Ranking the Best Pitching Seasons Ever
What was the best pitching season ever? We could look at the lowest ERA, but in some eras, ERAs were naturally low, like in the deadball era before 1920 when there were many seasons with ERAs under 2.00. We could look at the seasons with the most wins or highest winning percentages, but those are determined not just by the quality of the pitching but also by the run support a pitcher gets.
We could get around this problem by comparing a pitcher's ERA to the league average. Two pitchers might be judged equal if their ERAs are both 25% below the league average. Pitcher A might have an ERA of 3.00 with the league average being 4.00 while pitcher B has an ERA of 2.25 in a league with an average ERA of 3.00.
But a problem that often emerges in this approach is that the best seasons often come when runs per game were very high or very low. In extremely high scoring seasons, it may be easier to go far below the league average since the average is so high. Extremely low scoring seasons might increase the chances of any pitcher having a very low ERA.
One possible solution is to compare the best pitcher in the league to the other good pitchers in the league. If it is easy for one pitcher to go far below the league average, it should be easy for a few others. By comparing the league leader in ERA (or any measure of pitching quality) to the other very good pitchers, the problem mentioned above might be lessened.
ERA can also be affected by the home ballpark of the pitcher. So in addition to comparing the best pitchers to other good pitchers, their performance should be adjusted for park effects. Pitchers in high scoring parks will have their runs allowed adjusted downward and vice versa.
One measure that allows for this is called RSAA. It comes from the Lee Sinins Sabermetric Encyclopedia, a commercial database that can be purchased by any baseball fan. Here is the definition: "Runs saved against average. It's the amount of runs that a pitcher saved vs. what an average pitcher would have allowed."
I looked at how the RSAA of league leaders since 1900 compared to the average RSAA of the pitchers who finished 2-10 (hence, the idea of comparing top pitchers to other good pitchers). For example, Walter Johnson had 75 RSAA, meaning he allowed 75 runs less than the average pitcher. The next 9 best pitchers in 1913 averaged 25.56. So Johnson was 49.44 better.
But having, say, 30 more RSAA than the next best nine pitchers might mean more in a low scoring year than a high scoring year. In a low scoring year it will take a lower number of runs to add one over the course of a season. But how many? I used the formula which says it takes 10 times the square root of the number of runs scored per inning by both teams (found in Total Baseball, 5e). If each team scores .5 runs per inning, the total is one. The square root is 1 and 10 times that is 10, so it would take 10 additional runs over the course of a season to win one more game. The Lee Sinins Sabermetric Encyclopedia can call up the top 10 each season in RSAA.
Who were the top pitchers according to this method? The top 10 in the AL are listed below:
Pitcher Year RSAA RSAA Diff* R/W Extra Wins
Walter Johnson 1913 75 25.56 49.44 9.39 5.26
Pedro Martinez 2000 77 24.56 52.44 10.89 4.81
Lefty Grove 1931 75 26.67 48.33 10.73 4.50
Lefty Grove 1932 75 26.70 48.30 10.81 4.47
Walter Johnson 1912 74 29.56 44.44 10.02 4.44
Walter Johnson 1918 56 16.92 39.08 8.97 4.36
Cy Young 1901 72 25.11 46.89 11.08 4.23
Pedro Martinez 1999 71 28.11 42.89 10.87 3.95
Lefty Grove 1926 62 23.33 38.67 10.34 3.74
Hal Newhouser 1945 59 24.44 34.56 9.34 3.70
Lefty Grove 1936 70 29.78 40.22 11.32 3.55
In 1913 it took 9.39 runs to win one more game. Since 49.44/9.39 = 5.26, Johnson added 5.26 more wins than the average of the next best nine pitchers in the league (I have eleven pitchers here--Hal Newhouser's season was a war year, when many good pitchers may have been in the military).
For the NL, the top 10 were:
Pitcher Year RSAA RSAA Diff* R/W Extra Wins
Grover Alexander 1915 69 15.80 53.20 9.00 5.91
Dolf Luque 1923 66 24.00 42.00 10.40 4.04
Bob Gibson 1968 56 21.78 34.22 8.72 3.93
Greg Maddux 1995 64 24.80 39.20 10.17 3.86
Christy Mathewson 1905 61 25.33 35.67 9.62 3.71
Dwight Gooden 1985 58 22.75 35.25 9.51 3.71
Dazzy Vance 1930 64 22.36 41.64 11.33 3.68
Carl Hubbell 1933 52 18.10 33.90 9.44 3.59
Dazzy Vance 1924 56 20.78 35.22 10.07 3.50
Bucky Walters 1939 58 23.11 34.89 9.99 3.49
One problem can be seen--if you know some baseball history--is that we still see the best pitching performances coming from what are generally fairly high or fairly low scoring years. I really don't know the solution. Comparing players using standard deviations instead of simple averages might be better. I ran this study and ranked pitchers in ERA based on how many standard deviations below the average of all qualifying pitchers they were. Pedro Martinez in 2000 was the best, being 3.79 SDs below average.
Looking at ERA has an advantage over RSAA, since it only includes earned runs whereas RSAA includes both earned and unearned runs. Unearned runs may not be the fault of the pitcher. I also looked at the best ERAs relative to the 2-5 pitchers each year.
But both RSAA and ERA are, in part, determined by the quality of the fielding behind the pitcher. In his Win Shares methodology, Bill James attempted to rate pitchers solely on their contribution to winning, independent of the fielders. Using the electronic Win Shares database, I found the best seasons by taking the league leader and seeing how many Win Shares he had as percentage of the pitchers who finished 2-5.
Pitcher Year WS Ratio
Alexander 1915 43.32 1.87
Maddux 1994 25.96 1.86
W. Johnson 1913 50.28 1.81
Alexander 1917 39.11 1.77
Grove 1931 41.83 1.74
Walsh 1908 46.62 1.74
Alexander 1916 41.95 1.72
Maddux 1995 29.87 1.72
Vance 1924 35.57 1.68
W. Johnson 1915 39.34 1.67
Carlton 1972 40.38 1.67
Martinez 2000 28.86 1.64
Chesbro 1904 51.80 1.63
Luque 1923 38.97 1.63
Walters 1939 34.50 1.59
Gibson 1968 36.36 1.52
Clemens 1997 31.66 1.52
Martinez 1999 26.89 1.52
Perry 1972 39.04 1.50
The same pitchers seem to be near the top on all of these lists (including the lists at the links given above). There could be a problem that the quality of pitchers they are being compared to is relatively low (which might explain why they all do so well in simple comparisons to the league average as well). Maybe some years just did not have many good pitchers. I don't know how that could be determined. One suspicion I have about some of Pedro Martinez's good years is that there were no other very good pitchers. Roger Clemens was in decline. Randy Johnson was traded to the NL. But maybe the same could be said about Bob Gibson in 1968. Sandy Koufax was gone. Tom Seaver had yet to hit his prime. Maybe it could be said about any of these pitchers.
Some pitchers who stand out even among this crowd are Walter Johnson, Grover Cleveland Alexander, Greg Maddux, and Pedro Martinez. They each have two consecutive seasons that both appear near the top of these lists. They proved what they did was no fluke.
Cyril Morong teaches economics at San Antonio College and is a lifelong White Sox fan. A member of SABR since 1995, his articles have appeared in The Baseball Resarch Journal, By the Numbers and on line at The Chicago Sports Review.
[Additional reader comments and retorts at Baseball Primer.]