Measuring the greatness of Doc Gooden's 1985 season

The New York Mets retired the number of star 1980s pitcher Doc Gooden this week, prompting an interesting discussion on Brian Kenny’s MLB Network program, MLB Now. The thrust of the discussion amounted to this: Was Gooden’s 1985 season the best in recent baseball history?

It was an unusually subjective topic for the normally analytical Kenny to tackle, particularly because he did not try to delineate any objective methodology for answering the question; he merely opined. Yet it is possible to develop objective criteria to answer such a question. Let’s do it and see what the actual answer is.

It’s not as simple as comparing earned run averages or other common pitching metrics for obvious reasons: both pitching and hitting styles change over the years, and any of those changes can affect the relative gravity of performance. Just to take one illustration, the quality of a 3.50 ERA varies greatly depending on whether you were a pitcher in 1930 – when the average ERA was around 4.80 – or 1968, when it was 2.98

You can, however, arrive at valid apples-to-apples comparisons by calculating the standard deviation of a pitcher’s performance in any given year relative to his peers pitching under the same rules, conditions and expectations in the same season. By doing so you will arrive at a common and valid metric for every pitcher accurately reflecting the extent to which their pitching performance was exceptional by the standards of their time.

The table below uses that objective methodology to answer Kenny’s question in an unbiased fashion. For those unfamiliar with standard deviation, it is a measurement of exceptionality. Within any data set, about two-thirds of all data points will fall within one standard deviation plus or minus of the group average. About 93 percent will fall within two standard deviations, and virtually all the rest will fall within three standard deviations.

The basic question asked in this case is a straightforward one: Since 1970, and using ERA as the measurement standard, which pitchers who worked at least 162 innings achieved the largest standard deviation separation from their peers? Here are the top 10 seasons.

Pitcher, Year                      ERA        Std. Dev.

1              Pedro Martinez, 2000     1.74        3.79

2              Pedro Martinez, 1999     2.07        3.18

3              Ron Guidry, 1978              1.74        2.74

4              Zack Greinke, 2009          2.16        2.73

5              Greg Maddux, 1995        1.63        2.71

6              Johan Santana, 2004       2.61      2.62

7              Dwight Gooden, 1985    1.53        2.55

8              Greg Maddux, 1994        1.56        2.55

9              Roger Clemens, 1990      1.93        2.54

10           Kevin Brown, 1996           1.89        2.50

Pedro Martinez might've topped Dwight Gooden for greatest pitching season of all time

Anybody who was watching baseball in 1985 recalls how great Gooden was that year. He won 24 of his 28 decisions, pitched a league-leading 276.2 innings, struck out a league-leading 268 batters, and very deservedly won the Cy Young Award.

Yet as the table demonstrates, when adjusted for era-relevant exceptionality, Gooden’s performance – while great – wasn’t superior to all others. If you accept ERA as the governing metric, a half dozen other pitchers since 1970 have had more dominant seasons.

Pedro Martinez’s back-to-back 1999-2000 seasons are not only the most exceptional since 1970; they are in fact the most exceptional in the entirety of baseball history. The third best was Dutch Leonard’s 1914 season for the Boston Red Sox, in which his 0.96 ERA works out to have been 3.02 standard deviations better than his peers.

Perhaps, however, you’re not fond of ERA as the governing metric. Fine, let’s use ERA+, a relatively new stat that normalizes for era and ballpark. ERA+ adopts a scale on which a score of 100 equals a league average pitcher; therefore, a pitcher with an ERA+ of 200 is considered to have been twice as valuable as the average arm that season.

By the standard deviation of exceptionality using ERA+ as the metric, here are the 10 best seasons since 1970.

                Pitcher, Year                      ERA+     Std. Dev.

1. Greg Maddux, 1994 221         3.28

2.            Pedro Martinez, 2000     291         3.23

3.            Randy Johnson, 1999      184         3.07

4.            Roger Clemens, 1997      222         3.07

5.            Steve Carlton, 1972         182         2.98

6.            Steve Carlton, 1980         162         2.83

7.            Pedro Martinez, 1999     243         2.83

8.            Clayton Kershaw, 2013  194        2.81

9.            Dwight Gooden, 1985    229         2.76

10           Randy Johnson, 2001   188         2.73

Again, Gooden’s 1985 season is strong enough to make the top10, but far enough down that top 10 to deny the argument that it ranks as the best of the expansion era, or any era.

The striking thing to me about these two lists is how much both the order and the entries change depending on the metric used. Only four pitcher seasons – Martinez in 1999 and 2000, plus Maddux in 1994 and Gooden in 1985 – qualify as top 10 on both lists.

There’s a vital lesson in that: your evaluation of pitcher seasons depends greatly on which metric(s) you value. These rankings are based on ERA and ERA+, but they could as easily be based on wins, innings pitched, WAR, strikeouts, Win Probability Added, or on any combination of these metrics with absolutely no guarantee of similar results.

There is one other essential beauty to adopting standard deviation as the measuring stick for such cross-era ratings. When he posed the question of the exceptionality of Gooden’s 1985 season, Kenny limited the discussion to recent generations precisely because the science of pitching has changed so much. The mound is lower, expectations are different, workloads are less voluminous.

But because standard deviation assesses performance within the context of the prevalent trends of each season, it automatically normalizes for all those natural variations over time that would otherwise complicate and even confound a resolution of the question. If standard deviation is the yardstick, it is no longer necessary to exclude earlier seasons from the calculation because, while the expectations of modern pitchers versus earlier ones are far different, expectations within each individual season are virtually identical.

The question then ceases to be whether Gooden’s 1985 season was better than Guidry’s 1978 season or Dutch Leonard’s 1914 season, but whether Gooden’s 1985 season was more exceptional by comparison with other 1985 pitchers than Guidry’s or Leonard’s were by comparison with their 1978 or 1914 peers.  

10 amazing early-season pitching performances that have surprised MLB fans (calltothepen.com)