesday, July 06, 2005

Leroy Matlock

Leroy Matlock

Eligible in 1948.

John (You Can Call Me Grandma) Murphy Posted: July 06, 2005 at 09:21 PM | 67 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

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

   1. John (You Can Call Me Grandma) Murphy Posted: July 06, 2005 at 09:32 PM (#1453403)
Posted by Grandpa Simpson on July 06, 2005 at 5:32 PM

I want my Matlock!
   2. Chris Cobb Posted: July 06, 2005 at 10:17 PM (#1453497)
Thanks, John!

First, here's Matlock's data from Holway and Riley.

Leroy Matlock Data

From Riley

Born March 12, 1907

Teams: St. Louis Stars, 1929-31, Detroit/Homestead/Washington 1932, Pittsburgh Crawfords 1933-38, Santa Domingo 1937, Venezuela 1939, Mexico 1940-1942, New York Cubans 1942

Seasonal Data from Holway

1929 5-2, team 70-26, #5 in team decisons, -0.1 WAT
1930 10-3, #3 in league in wp, team 68-24, #2 in team dec. (tie), 0.5 WAT
1931 1-3, team 23-17, #5 in team dec., -1.4 WAT
1932 4-1 for Homestead (team 64-28), 2-5 for Was (team 16-35), 0.4 WAT
1933 11-4, 4th in wins, 4th in wp, 4th in TRA, 4th in K, team 49-31, #3 in team dec., 2.2 WAT, all-star
1934 14-3, 3rd in wins, 2nd in wp, 5 in K, team 63-22, #3 in team dec., 1.8 WAT
1935 17-0, 1st in wins, 1st in wp, 1st in TRA, team 47-17, #1 in team dec., 6.1 WAT, all-star, GSA
1936 9-3, 3rd in wins, 3rd in wp, 1st in TRA, team 39-27, #2 in team dec., 2.3 WAT
(Note, all his dec. came in the first half of the season. Pitched a lot then, and not at all thereafeter. Holway says he doesn’t know the reason, but Riley has him playing on a black all-star team that entered and won the Denver Post tournament.)
1937 4-1 in S.D., no team data, 1-0 for Pgh, team 12-16, 0.6 WAT
1938 3-7, team 24-18, #2 in team dec., -3.6 WAT
1939 No Data: in Venezuela
1940 15-10, 3.27 ERA, 101 K in Mexico
1941 15-7 in Mexico
1942 9-13 in Mexico, no data for NY Cubans

Career
77-31, .713 in NeL competition
4-1 in Santa Domingo
39-30 in Mexico
Black/Gray Ink 15/51 (15th in black ink, 18th in Gray)

Fuller Mexican data is available on the Nel Pitchers' thread (thanks to Dr. Chaleeko), as are rankings of Matlock's wins and winning percentage. I'll try to get that transferred over soon.

I'll note that Matlock's career rankings in the NeL underrepresent the scope of his career, in that he spent five seasons, 1937 and 1939-42, pitching mainly outside of the U.S. The evidence is that that he pitched quite well in at least 3 and probably 4 of those seasons.
   3. Chris Cobb Posted: July 06, 2005 at 10:20 PM (#1453505)
Now Matlock's MLE projections, using the same methods I've applied to Byrd, Brown, and Hilton Smith:


Year  IP DERA DERA+  snW-L      wp    ws
1929  78 4.50  100   4.5 - 4.5 .500   4.3
1930 163 3.98  113  10.5 – 8.2 .561  12.4
1931  91 6.43   70   3.4 – 7.1 .328   0.0 (-0.3)
1932 206 5.11   88  10.3 –13.4 .436   7.2
1933 239 3.72  121  16.3 –11.2 .594  21.5
1934 264 3.46  130  19.1 –11.2 .628  27.2
1935 303 2.65  170  25.9 – 8.9 .743  43.1
1936 281 3.44  131  20.4 –11.9 .632  29.0
1937 248 2.65  170  21.2 – 7.3 .743  35.2
1938 102 6.92   65   3.5 – 8.2 .297   0.0 (-1.1)
1939 206 4.02  112  13.1 –10.5 .556  15.8
1940 253 3.88  116  16.7 –12.4 .574  21.1
1941 222 4.41  102  13.0 –12.5 .510  13.6
1942 160 5.00   90   8.2 –10.2 .448   6.3
    2816 3.96  114 186.1-137.5 .575 236.7 
                  (182.2-141.5 .563 224.4)



Notes

1. There are two career lines listed. The top career line shows Matlock’s record as it adds up, season-by-season. The career DERA and DERA+ are calculated by converting the DERA and IP into runs allowed, adding up the runs, and using that total to calculate career DERA. The career w-l record and winning percentage, when added up, are higher than they would be if calculated off of career DERA+, so the second career line, in parentheses, shows the career line that the career DERA+ indicates. A fraction of the difference has to do with the two seasons in which Matlock earned negative win shares. These zero out in a seasonal count, but they are factored into the career runs allowed totals. The rest probably has to do with upward distortions of Matlock's value in his two years of 170 DERA+: using pythaganpat/port (the variable exponent thing), would probably correct for this.

2. Two seasons in this estimate are particularly sketchy: 1937 and 1939. In 1937, most of Matlock’s play was in Santa Domingo, where he went 4-1. He was also 1-0 for the Pgh Crawfords. I don’t have a team record for the Dominican to work with, so I estimated a .500 team. This could well be low or high. In 1939, Matlock pitched in Venezuela, and no data is available for that season. So I took the average of the four surrounding seasons to get the totals for that year, just as I would do in the regression analysis for position players.
   4. John (You Can Call Me Grandma) Murphy Posted: July 06, 2005 at 10:34 PM (#1453527)
Thanks, John!

No problem, Chris. You should also thank Marc for pointing out that you had requested a thread for Matlock earlier.

Boy, that '35 season was something, wasn't it?
   5. Michael Bass Posted: July 06, 2005 at 10:39 PM (#1453539)
Interesting, Chris...certainly a good comp for Hilton Smith I think.

I'm curious about the 1937 season...the record seems awfully thin to support a hefty WS estimate.

Man, that 1935 is impressive, though.
   6. Dr. Chaleeko Posted: July 06, 2005 at 10:43 PM (#1453554)
Just to double-check something because I don't have my source materials in front of me at this time...

Did Matlock quit baseball because by 1942 he was stinky? Or did an injury force him into retirement? Or did he serve in the military?

Also just as a way of offering some context for Matlock.

Here's seasons where his DERA+ would have ranked in the NL's top 5 of ERA+:
-1933 t-fifth with French (leader Hubbell at 193)
-1934 fifth (leader Hubbell at 168)
-1935 first (leader Blanton at 159)
-1936 third (leader Hubbel at 169)
-1937 first (leader Turner at 151)

So that's not a bad peak at all. In addition he had shoulder years of 116, 113, 112 to round out his prime, plus two more years at 102 and 100 to give him ten very solid years.

Then there's the three crappy years....
   7. Chris Cobb Posted: July 07, 2005 at 02:33 AM (#1454333)
Did Matlock quit baseball because by 1942 he was stinky? Or did an injury force him into retirement? Or did he serve in the military?

There's no information in Riley's bio about any of this, so if there is anything more need to be known, we'll need help from an expert. Gary A., gadfly, any knowledge about the ending of Matlock's career?

Since he was 36 in 1943, it seems more likely that he was washed up or injured than that he entered military service, but it's possible that he did, or that wartime industrial jobs paid better than what he could make playing baseball at that point in his life.
   8. Chris Cobb Posted: July 07, 2005 at 03:04 AM (#1454397)
I'm curious about the 1937 season...the record seems awfully thin to support a hefty WS estimate.

There are more uncertainties here than for many seasons, but there's also quite a bit of evidence that can be pieced together.

Well, here's my thinking about Matlock's 1937 season, looking at evidence of durability (for the IP projection) and effectiveness (for the winning percentage and durability projections).

1) There's no report of any injury.

2) He went 4-1 in Santa Domingo, in a 30-game season; not all his decisions were necessarily recorded. This suggests that he was pitching regularly.

