Sunday, December 26, 2010

Catcher Defense - Part 1

The catching position is the most difficult to quantify defensively.  Instead of the physical range characteristics cited for infielders and outfielders in earlier posts, the handling of pitchers is believed my many insiders to be the most important skill of any catcher.  By pitcher handling, I mean studying opposing batters, game calling, understanding pitcher abilities and tendencies, helping the pitchers maintain focus and other duties unique to the catching position.  These things are hard to measure because it's difficult to know how much of good or bad pitching is due to the pitcher and how much is due to the catcher. 

Bill James attempted to measure pitcher handling when he created the catcher ERA (CERA) statistic in the 1980's.  CERA is simply the ERA of a team's pitching staff when a given catcher is behind the plate.  The idea is that pitching staffs should have lower ERAS when a superior defender is catching. 

A limitation of CERA is that different pitcher/catcher combinations do not accumulate enough innings over the course of a season for it to be considered reliable.  Another concern is that it can be biased by which pitcher the catcher's catch.  For example, if a catcher was the personal catcher for the team's best pitcher, his CERA would be artificially deflated by the quality of the pitcher, instead of his own skill. 

To address the bias issue of CERA, John Dewan introduced the earned runs saved statistic in The Fielding Bible - Volume II.  Simply stated earned runs saved is the number of earned runs that a catcher saves his pitching staff.  For example, Mike Mussina, had the the following statistics pitching for the Yankees in 2008:

IP 200 1/3
ER 75
ERA 3.37

Jose Molina caught 190 1/3 of Mussina's innings and the Mussina/Molina combination posted the following numbers:

IP 190 1/3
ER 68
ERA 3.22

Now, suppose Mussina actually had an ERA of 3.37 (his ERA for the season) in the 190 1/3 innings that Molina caught.  In that case, the Mussina/Molina duo would have allowed 71.3 earned runs.  Subtracting the 68 actual runs allowed by the battery from the 71.3 yields 3.3 earned runs saved for Molina in games pitched by Mussina.  Summing Molina's earned runs saved for all the pitchers he caught yields 31 earned runs saved for the season.  Using the same technique, Brandon Inge cost his staff 37 runs (-37) in 2008.

Because it adjusts for quality of pitchers (based on their ERA for the season), earned runs saved is less biased than CERA.  However, it is still limited by small sample sizes for pitcher/catcher combinations.  Thus, Dewan arrives at a more conservative estimate by regressing to the mean:

(1)  He multiplies earned runs saved by .33 (31 x .33 = 12.2).  He admits that the .33 is arbitrary.  The idea is to give credit to Molina for saving runs without relying too heavily on an extreme number produced by small samples. 

(2) He further regresses to the mean based on the number of innings caught.  The more innings a catcher catches, the more credit he gets for his earned runs saved.  A full season is roughly 1,440 innings.  Molina caught 737 innings which is roughly half a season.  So he regresses by half ( 12.2 x .5 = 6.1)

Dewan also adjusts for ballpark.  In Molina's case, he ended up with 5 adjusted earned runs saved for the season. The adjusted earned runs saved can be found at Baseball-Reference (under RerC).  The 2010 leaders and trailers are shown in Tables 1 and 2 respectively. 

Table 1: Catcher Adjusted Earned Runs Saved Leaders for 2010

Table 2: Catcher Adjusted Earned Runs Saved Trailers for 2010

According to these numbers, new Tigers catcher/designated hitter Victor Martinez cost the Red Sox pitchers 5 runs with his pitcher handling in 2010.  Alex Avila cost the Tigers staff 2 runs last year.  Gerald Laird saved the Tigers 2 runs.  

Other facets of catcher defense beyond pitcher handling will be explored in a later post. 



  1. I have never seen such a useless formula, it disregards so many variables (such as opponents faced etc) while also using variables with no basis whatsoever (like multiplying by .33).

  2. Rex,

    This is why it is part 1 of the series!

    I don't know if I'd call it useless but even Dewan admits that it is a work in progress. It is also limited by small sample sizes. Even with refinements, it will probably require several years of data before it becomes reliable.

    I agree that the.33 multiplier is whimsical. it is simply used to regress towards the mean. Dewan is giving catchers with high earned runs saved credit for their work but wants a conservative estimate.

    As far as opponents faced, pretty much all statistics (batting average, ERA, wOBA, etc.) disregard opponents faced. So, this statistic is no different on that basis.




Blog Archive


My Sabermetrics Book

My Sabermetrics Book
One of Baseball America's top ten books of 2010

Other Sabermetrics Books

Stat Counter