Part II of DIPS: The Best, the Lucky, and the Not-So-Lucky

Thanks for all of the responses to Part I. Sorry for the delay. I was ready to post this in the morning until I found an error in my data, so I had to redo it all. Now I am back with a second round of numbers. The main critique of my first round was that I did not control for different seasons. I knew this would be a problem, but I really wanted to see a list that had two qualities. First, it must express ERA like we read everyday ERA. I really wanted to avoid an AERA or ERA+ number, just because it would be hard to translate, especially with DIPS-corrected numbers. Second, I wanted to avoid distorting DIPS ERA numbers due to differing values of runs across seasons. This is why I wanted to run one regression to generate the coefficients. However, the main thing I am missing is some sort of component to capture the quality of hitters in a given season. I think I found a good way to do what I want. And the results are different from what I posted earlier.

Here is the solution. I developed deflators for ERA, walks, strikeouts, and home runs based on the league averages for that year. I was then able to convert pitcher statistics into a single ERA as if all pitchers were pitching in 2003. For example, in 1968 the average ERA for pitchers in this sample was 3.03, compared to 2003′s average ERA of 4.51. The ratio of 2003 to 1968 ERAs is about 1.48. I can take this number and multiply it by all pitcher ERAs of 1968 to make them comparable to 2003 stats. This is similar to the method used for calculating price deflators (such as the CPI) to adjust money values for inflation. I use deflator indices to convert all past pitcher stats (walks, Ks, HRs, and ERA) to values comparable to the 2003 averages. Now, I’m ready to estimate the regression using the fielding independent components per 9 innings.

pERA = 2.80 + 0.46*BB – 0.17*K + 1.14*HR

Comparing these values to the previous estimates, I would say the adjusting the components for each year is an important step. I had figured it might just wash out, but that does not seem to be so. First, I present the top-25 pERA seasons from 1921-2003.

Rank First Last Year ERA Defl_ERA pERA
1 Pedro Martinez 1999 2.07 1.91 1.66
2 Dazzy Vance 1925 3.53 3.76 1.85
3 Pedro Martinez 2000 1.74 1.58 2.14
4 Pedro Martinez 2001 2.39 2.37 2.14
5 Lefty Grove 1930 2.54 2.48 2.18
6 Dazzy Vance 1924 2.16 2.43 2.19
7 Greg Maddux 1995 1.63 1.64 2.37
8 Harry Brecheen 1948 2.24 2.52 2.43
9 Cy Blanton 1935 2.58 2.84 2.43
10 Greg Maddux 1997 2.20 2.23 2.45
11 Kevin Brown 1998 2.38 2.34 2.46
12 Greg Maddux 1994 1.56 1.56 2.46
13 Dwight Gooden 1984 2.60 2.95 2.49
14 Carl Hubbell 1933 1.66 2.00 2.50
15 J.R. Richard 1980 1.90 2.15 2.53
16 Pedro Martinez 2002 2.26 2.31 2.55
17 Randy Johnson 1998 1.28 1.26 2.56
18 Roger Clemens 1997 2.05 2.08 2.57
19 Randy Johnson 1995 2.48 2.49 2.58
20 Pedro Martinez 2003 2.22 2.22 2.59
21 Lefty Grove 1928 2.58 2.94 2.61
22 Bill Gullickson 1981 2.80 3.42 2.62
23 Babe Adams 1922 3.57 3.99 2.62
24 Randy Johnson 2001 2.49 2.46 2.63
25 Dazzy Vance 1928 2.09 2.38 2.64

Though pitchers of recent history still dominate this list, it is not as extreme as the previous list. Pedro is still the king of pERA with 5 top-25 seasons with 3 in the top-5. Though he must be in decline since last season was only the 20th best pERA season ;-). Greg Maddux, Randy Johnson, and Dazzy Vance have 3 seasons a piece.

Again, for comparison here are the top-25 Deflator-Adjusted ERAs of the sample.