3) His 4-1 record was one of the best in a league with a lot of very good pitchers; he tied Paige for best winning percentage and is in the middle of the pack on recorded decisions: Here's the list from Holway, reordered by # of decisions:

Ramon Bragana 4-7
Satchel Paige 8-2
Martin Dihigo 6-4
Bertrum Hunter 4-5
Rodolfo Fernandez 4-5
Leroy Matlock 4-1
Chet Brewer 2-3
Luis Tiant 1-3
Bob Griffith 2-1
Lazaro Salazar 0-2
Silvio Garcia 0-2
Spoon Carter 1-0

(See Dr. Chaleeko's post above for more on the Dominican season).

5) According to Holway, when they got back to the U.S., the players who had jumped to the Dominican were blacklisted. This explains the lack of NeL data for the rest of Matlock's 1937 season. However, there's a bit more evidence to consider. The blacklisted players formed a barnstorming team, headlined by Paige, of course, "The Dominican All Stars." They entered and won the Denver Post Semipro tournament (in which Paige and Matlock had also pitched in the second half of the 1936 season). Holway's concluding note on the tournament is: "Leroy Matlock was voted best pitcher." So, although there are no Negro-League stats for the rest of his season, there is more than anecdotal evidence that he continued to pitch very well for the rest of this year. It appears that the Dominican All Stars continued to barnstorm after the tournament; Holway describes them playing a series at the Polo Grounds in September against a squad of NNL all-stars.

5) Given that he was pitching regularly and highly successfully, I see no case that he shouldn't be projected out to a full season of highly successful work. So I used my normal projection methods as best I could on the limited data. This is a case where some good method of regressing pitchers would be helpful. Matlock definitely would have a very high baseline to regress towards for this season, though!

That's the picture as I saw it when I did the projections. I hope that makes the projection seem a bit more reasonable.

However, there's another factor:

In reviewing all the evidence, I noticed that Holway strongly implies that Matlock and Paige were teammates on Trujillo's team in the Dominican, which won the pennant by 3 games. I had not picked up on this the first time through, so I had projected Matlock's team as a .500 team. They were certainly better than .500, so Matlock's projected DERA+ should be lowered somewhat from 170. With Paige and Matlock going 12-3 in half the team's games, though, I'd be inclined to give team pitching slightly more than the usual 35% credit for team success.

If anyone can find the standings for this season, that'd be a big help in making this projection more precise.
   9. John (You Can Call Me Grandma) Murphy Posted: July 07, 2005 at 02:51 PM (#1454923)
I don't see him as HoM material, but he does look better than Dizzy Dean, IMO.
   10. karlmagnus Posted: July 07, 2005 at 03:15 PM (#1454972)
No, no. Dean had an ERA+ of 130, not 114. It's not close. Trust me, Joss is the short career guy with the best case if you give ANY credit for missing seasons (and I see no justification for crediting WWII and not Dead).

Joss, interestingly, is also SIGNIFICANTLY better than Koufax -- 2327@142 vs. 2324@131. It's not close.
   11. TomH Posted: July 07, 2005 at 03:37 PM (#1455029)
Joss, interestingly, according to strictly using ERA+ and career IP, was SIGNIFICANTLY more valuable in his day than Koufax was in his; -- 2327@142 vs. 2324@131. It's not close.

But I laugh at the thought that this makes him Better.
   12. John (You Can Call Me Grandma) Murphy Posted: July 07, 2005 at 03:38 PM (#1455035)
No, no. Dean had an ERA+ of 130, not 114.

...but almost 1,000 less innings.

Joss, interestingly, is also SIGNIFICANTLY better than Koufax -- 2327@142 vs. 2324@131. It's not close.

Joss was not better than Koufax. Compare where they placed in the top-ten for innings. Koufax was the more durable pitcher, without a doubt.
   13. karlmagnus Posted: July 07, 2005 at 03:45 PM (#1455055)
It's difficult to be durable when you're dead.

Koufax was overworked, and blew his arm out -- normal baseball injury at that time. In modern conditions, he might have had successful surgery, but would have pitched many fewer innings/year. Joss, if transported to 2005, would have stayed alive and probably doubled his career totals to well over 300 wins (which he wouldn't have if still alive in 1911-19 -- pitchers pitched more innings, but didn't last as long; the higher "top 10 for innings" is pretty well exactly counteracted in 1900-10 (not in 1885) by shorter careers.

Joss is my new pet! (albeit only #7 on my tentative 1956 ballot).
   14. John (You Can Call Me Grandma) Murphy Posted: July 07, 2005 at 03:51 PM (#1455070)
It's difficult to be durable when you're dead.

I was talking about comparing the two per season, not for their careers. Koufax was in the top-five four times, while Joss made the top-ten only twice.
   15. John (You Can Call Me Grandma) Murphy Posted: July 07, 2005 at 04:43 PM (#1455236)
Win Share Comparison:

Dean: 37, 31, 31, 24, 22, 17, 9, 7, 1, 1, 0, 0

Matlock: 43, 35, 29, 27, 22, 21, 16, 14, 12, 7, 6, 4, 0, 0
   16. karlmagnus Posted: July 07, 2005 at 05:21 PM (#1455373)
Knock off the 43; that's presumably 1937. Justifying that kind of extrapolation from a 4-1 record in a 3-team Dominican Republic is cuckoo, as Chris Cobb very sensibly said.

WS is a useless and unreliable metric anyway; W/L and ERA+ work much better.
   17. TomH Posted: July 07, 2005 at 05:24 PM (#1455384)
Even if Matlock's projected WS as given above WERE in the ballpark, and we were convinced he was better than Diz, that wouldn't put Leroy in HoM territory, since Dean finished 40th last week.
   18. Dr. Chaleeko Posted: July 07, 2005 at 05:39 PM (#1455423)
All this said, I'd still like to know what happened to him after 1942.

Calling the Gadfly! Come in, Gadfly!
   19. John (You Can Call Me Grandma) Murphy Posted: July 07, 2005 at 05:41 PM (#1455438)
Knock off the 43; that's presumably 1937. Justifying that kind of extrapolation from a 4-1 record in a 3-team Dominican Republic is cuckoo, as Chris Cobb very sensibly said.

Fine. Knock it down to half and his career would still be at least comparable, if not still better than Dean.

WS is a useless and unreliable metric anyway; W/L and ERA+ work much better.

You keep ignoring IP for some reason, while W/L is "useless and unreliable" if not placed in the proper context.

Even if Matlock's projected WS as given above WERE in the ballpark, and we were convinced he was better than Diz, that wouldn't put Leroy in HoM territory, since Dean finished 40th last week.

I 100% agree, Tom.
   20. John (You Can Call Me Grandma) Murphy Posted: July 07, 2005 at 05:45 PM (#1455452)
All this said, I'd still like to know what happened to him after 1942.

Calling the Gadfly! Come in, Gadfly!


I agree with you, Eric. It would be nice to know if he went into the military or if his career just flamed out in '42.
   21. sunnyday2 Posted: July 07, 2005 at 06:08 PM (#1455547)
You've got your REALLY short careers at <2,000 IP, and then you've got your short careers at <3,000 IP.

Being a peak voter, Dean ranks better than 40th for me, but not in my top 25 at the present time.

If we should even think seriously about a short career, I still think it has to be Joss or Waddell.

Joss ERA+ 142 in 2,327 IP
Waddell 135 in 2,961
Dean 130 in 1,967

Anybody with a 114 is going to need 3,000+ IP AND a pretty good peak. Matlock has the peak but not the career.

Joss and Waddell have two out of three--which is to say, 1) effectiveness and 2) peak. Dean has one of three, which is a nice peak. But even in a REALLY short career his career effectiveness falls short.

But since Dean is a peak candidate, the fact that Joss, Waddell and Matlock all had better peaks is a killer.

Prime Performance

Joss 148 ERA+ in 2,220 IP
Waddell 152 in 1,772
Dean 133 in 1,727

Silver King 153 in 1,894
Reulbach 156 in just 1,262
Bridges 134 in 2,284
Gomez 138 in 1,977
Luque 141 in 1,397
Matlock ~144 in 1,335

Matlock looks like Dolf Luque to me for his peak/prime but falls way short on total career.
   22. John (You Can Call Me Grandma) Murphy Posted: July 07, 2005 at 06:42 PM (#1455696)
If we should even think seriously about a short career, I still think it has to be Joss or Waddell.

Waddell appears appreciatively better to me than Joss because he was far more durable than Joss.
   23. karlmagnus Posted: July 07, 2005 at 07:17 PM (#1455845)
Using the discredited method of differences again, the differences between Joss' prime (i.e. all but 100 innings at the end when he was ill) and Waddell's career is 741 innings at an ERA+ of 96 with a W/L of 33-46. Not without value, but more than counterbalanced by Waddell's historically high level of unearned runs, his inability to focus, and the phantom career Joss never had due to early death.
   24. karlmagnus Posted: July 07, 2005 at 07:20 PM (#1455858)
With 25% credit for posthumus play, Joss is 2777IP, 190-120 and an ERA+ of 136. Given the unearned runs, that beats Waddell's 2961IP, 193-143 and 135.
   25. sunnyday2 Posted: July 07, 2005 at 07:26 PM (#1455889)
karl,

Since both of your methods are in fact discredited--ie. the method of differences and the death credit--I'll still take Waddell. But my current pitcher list is:

1. Ray Brown--soon to be off my candidate list, leaving no obvious newbie and...

2. Tommy Bond--talk about discredited methods, my method is that he actually pitched baseball

3. Waddell
4. Jose Mendez
5. Joss
   26. John (You Can Call Me Grandma) Murphy Posted: July 07, 2005 at 07:32 PM (#1455917)
Not without value, but more than counterbalanced by Waddell's historically high level of unearned runs, his inability to focus, and the phantom career Joss never had due to early death.

The unearned runs is a reasonable concern, the inability to focus will show up in his stats already, and I refuse to give credit for Joss dying early.

Using the discredited method of differences again,

Well, you have discredited the method :-), but who else has? BTW, if you remove those 760 innings off of Waddell's career, what would his ERA+ be? I would think it would be a lot better than 135+.
   27. karlmagnus Posted: July 07, 2005 at 07:48 PM (#1455997)
Can't see how Bond beats McCormick, or indeed Welch, but to each his own.

Waddell's inability to focus shows up in his 193-143 W/L, which given he pitched mainly for Mack's good teams is my main problem with him. I don't havecareer WAT, but Joss's must be higher.
   28. sunnyday2 Posted: July 07, 2005 at 08:21 PM (#1456140)
John, check #21. I don't have Waddell for 2300 IP but for his prime, as I define it, I have him at 152, not the highest available (see Silver King, Ed Reulbach, and there's always Guy Hecker and people like that) but within a few points, and a career besides.
   29. Tiboreau Posted: July 07, 2005 at 08:22 PM (#1456142)
Knock off the 43; that's presumably 1937.

Dude, all you had to do was scroll up to check--it's his 1935.

If we should even think seriously about a short career, I still think it has to be Joss or Waddell.

Joss ERA+ 142 in 2,327 IP
Waddell 135 in 2,961
Dean 130 in 1,967


There are two problems I have with that: it doesn't consider era differences and season IP totals

For example, Joss season IP total is 338.2 (1907), Waddell's is 383 (1904), and Dean's is 325.1 (1935). And as you show, both Joss and Waddell had more IP during their peak than Dean. However, Joss was only among the IP leaders twice (1907, 5th; 1908, 2nd) and Waddell only four times (3rd, two 4th, and 10th), while Dean dominated season IP totals during his five peak years with 3 1st, a 2nd and a 3rd place finish.

And while Joss and Waddell have the advantage in ERA+, DERA has a different perspective:
car
Joss      3.34 3.36 3.41 3.57 3.68   3.65
Waddell   2.73 3.05 3.23 3.33 3.53   3.63
Dean      3.08 3.41 3.42 3.45 3.50   3.48 

Dean's best year easily tops Joss's, while they maintain similar DERA in their next four best years (Joss slightly better 2nd & 3rd, Dean decently better 4th & 5th). Waddell is pretty solidly better than both in top 5, however don't forget he was the least reliable of the 3. Of course, the more IP, the greater chance the numbers will fall toward the mean, which is reflected in the career DERA totals.

(FWIW, I used BPro's adjusted for season DERA, as opposed to adjusted for all-time. I don't know if there is any league quality issues affecting the former, but it's certainly less than the timeline on the latter.)
   30. John (You Can Call Me Grandma) Murphy Posted: July 07, 2005 at 08:40 PM (#1456201)
Marc:

I'm confused. None of my posts from #21 on referenced anything you said, except for #22 where I stated that I liked Waddell better. Did you mean karlmagnus instead?
   31. sunnyday2 Posted: July 07, 2005 at 08:44 PM (#1456210)
John in #26 you asked what Waddell's ERA+ would be in the same number of IP (2300) that Joss got. Your point was correct. It's not just what Pitcher B did in the extra IP, it's more important to know whether he was better in the bulk of his career, and I think Waddell was marginally.

BTW how many 182-142 pitchers have had 31 posts in their thread?
   32. Chris Cobb Posted: July 07, 2005 at 08:49 PM (#1456219)
How many of the 31 posts are actually about Leroy Matlock? :-)
   33. John (You Can Call Me Grandma) Murphy Posted: July 07, 2005 at 08:51 PM (#1456224)
It's not just what Pitcher B did in the extra IP, it's more important to know whether he was better in the bulk of his career, and I think Waddell was marginally.

Gotcha. I agree, Marc.

BTW how many 182-142 pitchers have had 31 posts in their thread?

:-)
   34. Dr. Chaleeko Posted: July 07, 2005 at 11:00 PM (#1456486)
I've got a super-weird system for pitchingthat I've described previously. It tries to adjust for the fluctuation in innings loads and the resultant effect on PWS by adjusting all pitcher's win shares toward an historical midpoint based on how the league's leading pitchers fared in WS. Crude? Yes. Better than anything else I've tried? Probably. (NgL hurlers are based on a combination of NL and AL leaders.)

Here's what the season-by-season WS for this clutch of hurlers we're discussing looks like once they get churned through my machine. Seasons move from best to worst, pitchers sorted by career length for ease of viewing.
NAME       1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19
Mendez    37 31 29 28 26 22 15 10  8  7  6  5  5  4  3  2  2  0  0
Walters   39 33 31 27 20 14 14 14 14 12 11 10  6  4  3  1  1  0  0
Ferrell   35 29 27 26 24 23 18 17 14  6  2  1  1  0  0  
Matlock   43 36 30 26 22 21 17 14 12  7  6  4  0  0  
Waddell   27 25 25 20 17 16 14 14 13 10  6  1  1 
H Smith   36 34 26 25 11 11  9  7  7  2  0  0  0
Dean      34 30 29 23 21 17  9  7  2  1  0
McCormick 33 23 21 20 20 20 18 11 11  4 
Joss      26 22 19 18 16 16 13 12  6

So my system sees Matlock as a total stud. Is it right? Don't know. Is anyone else's? Don't know that either. ; )
   35. sunnyday2 Posted: July 08, 2005 at 01:50 AM (#1457109)
Doc, sorry but can you describe your system again? I am intrigued.

And a question. Where did the Mendez WS come from? I don't remember seeing them before.

And whast is the range of adjustments? e.g. 0.5 to 2.0? or a tighter range? Do you know off hand what the total career adjustment is for these guys? I mean, I could calculate it easy enough--Dean is 180, right? And in your system 173? So his adjustment appears to be 0.96?

Matlock goes from 237 to 238 for an adjustment of .004?
   36. Dr. Chaleeko Posted: July 08, 2005 at 01:45 PM (#1457677)
Sunnyday,

Like I said, it's a little cocamamie, but here's what I do. I'd love some feedback about how to improve it.

-First, I figured the average league-leading and fifth-place WS totals for a typical MLB season from 1893-2004, using 5-year rolling averages to reduce the impact of unusually high or low seasons.

-Then I add the typical first- and fifth-place WS together to create the context of a "typical" historical league. Turns out that this number is 51 (30 WS for the leader, 21 for 5th place).

-Next, for each season in each league, I divide 51 (the historically average league) by the sum of the 5-year rolling average of the first- and fifth-place teams (centered on the season in question). This gives me the adjustment factor for that particular league season.

-Finally, I adjust a pitcher's WS by this adjustment factor.

I chose to use 1893-present because I wanted to get as much consistency as possible, so I'm not totally sure if this method works well for pre-1893 pitchers. And for 1893-1894, 1901-1902 (AL), and 2003 and 2004, I use three or four year rolling averages when no fifth year of data is available.

OK, to give a quick example, here's Joss's career adjusted this way
         #1  #5  #1  #5   SUM OF   ADJ
YEAR LG  WS  WS  AVG AVG  1 AND 5  FACTOR
-------------------------------------------
1901 AL  41  23  39  25    64       .80
1902 AL  38  27  43  26    69       .74
1903 AL  38  25  41  26    67       .76
1904 AL  53  29  40  27    67       .77
1905 AL  35  28  40  27    67       .77
1906 AL  36  24  42  26    68       .75
1907 AL  37  27  37  25    62       .82
1908 AL  47  23  38  24    62       .82
1909 AL  31  22  37  25    62       .83
1910 AL  37  26
1911 AL  31  26

NOTE: I've rounded everything for this table.



<pre>
        ADJ    JOSS'S   ADJ
YEAR    FACT    WS      WS 
---------------------------
1901    0.80    17      14
1902    0.74    20      15
1903    0.76    16      12
1904    0.77    25      19
1905    0.77    23      18
1906    0.75    28      21
1907    0.82    35      29
1908    0.82    20      16
1909    0.83     7       6
===========================
TOTAL          191     149

For pitchers like Joss and Waddell, this method really helps me see how the usage patterns of their era contribute to WS (and my own) perceptions of their value.

Someone pointed out that Joss didn't fare well amongst league leaders in a couple categories, and so this helps situate him relative to other pitchers in his era and others.

To get another point of view on this, here's what my system does with Roger Clemens:
        ADJ   RAWJA'S  ADJ
YEAR    FACT    WS     WS 
---------------------------
1984    1.10     8      9
1985    1.09     8      9
1986    1.11    29     32
1987    1.10    28     31
1988    1.09    22     24
1989    1.09    18     20
1990    1.11    28     31
1991    1.09    26     28
1992    1.07    26     28
1993    1.08    11     12
1994    1.09    16     17
1995    1.07    10     11
1996    1.09    20     22
1997    1.13    32     36
1998    1.13    25     28
1999    1.15    10     11
2000    1.18    16     19
2001    1.18    19     22
2002    1.17    11     13
2003    1.18    15     18
2004    1.11    20     22
=========================
TOTAL          398    443


In Clemens's case, each year gives him a boost because he's pitching in an era when workloads are lower. Those 443 WS are comparable to Warren Spahn, Greg Maddux, Pete Alexander, Cy Young, and Smokey Joe Williams, and not too far away from Walter Johnson. I personally believe that's appropriate company for a man who has pitched 21 seasons at such a consistenly high level, including 4 30+adjWS seasons and 9 20+adjWS seasons (i.e. 4 MVP-type seasons and 9 All-Star-type seasons).


One nice thing about this system is that I don't feel compelled to adjust for season length because that's inherent in the rolling averages: shorter seasons have typically meant higher workloads for pitchers.

Thanks in advance for any ideas for how I can improve this system!
   37. karlmagnus Posted: July 08, 2005 at 01:53 PM (#1457700)
This system adjusts for greater usage per annum in Joss's day but doesn't adjust for the shorter careers due to inadequate medical care -- in other words you rip him off on the swings without giving it back on the roundabouts. Clemens would NOT have pitched to 42 in 1900-20.
   38. John (You Can Call Me Grandma) Murphy Posted: July 08, 2005 at 02:10 PM (#1457739)
I think karlmagnus is correct in the sense that different eras reduced or expanded careers due to different training or medical capabilities, so any system needs to acknowledge that when comparing pitchers from different eras, IMO.

But would Clemens not have made it to age 42 had he pitched during the Deadball Era? I would say the odds would have been greatly reduced, but impossible? I wouldn't go that far.
   39. sunnyday2 Posted: July 08, 2005 at 02:15 PM (#1457752)
I agree with karl, though as more of a peak/prime voter it might not matter much to me. I think this system might work better for a peak/prime eval. than for a career eval. That is off the top.

I'd be curious to see more e.g.'s and especially some superficially comparable pitchers. I mean Joss and Clemens are not particularly comparable anyway. For Joss (191 career WS), how about the usual suspects Koufax (194) and Dean (188).

For Clemens (398). how about Grove and Spahn and Mathewson.

That will give us a sense of how pitchers from different eras fare in this system. If you provide their annual numbers, I can compute their peak/prime for myself.

Also can you provide your all-time list and also your adjWS for currently eligible pitchers?

I don't ask for much, right?

All in all, I think this is an honest attempt to deal with one of the toughest problems.Imean, it's hard enough dealing with the fact that players today are so much healthier than in the past and have an obvious advantage in career length, which is karl's legitimate concern. I've never found that to be an insurmountable difficulty, though, because to a large extent our sense of greatness is not absolute but relative--relative to one's peers from the same age cohort.

But with pitchers you've got that plus the rapidly declining IP. In a sense those could be counter-acting and thus making things easier. But it would be superficial to take it that way. Besides, for a peak/prime voter, it doesn't help at all.

Thanks for the details.
   40. karlmagnus Posted: July 08, 2005 at 02:30 PM (#1457785)
Not-Grandma is right, as almost always. Cy Young pitched effectively at 42-44. However, there are currently three 42 year olds (Clemens, Wells, Maddux) pitching well in the majors, which there certainly weren't in 1910. It's a question of probabilities; the medical odds against a long career were greater in 1910, as Joss proved, but they weren't insuperable. However, I don't think it's "superficial" to balance the extra innings off against the shorter careers, and I do think it's actively misleading to correct for one and not the other.

The reality is that Joss was universally considered one of the top pitchers of his day, like Dean and Koufax. His career was cut short, like both of theirs. If you give 25% death/terminal injury credit, that does not pull Dean into the HOM ranks (he had a VERY short career, and was overrated because of the success of his teams) but it does pull both Koufax (also overrated because of the success of his teams and the pitcher-friendly Dodger Stadium) and Joss (who didn't have either effect at the time, though the HOF 50 years later may have corrected inadequately for the dead ball era) into the HOM orbit.

I'd be very surprised if we don't elect Koufax, and if we elect Koufax I think it's very difficult not to elect Joss, who was extremely close to Koufax, and to me looks a bit better.
   41. karlmagnus Posted: July 08, 2005 at 02:41 PM (#1457810)
Add 450 innings of 110 ERA+ and a 30-25 record (roughly 1/4 of a potential remaining career) to all 3 and you get:

Joss: 2777IP, ERA+136, 190-122
Dean: 2417IP, ERA+125, 180-108
Koufax: 2774IP, ERA+126, 195-112

I think that puts Joss in, Dean clearly out and Koufax right on the edge.
   42. John (You Can Call Me Grandma) Murphy Posted: July 08, 2005 at 02:54 PM (#1457840)
However, I don't think it's "superficial" to balance the extra innings off against the shorter careers, and I do think it's actively misleading to correct for one and not the other.

When we get to the point that the current crop of pitchers are eligible (certainly the Seaver-Carlton-Ryan generation of pitchers), my system will take into account the longer careers. Not to do so would be unfair, IMO, just as it's unfair to compare the Radbourn of '84 with Pedro Martinez without adjustments being made.

As for Joss, he would most likely have been elected already, not if he had survived his illness, but if he had pitched more innings per season. The latter can't be corrected for him, though.
   43. Dr. Chaleeko Posted: July 08, 2005 at 03:04 PM (#1457873)
In response to one specific query, here's how Koufax fares in my system:
YEAR    ADJ    SKWS   ADJ WS
-----------------------------
1955    1.07     3     3
1956    1.11     1     1
1957    1.13     6     7
1958    1.14     7     8
1959    1.15     9    10
1960    1.12     9    10
1961    1.09    20    22
1962    1.06    15    16
1963    1.00    32    32
1964    0.95    24    23
1965    0.96    33    32
1966    0.94    35    33
============================
TOTAL          194   197


Koufax's best peak seasons in the mid-60s are adjusted downward because pitchers began to hurl more innings from the mid 60s through the early to mid 70s before the five-man rotation reduced workloads toward contemporary levels. On the other hand, a he is adjusted upwards toward the historical average in prior seasons where league workloads were a bit more historically normal.

That feels like a fair assessment of the condidtions in which he pitched.

To throw another short-career name out there, here's Lefty Gomez:
YEAR  ADJ FAC  LGWS   ADJ WS
-----------------------------
1930    0.91     1     1
1931    0.92    20    18
1932    0.92    17    16
1933    0.93    16    15
1934    0.98    31    30
1935    1.00    16    16
1936    1.00    11    11
1937    1.00    29    29
1938    1.01    19    19
1939    1.02    13    13
1940    1.02     0     0
1941    1.00    10    10
1942    0.97     2     2
1943    0.94     0     0
=============================
TOTAL          185   181


You've got my adjWS for Dean and several other short-career pitchers above. Admittedly, I'm on shaky ground with Jim McCormick because I'm using the same system for him that's really designed for post-1892 pitchers.

I'll run a chart with the longer-career pitchers in another post.
   44. Dr. Chaleeko Posted: July 08, 2005 at 03:57 PM (#1458074)
Sunnyday asked for Grove, Spahn, and Mathewson, and I'll add on the ones I mentioned in my Clemens comments.

GROVER
YEAR  ADJ FAC  LGWS   ADJWS
------------------------------
1925    0.99     9     9
1926    1.01    25    25
1927    1.01    24    24
1928    0.96    27    26
1929    0.91    28    26
1930    0.91    37    34
1931    0.92    42    39
1932    0.92    33    30
1933    0.93    23    21
1934    0.98     2     2
1935    1.00    29    29
1936    1.00    29    29
1937    1.00    27    27
1938    1.01    17    17
1939    1.02    23    23
1940    1.02    11    11
1941    1.00     5     5
==============================
TOTAL          391   378


SPAHN
YEAR  ADJ FAC  WSWS   ADJWS
------------------------------
1942    1.00     0     0

(MILITARY SERVICE 1943-1945)

1946    1.04     9     9
1947    1.09    32    35
1948    1.07    14    15
1949    1.05    24    25
1950    1.06    21    22
1951    1.03    26    27
1952    1.01    22    22
1953    1.02    31    31
1954    1.02    23    24
1955    1.07    19    20
1956    1.11    24    27
1957    1.13    22    25
1958    1.14    28    32
1959    1.15    23    27
1960    1.12    16    18
1961    1.09    25    27
1962    1.06    23    24
1963    1.00    22    22
1964    0.95     1     1
1965    0.96     7     7
===========================
TOTAL          412   440  

MATHEWSON
YEAR  ADJ FAC  CMWS  ADJWS
------------------------------
1900    0.81     0    0
1901    0.84    21   18
1902    0.84    22   19
1903    0.82    37   30
1904    0.82    34   28
1905    0.82    39   32
1906    0.83    20   17
1907    0.85    29   25
1908    0.88    39   34
1909    0.88    34   30
1910    0.88    30   26
1911    0.91    32   29
1912    0.90    31   28
1913    0.86    30   26
1914    0.84    19   16
1915    0.82     5    4
1916    0.85     4    3
=============================
TOTAL          426  364

ALEXANDER
<pre>
YEAR  ADJ FAC  PAWS  ADJWS
------------------------------
1911    0.91    34    31
1912    0.90    24    22
1913    0.86    27    23
1914    0.84    26    22
1915    0.82    43    35
1916    0.85    44    37
1917    0.87    40    35
1918    0.89     2     2
1919    0.94    26    24
1920    0.98    36    35
1921    0.94    22    21
1922    0.93    18    17
1923    0.97    27    26
1924    0.98    14    14
1925    0.98    20    20
1926    0.99    16    16
1927    1.02    28    29
1928    1.04    19    20
1929    1.03    10    10
1930    1.02     0     0
=============================
TOTALS         476   438


YOUNG
YEAR LG  ADJ FAC  CYWS  ADJWS
------------------------------
1890 NL    0.61     8     5
1891 NL    0.61    28    17
1892 NL    0.61    44    27
1893 NL    0.63    35    22
1894 NL    0.64    39    25
1895 NL    0.66    37    25
1896 NL    0.66    43    28
1897 NL    0.70    28    20
1898 NL    0.75    34    26
1899 NL    0.79    35    28
1900 NL    0.81    22    18
1901 AL    0.80    41    33
1902 AL    0.74    38    28
1903 AL    0.76    38    29
1904 AL    0.77    35    27
1905 AL    0.77    28    22
1906 AL    0.75    13    10
1907 AL    0.82    27    22
1908 AL    0.82    27    22
1909 AL    0.83    20    17
1910 AL    0.81    10     8
1911 AL    0.79     4     3
=============================
TOTAL             634   459

MADDUX
YEAR  ADJ FAC  GMWS  ADJWS
------------------------------
1986    1.16     1     1
1987    1.17     1     1
1988    1.26    20    25
1989    1.29    20    26
1990    1.25    15    19
1991    1.23    17    21
1992    1.17    27    32
1993    1.09    25    27
1994    1.06    26    28
1995    1.06    30    32
1996    1.05    23    24
1997    1.07    26    28
1998    1.10    25    28
1999    1.12    17    19
2000    1.11    24    27
2001    1.12    20    22
2002    1.12    19    21
2003    1.12    11    12
2004    1.11    12    13
============================
TOTAL          359   406

JOHNSON
YEAR  ADJ FAC  GMWS  ADJWS
------------------------------
1907    0.82     4     3
1908    0.82    20    16
1909    0.83    12    10
1910    0.81    36    29
1911    0.79    31    25
1912    0.78    47    36
1913    0.77    54    42
1914    0.77    38    29
1915    0.78    42    33
1916    0.83    36    30
1917    0.84    29    24
1918    0.86    38    33
1919    0.84    27    23
1920    0.87    10     9
1921    0.88    23    20
1922    0.90    21    19
1923    0.93    17    16
1924    0.98    29    28
1925    0.99    26    26
1926    1.01    15    15
1927    1.01     5     5
=============================
TOTALS         560   471
   45. John (You Can Call Me Grandma) Murphy Posted: July 08, 2005 at 04:08 PM (#1458117)
GROVER

YEAR ADJ FAC LGWS ADJWS
------------------------------
1925 0.99 9 9
1926 1.01 25 25
1927 1.01 24 24
1928 0.96 27 26
1929 0.91 28 26
1930 0.91 37 34
1931 0.92 42 39
1932 0.92 33 30
1933 0.93 23 21
1934 0.98 2 2
1935 1.00 29 29
1936 1.00 29 29
1937 1.00 27 27
1938 1.01 17 17
1939 1.02 23 23
1940 1.02 11 11
1941 1.00 5 5
==============================
TOTAL 391 378


Looks more like Super Grover to me! ;-)
   46. Dr. Chaleeko Posted: July 08, 2005 at 05:14 PM (#1458348)
John,

That's so funny, I didn't even realize I'd put the extra r on Grover! I was thinking of making a Super Grover reference there, but decided against...obviously my subconscious got the better of me.
   47. Dr. Chaleeko Posted: July 08, 2005 at 05:15 PM (#1458357)
John,

That's so funny, I didn't even realize I'd put the extra r on Grover! I was thinking of making a Super Grover reference there, but decided against...obviously my subconscious got the better of me.
   48. Chris Cobb Posted: July 08, 2005 at 05:33 PM (#1458427)
fwiw, I handle the adjustment of pitchers due to changing usage patterns in a way that is somewhat similar to Dr. Chaleeko's. Rather than adjusting from the top, however, I adjust from the middle. Through 1940, I've worked out the average number of innings pitched for a full-time starting pitcher and multiplied that by the ws/ip earned by an average pitcher in the win shares system to find the seasonal value of an average pitcher. I then subtract that from pitchers' seasonal win shares to measure their peak value.

