How Reliable are Won-Lost Records
Part IV
For Better or Worse
This part of the article would be a lot more interesting if we hadn’t already done this kind of stuff 300 times, one hundred by me and 200 by my friends and cohorts.
This is completely off the subject, but my mother-in-law, who passed away in May at the age of 98, liked the word "cahoots", which I assume is derived from "cohorts." One time she saw in the newspaper that somebody we knew "was joining the law firm of" whoever it was. She reported this to the rest of us that he "was going into cahoots with." For some reason I always thought that was hilarious.
Anyway, the subject of a pitcher’s expected winning percentage, whether he has been lucky or unlucky, true value. . .all of this is well-plowed ground. OK, I have a new method to measure these things, and it is a very good method that looks at the facts at a very fine-grained level, but still. . . .we’ve seen this stuff before. But Betty Lou’s got a new pair of shoes, and I’ve got a new system to rank pitchers, and this is the only dance I do, so here goes.
1. Pitchers whose Won-Lost Records
are Better than their True Winning Percentage
Number one on the list—I am sure you would figure this out yourself if you thought about it long enough—but number one on the list is Whitey Ford.
Ford was a good pitcher. His career True Winning Percentage was .573. That’s a Hall of Fame kind of number. Steve Carlton and Greg Maddux are at .582, Carl Hubbell at .578 based on the data that I have. Juan Marichal is at .575, Gaylord Perry at .570, Hal Newhouser .569, Jim Bunning .566. It’s not like Whitey is the only Hall of Famer below .600 or something.
Ford was the number one starter on great teams, teams that won the pennant virtually every year until Whitey was almost finished. There are very few pitchers in history who had that. Other guys were on a great team a few years here and there; some guys were #1 starters on very good teams. But Whitey got a LOT of help from the guys he played with.
Of the top 10 pitchers on that list, seven are Yankees, the others being Kirk Rueter, Sal Maglie and Mark Mulder.
First
|
Last
|
True WPCT
|
WPct
|
Discrepancy
|
Whitey
|
Ford
|
.573
|
.695
|
.123
|
Kirk
|
Rueter
|
.463
|
.584
|
.121
|
Vic
|
Raschi
|
.543
|
.663
|
.120
|
Spud
|
Chandler
|
.592
|
.708
|
.116
|
Allie
|
Reynolds
|
.534
|
.650
|
.116
|
Sal
|
Maglie
|
.551
|
.660
|
.110
|
Johnny
|
Allen
|
.570
|
.675
|
.105
|
Herb
|
Pennock
|
.532
|
.631
|
.099
|
Mark
|
Mulder
|
.536
|
.632
|
.096
|
Lefty
|
Gomez
|
.588
|
.681
|
.093
|
Sometimes the data isn’t complete. I don’t have ALL of Herb Pennock’s career starts in my collected-by-Retrosheet data; this is just based on the data that I have.
After you get out of the top 10, the Yankees no longer dominate to the same extent, but the same general principle applies. The guys who have winning percentages better than their true winning percentages are pitchers who pitched almost all of their careers for good teams—Dave McNally, Freddie Fitzsimmons, Lew Burdette, Andy Pettitte, Jack Billingham, Ed Lopat, Don Newcombe, Scott McGregor, David Wells, Ron Guidry, Bob Welch, Dave Stewart, Rick Porcello. There are 95 pitchers in my data who (a) won at least 100 games, and (b) had winning percentages at least 40 points better than their True Winning Percentages. Those 95 pitchers could be sorted into three classes:
1. Great pitchers made to look even greater,
2. Good pitchers made to look great or borderline great, and
3. Average and below-average pitchers made to look like good pitchers by deceptive won-lost records.
Great pitchers made to look even greater by their won-lost discrepancy include Lefty Grove, Roy Halladay, Mike Mussina, Ron Guidry, Lefty Gomez, Jim Palmer, Juan Marichal, Urban Shocker, Whitey Ford, Bob Feller and CC Sabathia.
Good pitchers made to look very good or great include Dwight Gooden, Warren Spahn, Jimmy Key, John Candelaria, Bob Lemon, Tim Hudson, Andy Pettitte, Sal Maglie, Catfish Hunter, Preacher Roe, Mike Cuellar, Orel Hershiser, Tom Glavine, David Wells, Herb Pennock, Bartolo Colon, Bob Welch, Dave McNally, Dave Stewart and Burleigh Grimes. (Obviously Gooden was legitimately great for the first couple of seasons, and Spahn was in some seasons.)
Sub-.500 pitchers (by True Winning Percentage) made to look significantly better than they actually were by their won-lost records include Scott McGregor, Bob Forsch, Jerry Reuss, Bob Buhl, Mike Torrez, Lew Burdette and Jack Billingham.
2. True Winning Percentage and the Hall of Fame
Before I started this study, one question I asked myself was, "Is it true that MOST Hall of Fame pitchers are made to look better than they really were by somewhat deceptive won-lost records?
