Statistical Anomalies To Figure Out Dan Fox’s Reliever Magic

After graduating from the Naval Academy, I was commissioned as a submarine officer.   Once I reported to my submarine, I was assigned to manage the very sensitive Quality Assurance program and all its paperwork.   This is one of the systems that has made submarines the gold standard for engineering practices. NASA adopted this system from the submarine program after the Space Shuttle Challenger exploded.

The officer I was relieving gave me a great piece of advice when reviewing the paperwork.   He said whatever the equipment or maintenance was, review several years worth of paperwork and look for anomalies and numbers that were different: valve torques, pounds of pressure, temperature, thicknesses, welds, should all be the same from year to year.  If there was a number that was different, that was a cause for concern or inquiry.

Well, when I was on Fangraphs looking at 2015 statistics, this chart on the front page caught my attention.

 

Of ALL the relievers in baseball, the Pirates had two of the top five in WPA in Melancon and Watson.

Per Fangraphs: “Win Percentage Added (WPA) captures the change in Win Expectancy from one plate appearance to the next and credits or debits the player based on how much their action increased their team’s odds of winning. Most sabermetric statistics are context neutral — they do not consider the situation of a particular event or how some plays are more crucial to a win than others.”  (WPA is not context neutral.)

So much like with my Quality Assurance paperwork, this WPA outlier caught my attention and caused me to dig deeper.

I first looked at individual relievers.   I looked at NL relievers (wanting to compare apples to apples and not involve the DH).  Of the 298 relief pitchers in the 2015 NL, the Pirates had:

The top two –  Melancon (1), Watson (2),

3 of the top 15 – Hughes (15)

5 of the top 30 – Caminero (29) and Bastardo (30)

7 of the top 33 – Soria (31) and Blanton (33)

As far as a statistical distribution, that’s pretty ridiculous…ridiculous in a good way.  This “red flag” lead me to look at Team Pitching WPA’s to see if I could find another outlier to explain this to me.

I copied and pasted what I found below for Team Pitching WPA as well as just Starting Pitcher WPA and just Relief Pitcher WPA.

 

Comments:

• The NL Central dominated overall WPA results, with the “lucky” Cardinals leading all baseball and being 30% better than the 2nd place Pirates.
• The Dodgers, not surprisingly, lead all starting pitcher WPA but their bullpen knocked them down to 6th in the overall rankings.

As far as the Pirates are concerned, the relievers crushed the rest of the league, including the acclaimed Royals bullpen.   This dominance of the Pirates’ relievers brought their overall WPA to second even though their starter WPA was average and slightly negative at -0.23.

So if this was the submarine I would go to the system expert: the nuke mechanic Chief or the electrician Petty Officer and ask “Why is this different?”  For the Pirates, the system expert is (most likely), Dan Fox.  Fox’s official title is Director of Baseball Informatics and he oversees six others in the Informatics/Analytics department for the Pirates.  But there’s no way that I’ll be able to extract that information from the notoriously close-to-the-vest Pirates.

In the baseball talking head community, I can pose the questions to our saber friends like Eno Sarris and Mark Simon: “Why is this different?”  Is the way the Pirates are selecting relievers or the way Clint Hurdle is using them leading to this trend?

I doubt Dan Fox is going to tell me.