One reason I like "low-peak," long career pitchers like Rixey and Ruffing is that, although their 20-22 win-share seasons don't look all that special when they are compared to position players (for whom average in the 1930s is 19), they are actually well above the win shares of an average pitcher, which are about 12.5 for the 1930s, as I see it.

Great pitchers generally have more value above average than great hitters, which compensates for them having generally less value above replacement (or above the WS zero point, as the case may be).

Overall, interesting discussion of short-career, high-peak pitchers. They've always been a tough group for us . . .
   49. John (You Can Call Me Grandma) Murphy Posted: July 08, 2005 at 05:57 PM (#1458504)
John,

That's so funny, I didn't even realize I'd put the extra r on Grover! I was thinking of making a Super Grover reference there, but decided against...obviously my subconscious got the better of me.


:-)
   50. sunnyday2 Posted: July 08, 2005 at 05:58 PM (#1458508)
Chris, I like value above average rather than above replacement but for a different reason. It's just a higher standard and it helps in differentiating players who are above average anyway.
   51. Dr. Chaleeko Posted: July 08, 2005 at 06:26 PM (#1458596)
Sorry for double post.

Here's the career leaders on my system's leaderboard. I have not done every pitcher, but I've got about 150 done.
-No NgLs included
-* means bulk of career before 1893, so I'm not as confident about the result
-No league-quality deductions
-Credits or debits for wartime noted in parentheses, but players who are not current candidates have not been credited or debited for any service time.

name             ws total
------------------------------
w johnson                 471
alexander (1918 @ 25))    462
young                     460
clemens                   453
spahn                     440
maddux                    406
seaver                    388
grove                     378
carlton                   374
p niekro                  373
mathewson                 364
roberts                   362
g perry                   355
blyleven                  354
ryan                      342
ruffing                   336
lyons (+w.c. 15, 10, 5)   335
wynn                      333
sutton                    332
nichols                   331
rixey (1918-19 @ 17 ea.)  330
r johnson                 329
glavine                   320
john                      319
palmer                    312
jenkins                   310
gibson                    303
hubbell                   299
ford                      297
feller                    295
faber (1918-19@ 15 each)  290
grimes (1919 @ 15)        280
kaat                      279
pierce                    277
plank                     273
bunning                   273
k brown                   270
smoltz                    269
reuschel                  268
drysdale                  267
caruthers*                266
d martinez                264
mussina                   264
pedro                     263
walters (10% deduc 43-45) 250
j morris                  253
newhouser                 253
stieb                     252
schilling                 252
marichal                  252
hershiser                 250
lemon                     247
mullane (+1885 @25 )      247
w cooper                  246
finley                    246
hoyt                      245
willis                    242
vance                     241
tiant                     238
3-f brown                 237
luque                     236
cone                      234
keefe*                    232
clarkson*                 230
newsome                   230
mays                      228
derringer                 228
galvin                    228
harder                    227
saberhagen                222
ferrell                   222
root                      222
pennock (+1918 @ 4)       222
s.s.jones                 221
adams                     220
appier                    220
mcginnity                 220
quinn                     220
coveleski                 219
bridges                   218
uhle                      216
j powell                  216
trout                     216
radbourn*                 215
warneke                   214
walsh                     212
blue                      208
rusie                     208
trucks                    207
cicotte                   202
haines                    201
mullin                    201
hunter                    200
welch*                    199
shocker                   198
bender                    198
griffith                  198
koufax                    197
whitehill                 192
rommell                   192
newcome (mlb only)        191
waddell                   188
orth                      186
dc white                  185
marquard                  184
gomez                     181
mccormick* (1884 @ 39)    180
brecheen (43-45 @10% off) 177
vaughn                    175
dean                      173
guidry                    166
th. lee (43-45 @10% off)  162
tudor                     161
chesbro                   160
phillipe                  160
ct davis (43-45 @10% off) 159
antonelli                 159
pfeffer                   158
s king*                   153
dineen                    153
rijo                      151
m cooper (43-45 @10% off) 148
joss                      148
parnell                   147
maglie (mlb only)         146
bond* (no UA)             137
lanier (mlb only)         132
   52. Mike Webber Posted: July 08, 2005 at 09:28 PM (#1459055)
Doc C, can I ask you a question about your method in #34?

From 1930 to 1945 no pitcher in the majors had 43 winshares in a season. 2 had 42, Trout in 1944 and Grove in 1931. In fact those were the only seasons where a pitcher had more than 38 win shares. Including those 2 seasons there were only 8 seasons where a pitcher had 35 or more winshares.

So suppose Matlock's big season was "only" 38 Win Shares, better than any season for Farrell or Hubble or Dean or Gomez or anyone except Grove once and Trout's WW2 season.

What would that do to him in your system?
   53. Dr. Chaleeko Posted: July 08, 2005 at 09:49 PM (#1459115)
Mike W.,

In Matlock's big year, he won every single case that went to court, got several dramatic acquittals, bought himself a new powder-blue suit, and outpointed Murder She Wrote.

Sorry, it's Friday, and I couldn't help myself.

Seriously, for Negro League pitchers, I use a combined AL/NL rolling average. For 1935, that adjustment is .97, meaning that Matlock's PWS would be 41.8.

If you used only the NL (since Chris translates into an NL environment), Matlock would come out at 40.4 because the NL's adjustment rate is .94.

Also, I keep forgeting whether Chris's WS estimates address batting. For a moment assume they don't. I looked at a bunch of pitchers and found that the typical pitcher-batter nets about 3% of his total WS value from batting. While Matlock was supposed to be a good hitter (per Riley), I'm assuming he was an average hitter in the absence of any further data. In which case adding 3 percent back to his new PWS would return him close to the original sans-hitting estimate.
   54. Dr. Chaleeko Posted: July 08, 2005 at 09:51 PM (#1459121)
Riley says that Matlock played at least one winter in Cuba (1936-1937 I think).

Does anyone have any data on him on that season? (Or on any others he might have played in Cuba?)
   55. OCF Posted: July 08, 2005 at 10:32 PM (#1459205)
Dr. Chaleeko in #51 has a systematic method for rating pitchers - adjusted win shares. I also have a system for rating pitchers, from which I can produce a single number, if I want: RA+ PythPat equivalent FWP. What does what we do say for the pitchers we have in common?

I'll mention who we don't have in common: I never did work up any 1880's pitchers this way, in part because I don't trust my system to be meaningful for them. And I'm just barely ahead of the election cycle- I have very few pitchers who aren't already eligible, although I did jump ahead to pick up Pedro Martinez, Randy Johnson, and Roger Clemens. There differences around the edges as to which marginal candidates are on the lists. The most significant ones on my list and not his are Leever and Shawkey. We both make a number of incidental adjustments, but for different reasons to different people.

In the end, there are 56 names in common. The average over these pitchers of the ratio of [adj. WS (Chaleeko)]/[adj. FWP (OCF)] is about 1.29, but it ranges all the way from .83 to 1.94. I would have predicted up front that my number would be friendlier to pre-1920 pitchers and his numbers friendlier to post-1920 pitchers, for one very big reason: he is adjusting innings pitched (or whatever other measure of playing time he uses) for the usage patterns of the times, and I'm not. (To a certain extent, I'm making that adjustment non-numerically in preparing ballots, as in "I know Willis's numbers look very good but I've got to depreciate his bulk some and I'm not voting for him.") That presumption is accurate.

The pitchers with the lowest WS/FWP ratio, in order (the ones that my system is friendliest to compared to Dr. Chaleeko's):

Young, Joss, Nichols, Rusie, Walsh, Plank, Waddell, W. Johnson, Phillippe, Pedro Martinez, McGinnity, Mathewson.

Note that 3 of the first 4 have not just pre-1920 time but also pre-1893 time. And what is Pedro doing on this list? Young is an outlier, at .83 to Joss's .90.

The pitchers with the highest WS ratio, in order (the ones that Chaleeko's system is friendliest to compared to mine):

Mullin, Grimes, S.S. Jones, Whitehill, Uhle, Ruffing, Lyons, Rixey, Derringer, Walters.

Hmm.. I voted for Lyons anyway, I am voting for Ruffing and Rixey, and Walters is just off my ballot. Interesting. Note that Dr. Chaleeko appears to be giving Grimes, Lyons, and Rixey missing season credit and I'm not.

Some pitchers with that WS/FWP ratio very near the 1.29 average:

Vance, Dean, Willis, Ferrell, Maddux, Warneke.

The issue this time is my adjustments. I adjusted Willis down for having a good defense behind him, and I adjusted Ferrell up for his own hitting.

Now that I think about it, what boosts Pedro in my system is that his RA+, and hence equivalent winning percentage, is so high that the nonlinearity of FWP becomes a major issue.
   56. OCF Posted: July 08, 2005 at 10:34 PM (#1459213)
I can make a chart of this. I will say that I think Dr. Chaleeko's numbers are probably better than mine, but for the sake of this chart I'll weight them evenly and rank the candidate by (WS+1.29*FWP)/2, where by "WS" and "FWP" I mean the adjusted numbers being talked about here.

Pitcher     Dr.C WS   OCF FWP  Avg.
Young        460       551     586
Johnson      471       473     541
Alexander    462       397     488
Clemens      453       365     462
Grove        378       349     415
Maddux       406       320     410
Mathewson    364       340     402
Nichols      331       363     400
Plank        273       289     324
Hubbell      299       255     315
Ruffing      336       205     301
Lyons        335       205     300
Rixey        330       202     296
Martinez     263       248     292
Faber        290       200     274
Walsh        212       225     251
Rusie        208       222     248
Vance        241       194     246
Willis       242       192     245
McGinnity    220       206     243
Brown        237       249     243
Coveleski    219       203     240
Hoyt         245       182     240
W. Cooper    246       180     239
Grimes       280       147     235
Adams        220       189     232
Walters      250       162     230
Bridges      218       181     226
Ferrell      222       176     225
Luque        236       164     224
Powell       216       179     231
Quinn        220       167     218
Cicotte      202       181     218
Waddell      188       191     217
Griffith     198       181     216
Warneke      214       165     214
Root         222       158     214
Shocker      198       173     211
Mays         228       150     211
Pennock      222       152     209
Harder       227       148     209
Derringer    228       141     205
Bender       198       160     203
Gomez        181       164     197
Jones        221       125     191
Rommell      192       147     191
Uhle         216       126     189
Haines       201       135     188
Vaughn       175       144     181
Phillippe    160       155     180
Joss         148       164     180
Dean         173       139     176
Chesbro      160       187     173
Orth         186       121     171
Marquard     184       121     170
Mullin       201       103     167
Whitehill    192       108     166
   57. OCF Posted: July 08, 2005 at 11:20 PM (#1459326)
The number in the second column for Brown should be 193, not 249. (249=193*1.29.) The number in the third column is correct. Brown ranks as low as he does for me becuase of a defensive support adjustment.
   58. Brent Posted: July 09, 2005 at 01:36 AM (#1459891)
Dr. Chaleeko asked:

Riley says that Matlock played at least one winter in Cuba (1936-1937 I think).

Does anyone have any data on him on that season? (Or on any others he might have played in Cuba?)


Yes, Matlock pitched one season in Cuba, 1938-39, splitting his time among two teams and going 6-8. With pennant-winning Santa Clara he was a teammate of Josh Gibson, Ray Brown, Sam Bankhead, Manuel García, and Lázaro Salazar. With third-place Cuba he was a teammate of Bill Holland, Ray Dandridge, and Bill Perkins.
Yr    Team         G CG  W  L Tm W Tm L Tm Pct Pennant  WAT % tm dec
38-39 Santa Clara  8  3  2  4   34   20   .630    *    -2.0      11%
38-39 Cuba        10  4  4  4   25   29   .463          0.3      15%
38-39 Total       18  7  6  8                          -1.7 
   59. Brent Posted: July 09, 2005 at 01:58 AM (#1460022)
Dr. Chaleeko wrote:

Like I said, it's a little cocamamie, but here's what I do. I'd love some feedback about how to improve it.

I make similar adjustments to pitching performance, but like Chris Cobb, I make my adjustments based on innings pitched rather than win shares. Because win shares mix performance with changes in conditions, I think use of win shares leaders brings some effects into your adjustment that you probably really don't want--particularly since you include both the #1 pitcher (who by definition is an outlier) and the #5. My innings-based adjustments are based on the #5 pitcher only. Of course taking a rolling (or moving) average helps remove the effects of outlying seasons, but I think they still may be affecting your adjustments, particularly during periods like 1912-16 or 1928-32 when Johnson and Grove were at their peaks. Using a moving average of innings pitched by the #5 pitcher, I get a smooth set of adjustments that I think is more a reflection of just changes in expected workloads, without mixing in the effects of unusual pitching performances.

Another difference - rather than taking the historical average league as the denominator, I take an average league from 1901-09. The reason is that during the aughts, pitchers and position players were about equally distributed among the league leaders in win shares--in the 1800s pitchers had an advantage, and after 1910 position players generally led the pack. So by using the aughts as my reference period, my adjusted win shares are more comparable to those of position players. I also make several other adjustments - actually I use an average of Warp1 and win shares, since each seems to have advantages (though for pitchers I'm leaning a little more toward Warp1).
   60. Brent Posted: July 09, 2005 at 01:59 AM (#1460027)
By the way, I posted the data on #5 pitcher in innings pitched on the pitchers thread several months ago.
   61. Brent Posted: July 09, 2005 at 02:29 PM (#1460523)
Mike Webber wrote:

From 1930 to 1945 no pitcher in the majors had 43 winshares in a season. 2 had 42, Trout in 1944 and Grove in 1931. In fact those were the only seasons where a pitcher had more than 38 win shares. Including those 2 seasons there were only 8 seasons where a pitcher had 35 or more winshares.

So suppose Matlock's big season was "only" 38 Win Shares, better than any season for Farrell or Hubble or Dean or Gomez or anyone except Grove once and Trout's WW2 season.


We observe this phenomenon in all of Chris’s pitcher MLEs – some seasons translate as huge – Ray Brown in 1938 is estimated to have had a DERA+ of 201 and 43 WS; Hilton Smith’s estimated DERA+ for 1947 is 248. These “great” seasons are sometimes mixed in with others that translate very poorly – Smith’s estimated DERA+ for 1946 is 98 and for 1948 is 60. While we expect some volatility in the season-to-season performance of major league pitchers, we don’t expect as much volatility as we see in these NeLg MLEs.

Chris obviously is aware of this issue—for example, he wrote the following on the Hilton Smith thread (# 18):

I wouldn’t place all that much weight on totals for any single season here. The number of recorded decisions for the Negro American League teams and their pitchers per season is smaller than anything we’ve dealt with since the formation of the Negro Leagues, so Smith’s short seasons (the most recorded decisions he ever had was 14 in 1938) would need to be regressed somewhat to really simulate a major-league season. He swings from very high values to very low values with very little middle ground. I think the career numbers, however, tell us basically what we need to know to place him, and a look at 5-7 years together, say, some chunk of 1936-1942 depending on how you like to measure, can give us a pretty good idea of the height of his peak.

In other words, we all know that for periods covering only a few decisions we can observe outliers, like Livan Hernandez winning 11 straight. We don’t want to extrapolate these outliers to make him look like Clements. So our preferred solution would be to do some kind of regression to damp down the outliers.

Although I don’t think I know enough to regress the pitcher MLEs—it apparently requires knowing the variance of the variable being regressed—perhaps I can achieve a similar effect by smoothing a pitcher’s career performances using a weighted moving average.

In the following example I will calculate a 5-year moving average of Matlock’s DERA, using a weight of 1/2 for Chris’s unadjusted MLE for the season and weights of 1/8 for the two preceding and two following seasons. (I selected these weights because they seem similar in effect to the examples of regression for batting statistics shown on the major league equivalencies thread.) As a final step, I then make a small proportional adjustment to ensure that the same career DERA.
Year   IP dera_2 dera_2+
1929   78   4.57     98
1930  163   4.41    102
1931   91   5.19     87
1932  206   4.59     98
1933  239   3.93    115
1934  264   3.47    130
1935  303   2.88    156
1936  281   3.55    127
1937  248   3.33    135
1938  102   5.03     89
1939  206   4.10    110
1940  253   4.33    104
1941  222   4.21    107
1942  160   4.55     99
     2816   3.96    114

So, what do you think? Does this present a more realistic picture of Matlock’s peaks and troughs? Can the method be improved?
   62. Brent Posted: July 09, 2005 at 02:39 PM (#1460524)
That should have said "Clemens" (not "Clements").
   63. Brent Posted: July 09, 2005 at 02:47 PM (#1460528)
I just realized the chart would be easier to read if it also showed Chris’s unadjusted MLEs. Columns 3 and 4 are Chris’s unadjusted dera and dera+, columns 5 and 6 are my smoothed estimates:
Year   IP dera dera+ dera_2 dera_2+
1929   78 4.50  100    4.57     98
1930  163 3.98  113    4.41    102
1931   91 6.43   70    5.19     87
1932  206 5.11   88    4.59     98
1933  239 3.72  121    3.93    115
1934  264 3.46  130    3.47    130
1935  303 2.65  170    2.88    156
1936  281 3.44  131    3.55    127
1937  248 2.65  170    3.33    135
1938  102 6.92   65    5.03     89
1939  206 4.02  112    4.10    110
1940  253 3.88  116    4.33    104
1941  222 4.41  102    4.21    107
1942  160 5.00   90    4.55     99
     2816 3.96  114    3.96    114
   64. Gadfly Posted: July 09, 2005 at 07:19 PM (#1460845)
18. Doc

LOL

Thanks for your confidence, but unfortunately I don't know everything. From Matlock's final season in 1942 until his death, I've got squat. Matlock, by the way, died in St. Paul, Minnesota on February 6, 1968.

Of course, one of the reasons that he's pretty much forgotten is that he died before the Negro Leagues became once again remembered.

It's possible that Matlock played semi-pro ball after 1942 up north and west of Chicago (an area famous for hiring great black pitchers to lead their teams from George Wilson to John Donaldson to Bismarck to the ManDak League), but his dying in St. Paul may also just be a coincidence.

Anyone here live in the Twin Cities and want to do some research and see if they can find his obit?

A Couple other notes on Matlock:

For Chris Cobb (Post#8)

1937- The Ciudad Dragones reportedly won the 1937 Santa Domingo Tournament with a record of 18-13 (.581). Interestingly the pitchers listed who played for CD (Paige 8-2, Fernandez 4-5, Matlock 4-1, Griffith 2-1, Garcia 0-2, Salazar 0-2) exactly equal a record of 18-13.

1937 Denver Post Tournament (Jul/Aug 1937)-

The Santa Domingo Stars, aka Negro League All-Stars, won their first six games straight to get to the finals. The team's rotation was Matlock, Bob Griffith, and Chet Brewer. Matlock won both of his starts, games 1 and 4 (in fact, Brewer, Matlock and Griffith threw 3 consecutive shutouts in Games 3 through 5).

Their seventh game was for the Championship (and 10 grand in prize money, which was a small fortune then). In addition, the winning pitcher was to get a thousand dollar bonus.

Satchel Paige, always money hungry, showed up and talked the tournament promoters into letting him pitch for the Stars and a shot at the bonus.

Because the tournament was double elimination, the Stars only had to beat their opponent once to win but their opponent (Duncan, OK) had to beat the Stars twice to win the tournament.

The Stars' players, pissed at Paige for trying to horn in and take the 1000 dollar bonus away from Matlock (who had been scheduled to pitch), reportedly deliberately lost the first game (by a score of 6-4).

Then they crushed Duncan, 11-2, with Matlock on the mound and Matlock got his grand (and finished 3-0 in tournament play).

The ridiculous ease with which the Stars tore through this Dever Post tournament led to a ban against independent black teams. Although there was possibly racism involved, it is also quite true that the black teams were simply way too good for the independent semi-pro white teams to compete against with any chance of success.

1939 Venezuela-

Matlock played for Vargas in the Venezuelan National Series. He finished with a record of 8-6 and an ERA of 2.07. He finished fourth in ERA behind Vidal Lopez (1.71), Bertrum Hunter (1.76), and Manuel "Cocaina" Garcia (1.88). However, Matlock pitched far more innings than either Lopez or Hunter (Garcia, though playing for a poor team, was undoubtably the best pitcher).

All games for these Championships were played in San Agustin Stadium in Caracas. It was a typical Latin American park of the time, i.e no outfield bleachers and the fences a long taxi ride away (reportedly all from 400 to 500 feet away). In other words, the Series was pitcher friendly.

Vargas, one of four teams competing, won the championship with a record of 14-8-1 (Only games played on Sunday in the National Stadium counted in the standings though the teams played during the week too). Vargas' three pitchers were Vidal Lopez (5-1), Matlock (8-6), and Bob Evans (1-1) who was also a quality Negro League pitcher.

Other Negro League and Cuban pitchers competing in the Series were: Bud Barbee, Oscar Estrada, Manolo Fortes, and Silvio Garcia.

Also on the Vargas team were Hall of Famer Ray Dandridge; Carlos Ascanio, the only Venezuelan to play in the Negro Leagues; Jesus Ramos, who would play in the Majors; and Alejandro Oms.

The batting leaders were:
.348 Carlos Blanco
.337 Carlos Ascanio
.321 Pedro Cepeda
.316 Fausto Fuenmayor
.310 Ray Dandridge
.307 Silvio Garcia
.304 Jesus Ramos

Hope that helps some.

Now I got to get around to what I signed on for, posting my 1955 ballot and some stuff on the Oms thread.
   65. Chris Cobb Posted: July 09, 2005 at 10:24 PM (#1461132)
gadfly,

Thank you for the additional information about Matlock! His 1937 season is quite an interesting story!

I'll work up new MLEs for Matlock for 1937 and 1939, including the new Dominican and Venezuelan data.

I'm hoping that you're going to give us some information about Oms in Venezuela, too! I've started work on MLEs for him this weekend, and it would be great to be able to fill in some of the gaps in his record in the 1930s with hard data as well as extrapolated seasons.
   66. Bleed the Freak Posted: January 20, 2011 at 03:21 AM (#3733047)
Chris Cobb, if you are available, I hope all is well, and that you can post finalized Leroy Matlock MLE's you mentioned in July of 2005, or maybe the MLE's have been posted to another thread and can be cross posted here.

Both Dr. Chaleeko and Cobb voted for Matlock in the 2008 election and worked on making MLE's for him.

Alex King, in the Hilton Smith thread, post 61, you mentioned that you ran Leroy Matlock and Leon Day through a WAR based MLE system?

Do you have either Matlock or Day to share, particularly Matlock, as he has sniffed the hall of merit in the past?

Thanks,
-Bleed
   67. Alex King Posted: January 21, 2011 at 08:29 AM (#3733874)
This isn't actually a separate MLE; I merely took Brent's DERA numbers and estimated Matlock's WAR from them, using the pythagorean formula to calculate a winning percentage and setting replacement level at .420, as is standard for bbref_WAR. So this is basically the same numbers as before, just recast in a different system.

Year WAR
1929 0.6
1930 1.6
1931 0.2
1932 1.6
1933 3.9
1934 5.9
1935 9.3
1936 5.9
1937 6.0
1938 0.3
1939 2.9
1940 2.8
1941 2.8
1942 1.3
Total 45.0

I posted my WAR estimates for Day in the 2011 ballot thread, post 199 (there's a small discussion of Day, posts 195-200). Based on that discussion, I think it's reasonable to give Day another 3 WAR or so for his work in the minors 1950-1952; here's a repost of his estimated WAR.

Year WAR
1935 4.3
1936 4.6
1937 8.3
1938 0.7
1939 3.3
1940 4.9
1941 5.8
1942 4.5
1943 1.1
1944 4.0
1945 4.0
1946 2.1
1947 0.6
1948 0.9
1949 0.3
1950 0.5
1951 2.6
1952 0.2
Total 52.8

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
rr
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 1.1005 seconds
51 querie(s) executed