A discussion of the value and shortfalls of the statistic called Catcher's Earned Run Average (CERA).

INTRODUCTION

Most experts (meaning managers, coaches, pitchers and catchers) believe that the aspect of the catcher's job that has the most impact is his game-calling, that is, his ability to work with pitchers and help them throw more effectively. The standard and most acceptable measure for a pitcher is the Earned Run Average (ERA). Baseball is a game that has statistics for virtually everything, but there seems to be precious little time and energy devoted to measuring how well catchers perform at calling the game. Rather, we see catchers’ defense measured by how many base stealers they throw out or how many passed balls or errors are charged against the backstop. A recent attempt at measuring a catcher's defensive skills is the CERA, which basically is the Earned Run Average of the battery (catcher and the pitchers on a team) for each specific catcher as compared to all other catchers and their batterymates.

            The most comprehensive published study on the subject is Craig Wright's "Catcher's ERA" in his book The Diamond Appraised. Craig defined a process whereby catchers on the same team can be compared by how well a common set of pitchers perform with each catcher. That is, Catcher A's and Catcher B's CERA for Pitcher 1 are compared for the differences. The resultant CERA can be used to draw a conclusion as to the intrateam value among catchers.

PROBLEMS WITH CERA

Furthermore, there is the situation of the alternate (backup) catcher being used as a late inning substitute and paired with mop-up bullpen hurlers, generally in a losing cause. The starting catcher would have very few innings with these bullpen guys (usually with a high ERA) while the backup catcher would have few innings with the #1 and #2 starting pitchers (who usually have lower ERA's). Then there is the phenomena of Grag Maddux. When he pitched for Atlanta he preferred to throw to backup catcher Eddie Perez instead of the number one guy, Javy Lopez. Because of Maddux's preference and low ERA, this would preclude any matched pairings or if pairings were ignored the scales would tip in Perez's favor.

The next concern with CERA (and by no means the last) is the way that CERA is now being captured and presented in various publications which form the core of the CERA statistical library. The Bill James Handbook formerly published by STATS, Inc. and now ACTA, do not use matched pairings, but rather capture all of a catcher's innings and earned runs regardless of the pitchers involved. It is a raw total report that in and of itself is very misleading.

In Table 1. below you have two equal catchers (A and B) who have the very same CERA for each and every pitcher (1, 2, and 3) they caught. The only difference between the catchers is in the number of innings caught for each pitcher although their cumulative total innings are identical. "CATCHER A" only caught 50 innings with "PITCHER 2" (ERA of 4.50) while "CATCHER B" caught 110 innings and had the identical CERA. "CATCHER B" is penalized (in his cumulative CERA of 3.86) for doing the same job as "CATCHER A" only because of the way the CERA raw total statistic is formulated.

Table 1

The summary columns (highlighted in green) would be what is published and used for analysis in comparing CATCHERS A & B. It would seem that CATCHER A with a 3.56 CERA is a far better defensive guy than CATCHER B. (CERA of 3.86) However, we know different by looking at the numbers inside the numbers. Both catchers are equal. Although Table 1. is completely fictional it does demonstrate one of the potential problems with using raw totals in computing CERA.

Do these problems make CERA an invalid measure for a catcher? The answer is YES and NO. Keith Woolner (Baseball Prospectus) concluded that there was no statistical significance between catchers' ERA and that the differences were purely a matter of chance variations or randomness. However, his study only used a subset of a subset of all catchers and pitchers because he only included batteries (pitchers and catchers) with 100 innings or more. He is correct in his conclusions, but only for the data set he chose to use. Perhaps a different approach using every catcher and pitcher in some weighted fashion would make CERA relevent. But, that's a lot of data to manipulate. Fortunately we have Retrosheet and its play-by-play data which eventually might give us true validity. In the meantime, we can use the raw CERA data (Innings & Earned Runs) to roughly describe comparisons and limit our convictions of the results by the known shortfalls.

PRESENTING THE "IFFY" CERA DATA

The following two tables list the TOP 25 CATCHERS in Best CERA for a Season and Cumulatively (data for the years 1990-2003 only). The list has been sorted by d-CERA which is the difference between an individual catcher's CERA and the other catcher's ERA (O-ERA). The O-ERA is computed as the Team ERA (T-ERA) components (INN and ER) minus the catcher's CERA components.

Table 2 - Season CERA Leaders (100 Games Caught Minimum)

Table 3 - Cumulative 1990-2003 CERA Leaders (500 Games Caught Minimum)

FINALE

The future of a game-calling measure, whether it be CERA or RPR (Run Prevention Rate) or some other formula derived from the play-by-play numbers, will still have to answer two questions:

• Do the differences in game-calling measures (ie. CERA) among catchers vary from what we'd expect solely from chance or are the variations statistically significant?
• Are the year-to-year game-calling measures for a catcher capable of being trended?

A catcher gains game-calling ability with time. The longer he's in the majors the more he's learned about his pitchers and the opposing hitters. A catcher must remember every pitch sequence to every hitter so as to avoid being predictable, a catalog that might stretch back several innings or several years. The longer he catches the more game-calling ability he possesses, which should show up in the numbers - some numbers somewhere. And that is the challenge - To identify the right numbers and assemble them in the right way.