January 12, 2016
It’s Catcher Day at Baseball Prospectus, as we celebrate the expansion—both in method and in scope—of our new catching statistics. Given the age and breadth of some of these stats, we truly feel as if we are debuting our large adult child.
The statistics both apply to and measure players other than catchers, but they are all perhaps most important to catchers as we measure their total value to a team. The statistics are four-fold, covering three critical catching skills:
1. Running Game
a. Swipe Rate Above Average (SRAA) – the effect of the player on base-stealing success;
b. Takeoff Rate Above Average (TRAA) – the effect of the player on base-stealing attempts;
2. Blocking Pitches
a. Errant Pitches Above Average (EPAA) – the effect of the player on wild pitches and passed balls;
3. Framing (AKA “Presenting”)
a. Called Strikes Above Average (CSAA) – the effect of the player on strikes being called.
Obviously pitchers, runners and even batters can be subjects of these measures. We’ll publish those as well, providing deeper insight into just how aggressive Rickey Henderson was. But today is about catchers and not an excuse to mention Rickey Henderson. You never really need a reason to mention Rickey Henderson . . . but we digress.
Minor League Catching Models! And the Catcher Way-Back Machine
We’ve written about ‘retro-framing’ over the years, but now we’re publishing it—MLB framing now goes back to 1988. The pitch-by-pitch data used to generate MLB framing data for 1988-2007 (i.e. not PITCHf/x or Trackman) is also available for some of the minor leagues over the past decade. Triple-A framing begins in 2006, while Double-A framing begins in 2008—but just for the Texas League. By 2012 both the Southern and Eastern Leagues are included. 2015 provided a surprise splash of data from the NY-Penn League.
And it’s not just CSAA that we can extend. We can actually go as far back as 1950 for MLB with each of our other metrics. And we also take EPAA, SRAA and TRAA back to 2005 for almost all levels of the minor leagues.
Pre-1988 data is challenging because official pitch-by-pitch data is not available. While CSAA becomes unavailable, the other statistics can still be calculated. SRAA is based on stolen base attempts (so no change required), TRAA is already based on play-by-play, and it turned out EPAA works almost as well without pitch-by-pitch data, because play-by-play data turns out to be largely sufficient. It does require we scale the EPAA rate and chances to match pitch-by-pitch years, and the results do not have the additional accuracy afforded by the tracking data from 2008 forward.
Extended Integration with FRAA and WARP
During their inaugural season, some of these statistics (namely SRAA and TRAA) were published solely as a percentage above or below average. That is useful for comparison purposes, but doesn’t tell readers what usually interests them most: the effect in runs the player has in each category. We’ve now remedied this, and all four statistics are expressed in runs as well. As to catchers, SRAA and TRAA are combined into the category of Throwing Runs, because controlling the running game involves both throwing guys out and deterring them from running.
TRAA only provides run value to catchers who are credited to have prevented runners from going—those who did go have their impact measured via SRAA.
SRAA runs are ATTEMPTS*SRAA*(SB_runs - CS_runs). TRAA runs are calculated as CHANCES*TRAA*((SB_runs * SB_rate) + (CS_runs * (1-SB_rate))). SB_runs and CS_runs are the linear weight values for stolen bases and caught stealing; SB_rate is the league’s success rate on attempts.
EPAA is expressed as Blocking Runs, and CSAA is Framing Runs.
For catchers, this information is contained in two complementary tables. The first table is Catcher Stats – Full Season. This table combines all results from all levels and teams for each catcher into one set of composite statistics for each catcher over each baseball season. The second table, Catcher Stats - with team stints, breaks down the data from the full-season table, so you can separate minor-league from major-league performance, and compare their tenures with different teams.
The runs categories, and their best / worst performers among MLB catchers for the 2015 season, are as follows. The run effect is in parentheses:
● Throwing Runs (SRAA + TRAA):
o Best: Russell Martin (+2.5)
o Worst: Kurt Suzuki (-3.8)
● Blocking Runs (EPAA):
o Best: Brian McCann (+0.6)
o Worst: Tyler Flowers (-0.8)
● Framing Runs (CSAA):
o Best: Yasmani Grandal (+25.5)
o Worst: Carlos Ruiz (-.18.4)
As you can see, the importance of catcher framing is difficult to overstate.
Revised Model Extraction
All four statistics now used an updated mixed model procedure that, rather than extracting the coefficients for each player, performs a “with or without you” (WOWY) that separately totals each participant’s most likely contributions over the course of each season, both with the player involved and with a predicted-average player participating in their stead. The differential between those two totals gives us the seasonal value for each contributor, be it a steal, a ball in the dirt, or a called strike. The models are otherwise basically the same as we’ve previously described them before, and we welcome you to revisit those articles if you’d like more information about them, at these links: CSAA, SRAA and TRAA, and finally the recently-updated EPAA.
Of course, these statistics also still benefit from being mixed models in the first place. By applying shrinkage principles, the estimated values provided by these statistics are more accurate than raw averages. And by controlling for all participants in each play—rather than just assuming the others somehow cancel each other out—these statistics control for the quality of the opponents the catcher deals with, as well as the quality of his teammates.
Now go read all the great articles we have available making use—and occasional sense—of this wealth of catching information.
Jonathan Judge is an author of Baseball Prospectus. Follow @bachlaw