The 2017 HOF Pitching Nominees

Clemens, Mussina, and Schilling

Recently the nominees for the 2017 Baseball Hall of Fame were announced and Baseball Reference has a nice statistical summary of these nominees.  Scanning the list, only eight pitchers appear and I’m particularly interested in three starters:  Roger Clemens, Mike Mussina, and Curt Schilling.  How do the careers of these pitchers compare to pitchers that have recently been inducted in the HOF?

One useful way to look at a pitcher is to decide on an appropriate metric and then plot this metric as a function of the player’s age — the so-called career trajectory.  What do we look for when we view a trajectory?  Generally, I expect a player’s performance level to increase (or decrease depending on the metric) until the late 20’s or early 30’s when he achieves peak performance, and then decrease until retirement.  The age at which he achieves peak performance is interesting — pitchers can peak at different ages.  Also I would like to see consistency of a pitcher’s performance from season to season.  Inconsistency or high variation in a pitcher could mean that there are other factors influencing the pitching metric.

I thought it would be interesting to compare Clemens, Mussina, and Schilling with the “Big Three” Braves pitchers Greg Maddux, John Smoltz, and Tom Glavine who recently were inducted in the HOF.

Win/Loss Percentage

Generally, a pitcher’s win/loss percentage is not a good measure of pitching performance — it better reflects the quality of the team.  So I would think that pitcher trajectories of win percentage would be pretty volatile.  And they do appear to be — you can gauge this inconsistency by the wide standard error bands surrounding the loess smoothers.  All of these pitcher had good win/loss records — it is interesting that Roger Clemens was successful towards the end of his career.

hof2017a.png

WHIP Trajectories

A better pitching measure is the number of hits and walks allowed  per inning (WHIP).  It is desirable that this is low, so one might expect a pitcher’s WHIP trajectory to have a local minimum around age 30.  We see this pattern for Maddux, Schilling, and Smoltz, although Schilling and Smoltz peaked later than Maddux.  Clemens, Mussina, and Glavine have relatively flat trajectories.  It it interesting that Clemens actually had a decrease in WHIP towards the end of his career.

hof2017b.png

Strikeout Rates

Of course, a WHIP value is influenced by the quality of a pitcher’s fielders.  A pitcher has sole control over strikeouts, so a plot of a pitcher’s SO Rate (SO divided by IP) might be of interest.  Clemens had an unusually flat SO Rate trajectory.  Schilling’s trajectory is a bit more inconsistent — perhaps injuries may have been a cause of this.  Maddux was not a strikeout pitcher compared with the rest, but his SO rates were very consistent over his career.

hof2017c.png

FIP (Fielding-Independent Pitching)

A modern measure of pitching performance is the FIP stat — it is an estimate of a pitcher’s ERA using only measures under his control such as HR allowed, SO, BB, and HBP.  Maddux has a well-defined trajectory which peaks around age 30.  Clemens in contrast has a FIP trajectory that is pretty constant but more unstable as a function of age.

hof2017d.png

What Have We Learned?

  1. These trajectory graphs could be improved.  For example, I did not adjust these for the seasons when these players played.  There have been substantial changes, say in home run hitting and in strikeouts, over seasons and it is better to make adjustments in these measures.  For example, one simple adjustment is to compare a z score that subtracts an average value and divide by the standard deviation.
  2. Generally, Clemens, Mussina, and Schilling have pitching careers that compare favorably with other players such as Maddux, Glavine, and Smoltz who are already in the HOF.  Of course, there are some other issues regarding Clemens and Schilling that might be considered among the baseball writers who cast their ballots.

R Code

These trajectory graphs are remarkably easy to construct using data from the Lahman package and the ggplot2 package makes it easy to compare these trajectories using facets.  The entire script I used can be found on my github site and I encourage the interested reader to plot his/her trajectories for other players in the 2017 HOF nominee class.

 

Advertisements

One response

  1. Very good presentation Jim!. Thanks for sharing the code to try with other players

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: