AJ Mass’s Rule of Sevens

Methods

I turned as usual to the Lahman database and the tidyverse packages.

library(Lahman)
library(dplyr)

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

library(ggplot2)
library(hrbrthemes)

Conceptually we want to know how many players reach this plateau each year. Ideally we’d want qualifying pitchers, which means they pitched 162 innings since the season went to this long in 1962 for both leagues.

data(Pitching)

I filtered the pitching list by years after 1961. I then created a K/BB variable. After this I filtered by the Rule of Sevens rules, ERA < 3.5 and K/BB > 3.5 and qualifying pitchers (IP > 162). Finally I grouped by year and saw how many pitchers fit this mark.

Results

The code for the analysis yielded the following plot. There have been an increasing number of pitchers meeting these criteria since the 1990s. I overlaid a loess smoothing line over the points.

Pitching %>% filter(yearID > 1961) %>%
  mutate(K_BB = SO/(BB + IBB)) %>%
  filter(ERA < 3.5 & K_BB > 3.5 & IPouts > 162*3) %>% 
  group_by(yearID) %>% 
  summarize(Rule_of_7s_players = n_distinct(playerID)) %>% 
  ggplot(aes(x = yearID, y = Rule_of_7s_players)) + 
  geom_point() + 
  geom_smooth(method = "loess") + 
  theme_ipsum() +
  labs(y = "Number of players",
         x = "Year",
         title = "Rule of 7s Players")

Who are these great pitchers? I looked at the 2015 players to get a sense. These were some pretty amazing seasons. As a Mets fan it was sad to see how fast Matt Harvey went from being one of the top pitchers in the game to virtually out of the big leagues. And what got into Wei-Yin Chen that year?!

data(Master)

Pitching %>% filter(yearID == 2015) %>%
  mutate(K_BB = SO/(BB + IBB)) %>%
  filter(ERA < 3.5 & K_BB > 3.5 & IPouts > 162*3) %>% inner_join(Master) %>%
  select(nameFirst, nameLast, IPouts, ERA, BB, SO) %>% arrange(ERA) %>%
  as.data.frame()

## Joining, by = "playerID"

##    nameFirst  nameLast IPouts  ERA BB  SO
## 1       Zack   Greinke    668 1.66 40 200
## 2       Jake   Arrieta    687 1.77 48 236
## 3    Clayton   Kershaw    698 2.13 42 301
## 4     Dallas   Keuchel    696 2.48 51 216
## 5      Jacob    deGrom    573 2.54 38 205
## 6     Gerrit      Cole    624 2.60 44 202
## 7       Matt    Harvey    568 2.71 37 188
## 8        Max  Scherzer    686 2.79 34 276
## 9    Madison Bumgarner    655 2.93 39 234
## 10     Chris    Archer    636 3.23 66 252
## 11   Wei-Yin      Chen    574 3.34 41 153
## 12       Jon    Lester    615 3.34 47 207
## 13      Jose  Quintana    619 3.36 44 177
## 14     Chris      Sale    626 3.41 42 274
## 15     Danny   Salazar    555 3.45 53 195
## 16     Corey    Kluber    666 3.49 45 245

How about the 2005 players? Other than Carlos Silva’s unreal control year (1.6 walks per 100 outs!), it was the usual suspects.

Pitching %>% filter(yearID == 2005) %>%
  mutate(K_BB = SO/(BB + IBB)) %>%
  filter(ERA < 3.5 & K_BB > 3.5 & IPouts > 162*3) %>% inner_join(Master) %>%
  select(nameFirst, nameLast, IPouts, ERA, BB, SO) %>% arrange(ERA) %>%
  as.data.frame()

## Joining, by = "playerID"

##   nameFirst  nameLast IPouts  ERA BB  SO
## 1      Andy  Pettitte    667 2.39 41 171
## 2     Pedro  Martinez    651 2.82 47 208
## 3     Chris Carpenter    725 2.83 51 213
## 4     Johan   Santana    695 2.87 45 238
## 5      Jake     Peavy    609 2.88 50 216
## 6       Roy    Oswalt    725 2.94 48 184
## 7    Carlos     Silva    565 3.44  9  71
## 8   Bartolo     Colon    668 3.48 43 157

In the 1990s the number of pitchers meeting Mass’s criteria were much fewer. I looked at 1995 for an example. Greg Maddux and Randy Johnson were dominant, but what a year for Shane Reynolds!

Pitching %>% filter(yearID == 1995) %>%
  mutate(K_BB = SO/(BB + IBB)) %>%
  filter(ERA < 3.5 & K_BB > 3.5 & IPouts > 162*3) %>% inner_join(Master) %>%
  select(nameFirst, nameLast, IPouts, ERA, BB, SO) %>% arrange(ERA) %>%
  as.data.frame()

## Joining, by = "playerID"

##   nameFirst nameLast IPouts  ERA BB  SO
## 1      Greg   Maddux    629 1.63 23 181
## 2     Randy  Johnson    643 2.48 65 294
## 3     Shane Reynolds    568 3.47 37 175

Finally which pitcher has the most seasons where he met AJ Mass’s criteria?

Pitching %>% filter(yearID > 1961) %>%
  mutate(K_BB = SO/(BB + IBB)) %>%
  filter(ERA < 3.5 & K_BB > 3.5 & IPouts > 162*3) %>%
  count(playerID) %>% arrange(desc(n)) %>% inner_join(Master) %>%
  select(nameFirst, nameLast, n ) %>% filter(n > 2) %>%
  as.data.frame()

## Joining, by = "playerID"

##    nameFirst   nameLast n
## 1      Randy    Johnson 7
## 2       Greg     Maddux 7
## 3      Pedro   Martinez 7
## 4       Curt  Schilling 6
## 5        Roy   Halladay 5
## 6     Fergie    Jenkins 5
## 7      Sandy     Koufax 5
## 8       Mike    Mussina 5
## 9       John     Smoltz 5
## 10     Kevin      Brown 4
## 11     Chris  Carpenter 4
## 12     Roger    Clemens 4
## 13   Clayton    Kershaw 4
## 14     Cliff        Lee 4
## 15      Juan   Marichal 4
## 16        CC   Sabathia 4
## 17     Johan    Santana 4
## 18   Madison  Bumgarner 3
## 19       Jim    Bunning 3
## 20      Zack    Greinke 3
## 21      Cole     Hamels 3
## 22       Dan      Haren 3
## 23     Felix  Hernandez 3
## 24      Bret Saberhagen 3
## 25     Chris       Sale 3
## 26       Max   Scherzer 3
## 27       Tom     Seaver 3
## 28    Javier    Vazquez 3
## 29    Justin  Verlander 3

This list demonstrates the sustained excellence for Randy Johnson, Greg Maddux, and Pedro Martinez (and even the sometime greatness of Kevin Brown).

Conclusions

AJ Mass’s rule of 7s is pretty good for the last few years, but before this decade, you would have to be a truly elite top 10 player to meet both criteria.