The answer to that question is "Yes, if by ‘better’ you accept 10 or 15 points better as better. More than half of Hall of Fame pitchers got at least a little bit of a ‘situational boost" in their won-lost records. Dese is da facts:
In my data there are 46 Hall of Fame starting pitchers with at least 100 wins within my data.
Of those 46, 28 had higher Winning Percentages than True Winning Percentages. 18 had lower Winning Percentages than True Winning Percentages. That’s not a big edge, but 60% had higher record book Winning Percentages than True Winning Percentages.
Stated another way, the edge looks much bigger. Fourteen Hall of Fame pitchers had Winning Percentages 40 points (.040) better than their True Winning Percentages. Only three had Winning Percentages 40 points worse than their True Winning Percentages. 14 to 3; that looks like a big thing.
To balance the scales, you would need to move the center by 18 points. If we subtract .018 from the advantage, then these charts balance. Hard to state this in a coherent sentence, but there are as many Hall of Fame pitchers who have record book Winning Percentages more than 18 points better than their True Winning Percentages as those who do not, and, at the extremes, there are eight Hall of Fame pitchers who have record book Winning Percentages 58 points better than their True Winning Percentages, and nine who have record book Winning Percentages 22 points worse than their True Winning Percentage. +58 is 18 points better than .040, and -22 is 18 points better than negative .040. So we can say that Hall of Fame pitchers have an 18-point advantage in record book Winning Percentage versus True Winning Percentage.
The ten Hall of Fame pitchers who were meaningfully BETTER than their won-lost records were Phil Niekro (-.021), Jim Bunning (-.022), Gaylord Perry (-.023), Ferguson Jenkins (-.023), Robin Roberts (-.025), Bob Gibson (-.029), Dennis Eckersley (-.031), Dazzy Vance (-.034), Red Faber (-.044), Bert Blyleven (-.051) and Nolan Ryan (-.078).
There is no Hall of Fame pitcher whose True Winning Percentage was less than .500. The lowest ever was a .519 wining percentage by Jesse Haines, followed by .520 (Burleigh Grimes) and .532 (Herb Pennock). While it is my opinion that none of those three men should be in the Hall of Fame, they were all from the earliest part of the study, where there is a lot of missing data.
3. Pitchers Whose True Winning Percentages
Were Much Better than their Won-Lost Records
Among pitchers who had 100 starts within my data, these five pitchers had the worst won-lost records compared to their True Winning Percentages:
First
|
Last
|
True WPCT
|
Wins
|
Losses
|
WPct
|
Discrepancy
|
Jose
|
DeLeon
|
.551
|
73
|
110
|
.399
|
-.152
|
George
|
Caster
|
.490
|
37
|
69
|
.349
|
-.141
|
Jay
|
Hook
|
.432
|
26
|
62
|
.295
|
-.137
|
Ryan
|
Franklin
|
.492
|
27
|
49
|
.355
|
-.137
|
Skip
|
Lockwood
|
.457
|
25
|
53
|
.321
|
-.136
|
DeLeon you probably remember; he was the pitcher who went 2-19 for the Pittsburgh Pirates in 1985, supported by a robust 2.32 runs per start. George Caster pitched for the St. Louis Browns in the 1930s, and Jay Hook pitched for the stuff of legends, the 1962 Mets. Ryan Franklin went 4-16 for the Seattle Mariners in 2004, a team which lost 99 games just three years after winning 116. Skip Lockwood as I recall was a Kansas City A's bonus baby as an infielder who switched to pitching. He had some difficult years with the expansion Pilots/Brewers, later was a decent relief pitcher.
These are the bottom five if we increase the minimum starts to 150:
First
|
Last
|
True WPCT
|
Wins
|
Losses
|
WPct
|
Discrepancy
|
Jose
|
DeLeon
|
.551
|
73
|
110
|
.399
|
-.152
|
Matt
|
Young
|
.476
|
42
|
79
|
.347
|
-.129
|
Jim
|
Beattie
|
.504
|
52
|
86
|
.377
|
-.128
|
Al
|
Hollingsworth
|
.478
|
44
|
79
|
.358
|
-.120
|
Jeff
|
Samardzija
|
.526
|
59
|
86
|
.407
|
-.120
|
And here are the bottom five if we increase the minimum starts to 200:
First
|
Last
|
True WPCT
|
Wins
|
Losses
|
WPct
|
Discrepancy
|
Jose
|
DeLeon
|
.551
|
73
|
110
|
.399
|
-.152
|
Jeff
|
Samardzija
|
.526
|
59
|
86
|
.407
|
-.120
|
Si
|
Johnson
|
.467
|
72
|
133
|
.351
|
-.116
|
Milt
|
Gaston
|
.485
|
68
|
115
|
.372
|
-.114
|
Jason
|
Johnson
|
.465
|
55
|
96
|
.364
|
-.100
|
4. Pitchers Who Might Possibly Be in the Hall of Fame
With Better Fortune
Without saying that any of these pitchers SHOULD be in the Hall of Fame, here are eleven pitchers who
(a) Had long careers,
(b) Had True Winning Percentages over .500, but
(c) Had Won-Lost records significantly worse than their True Winning Percentages would predict.