Rank First Last Year ERA Defl_ERA pERA
1 Randy Johnson 1998 1.28 1.26 2.56
2 Greg Maddux 1994 1.56 1.56 2.46
3 Doyle Alexander 1987 1.53 1.58 3.52
4 Pedro Martinez 2000 1.74 1.58 2.14
5 Greg Maddux 1995 1.63 1.64 2.37
6 Bob Gibson 1968 1.12 1.66 3.04
7 Dwight Gooden 1985 1.53 1.73 2.74
8 Red Munger 1944 1.34 1.77 3.48
9 Kevin Brown 1996 1.89 1.81 2.69
10 Steve Rogers 1973 1.54 1.81 3.90
11 Pedro Martinez 1999 2.07 1.91 1.66
12 Pedro Martinez 1997 1.90 1.93 2.76
13 George Witt 1958 1.61 1.93 3.93
14 Carl Hubbell 1933 1.66 2.00 2.50
15 Joel Pineiro 2001 2.03 2.01 3.18
16 Dean Chance 1964 1.65 2.05 3.38
17 Nolan Ryan 1981 1.69 2.07 3.15
18 Jim Hearn 1950 1.94 2.07 3.72
19 Ron Guidry 1978 1.74 2.08 2.85
20 Roger Clemens 1997 2.05 2.08 2.57
21 Cal Eldred 1992 1.79 2.10 3.33
22 Sandy Koufax 1966 1.73 2.13 3.09
23 J.R. Richard 1980 1.90 2.15 2.53
24 Sandy Koufax 1964 1.74 2.17 3.08
25 Tiny Bonham 1940 1.90 2.17 3.13

Now to the top-25 luckiest pitchers, as denoted by the pRatio.

Rank First Last Year Defl_ERA pERA pRatio
1 Doyle Alexander 1987 1.58 3.52 2.23
2 Steve Rogers 1973 1.81 3.90 2.15
3 Randy Johnson 1998 1.26 2.56 2.03
4 George Witt 1958 1.93 3.93 2.03
5 Leo Dickerman 1924 2.71 5.49 2.03
6 Red Munger 1944 1.77 3.48 1.96
7 Bob Gibson 1968 1.66 3.04 1.83
8 Jim Hearn 1950 2.07 3.72 1.80
9 Andy Benes 2002 2.84 5.02 1.77
10 Brian Bohanon 1998 2.36 4.18 1.77
11 Roger Craig 1959 2.38 4.17 1.75
12 Al Benton 1949 2.38 4.13 1.73
13 Freddie Fitzsimmons 1941 2.44 4.20 1.72
14 Steve Sundra 1939 3.00 5.14 1.72
15 Jim Konstanty 1944 3.70 6.31 1.70
16 Ken Holtzman 1967 3.38 5.70 1.69
17 Floyd Youmans 1985 2.77 4.66 1.68
18 Woody Williams 2001 2.26 3.79 1.68
19 Rube Melton 1946 2.70 4.50 1.66
20 Joey Jay 1958 2.57 4.27 1.66
21 Greg Harris 1991 2.48 4.12 1.66
22 John Candelaria 1977 2.56 4.25 1.66
23 Dean Chance 1964 2.05 3.38 1.64
24 Hal Dues 1978 2.82 4.63 1.64
25 Pete Smith 1992 2.41 3.95 1.64

It is interesting that this list is not all that different from the first list. Finally, here is the unlucky top-25.

Rank First Last Year Defl_ERA pERA pRatio
1 Dazzy Vance 1925 3.76 1.85 0.49
2 Chris Zachary 1971 6.72 3.57 0.53
3 Ramiro Mendoza 1996 6.50 3.53 0.54
4 Seth Morehead 1958 7.03 3.97 0.57
5 Lefty Grove 1934 7.02 3.97 0.57
6 Camilo Pascual 1955 6.95 3.99 0.57
7 Bobo Newsom 1942 6.42 3.70 0.58
8 Herman Besse 1942 8.02 4.63 0.58
9 Jon Lieber 1995 6.35 3.71 0.58
10 Johnny Babich 1935 7.32 4.31 0.59
11 Slim Harriss 1926 5.13 3.06 0.60
12 Benny Frey 1935 7.53 4.51 0.60
13 Micah Bowie 1999 9.20 5.52 0.60
14 Ken Holloway 1926 5.89 3.54 0.60
15 Dutch Leonard 1949 4.66 2.80 0.60
16 Ted Blankenship 1924 5.63 3.39 0.60
17 Bob Muncrief 1946 6.78 4.10 0.60
18 Milt Pappas 1968 8.32 5.05 0.61
19 Rick Wise 1968 6.76 4.10 0.61
20 Mike Parrott 1981 6.21 3.83 0.62
21 Roy Halladay 2000 9.64 5.94 0.62
22 George Murray 1926 6.49 4.00 0.62
23 Brad Havens 1983 9.36 5.78 0.62
24 Milt Gaston 1928 6.28 3.88 0.62
25 Hal Gregg 1947 7.03 4.36 0.62

Again, I see some familiar faces. And Dazzy Vance’s second best pERA of all time is also the most unlucky season. I guess that is not surprising.

Anyway, this has been a fun exercise, and I have appreciated all of the suggestions and comments that I have received. I am happy to listen to more.

Comments are closed