For Carter, a big part of his case for the top 100 list is in his defense. To do some analysis on that, I bring back the tool I used for catcher defense in the Carlton Fisk post – namely generalized linear mixed effect models. Briefly, in the baseball context, it works as a sort of automated WOWY analysis, with regression to the mean.
The specific thing I want to look at for Carter is preventing stolen bases. This means both throwing runners out and causing them to not run in the first place. I start with retrosheet play-by-play data, from 1921 – 2017. In order to isolate steal situations, and also provide some standardization in the situational context, I take only cases where
- there’s a runner on first
- second and third are empty
- the score differential is 4 runs or less
- the inning is the 8th or earlier
For each of these I model whether a stolen base occurred, based on the identities of the catcher, the pitcher, and the runner on first. This means a catcher gets credit for the case a runner chooses not to attempt a steal, as well as the cases he throws the runner out. The output of the model is the estimated contribution to the probability of a successful steal due to each of the catcher, pitcher, and runner – which can be thought of as an estimate of the “true talent” for preventing (or allowing) steals. For analysis purposes, I translate this into a “probability added” over what the probability would have been with an average player instead. There’s a lot of context this is glossing over and it’s far from perfect – but I don’t believe in letting the perfect be the enemy of the good.
With this framework, the following are the top 20 catcher season in terms of reduced steal probability
We see Carter’s 1979 right there at number 3, with his 1978 at number 20. On a per opportunity basis, the list is as follows
And we see that limited-playing-time 1976 season joining the list at #12.
With these mixed effects models, where you control for the context, you necessarily end up measuring the context – in this case leading to some interesting observations about runners and pitchers as well. As one example, following is the list of pitcher seasons that allowed the most extra stolen bases – a bunch of fireballers and knucklers!
As one more example, following is the top per-opportunity increase in stolen base probability for runners. The Terrance Gore 2016 season is fascinating – 11 steals in 17 G and 3 PA.