1. Bob Friend, Pirates of the 1950s and early 1960s
2. Charlie Hough, Texas Rangers knuckleball pitcher of the 1980s
3. Frank Tanana, Angels’ flamethrower of the late 1970s who transitioned into a crafty lefty and pitched for years
4. Jerry Koosman, Tom Seaver’s running mate with the Mets
5. Mark Langston
6. Kevin Appier
7. Chuck Finley
8. Mickey Lolich
9. Dave Stieb
10. Camilo Pascual
11. Rick Reuschel
5. The Best Seasons Ever, by True Winning Percentage
Among pitchers with 30 or more starts, the highest True Winning Percentages ever for a season (within my data) are:
1. Randy Johnson, 1999 .785
2. Dazzy Vance, 1924 .770
3. Pedro Martinez, 1997 .768
4. Randy Johnson, 2001 .768
5. Randy Johnson, 1995 .761
6. Lefty Grove, 1931 .753
7. Roger Clemens, 1997 .749
8. Randy Johnson, 2000 .736
9. Randy Johnson, 2002 .735
10. Bob Gibson, 1968 .731
11. Dwight Gooden, 1985 .731
12. Roger Clemens, 1990 .731
There is a reason most of these seasons are concentrated in the steroid era, but I don’t know if this is the right place to try to explain it. When more runs are scored, that acts as a magnifying glass on the differences between pitchers. It pushes the best pitchers further ahead of the average pitchers. It is not a distortion in the sense of being a measurement flaw; what is being measured is real, and has real consequences. It is more like an unfair reality.
Randy Johnson has the highest True Winning Percentage ever with 30 or more starts; however, Pedro Martinez beat Randy’s number twice with 29 starts. The highest True Winning Percentages ever with 20 or more starts are:
1. Pedro Martinez, 2000 .852 (29 starts)
2. Pedro Martinez, 2001 .818 (29 starts)
3. Greg Maddux, 1995 .771 (28 starts)
6. The Luckiest Seasons In My Data
I hate to use the term "Luck" here, because that is not EXACTLY what we are measuring. More accurately stated, we are measuring the difference between True Winning Percentage and Won-Lost record, but that’s too long for a headline.
The most notable seasons in my data in this regard were by:
1. Whitey Ford, 1961
Was:   25-4
Should have been: 17-12
Despite a should-have-been won-lost record of "just" 17-12, this method will show Whitey to have been the most deserving Cy Young candidate of 1961. It’s a very odd year; I think the second-best candidate for that season is Camilo Pascual. Baseball Reference shows the league leader in pitcher’s WAR for that season to be Jack Kralick, and does not show Whitey Ford among the top 10. I think we argued about that in 2017, probably.
2. Catfish Hunter, 1973
Was: 21-5
Should have been: 14-12
3. Mike Nagy, 1969
Was: 14-2
Should have been: 7-7
4. Bob Welch, 1990
Was:   27-6
Should have been: 18-15
5. Preacher Roe, 1951
Was:   21-3
Should have been 14-12
Preacher Roe in 1951 was 22-3, but one win was in relief.
6. Phil Niekro, 1985
Was: 17-4
Should have been: 11-10
7. Nathan Eovaldi, 2015
Was: 14-3
Should have been: 8-9
8. Braden Looper, 2009
Was:   14-7
Should have been: 8-13
Braden Looper was 14-7 in 2009, was a free agent after that season and did not receive a significant offer. That was one of the moments at which I realized that sabermetrics had fully penetrated major league front offices.
9. Russ Meyer, 1953
Was: 15-4
Should have been: 10-9
Meyer in 1953 was 15-5, but one of the losses was in relief. The 10-9 record actually calculates as 9.505 to 9.495.
10. Gaylord Perry, 1978
Was: 21-6
Should have been: 15-12
Three of these pitchers won the Cy Young Award.
7. The Un-luckiest Seasons In My Data
Apologizing again for the fact that this kind of stuff has been written up so many times before; I’m hoping it is still a little bit interesting.
1. Jerry Koosman, 1978
Was: 3-15
Should have been: 9-9
2. Nolan Ryan, 1987
Was: 8-16
Should have been: 15-9
3. Jose DeLeon, 1985
Was: 2-18
Should have been: 10-10
4. Vidal Nuno, 2014
Was: 2-12
Should have been: 7-7
5. Matt Cain, 2007
Was: 7-16
Should have been 14- 9
6. Corey Kluber
Was:   9-15
Should have been: 15-9
7. Paul Derringer, 1933
Was:   7-26
Should have been: 16-17
8. Shelby Miller, 2015
Was:   6-17
Should have been: 12-11
9. Sherry Koosman, 1977
Was:   8-20
Should have been: 16-12
10. Ross Ohlendorf, 2010
Was: 1-11
Should have been: 6- 6