Monday, October 09, 2017
Statcast Lab: WOWY Battery
Prelude: introduction to WOWY (With or Without You).
***
Since 2012, Roberto Perez is among the best catchers at throwing out basestealers. With 114 stolen base opportunities of 2B, basestealers are successful only 55% of the time, well below the 72% league average.Perez has had four primary pitchers (all RHP as it turns out) in that time:
24 opps, 8 SB with Bauer (33%)
16 opps, 7 SB with Carrasco (44%)
14 opps, 4 SB with Kluber (29%)
12 opps, 11 SB with Shaw (92%)
48 opps, 33 SB with rest (69%)
As we can see, the Perez combos with Bauer, Carrasco, Kluber is well better than league average as well as the average of Perez with anyone else. So, some of that low 55% stolen base rate is a result of Bauer+Carrasco+Kluber. Or... is it?
We can also repeat the exercise going the other way, showing how these pitchers perform without Roberto Perez:
36 opps, 29 SB Bauer-noPerez (81%)
42 opps, 26 SB Carrasco-noPerez (62%)
72 opps, 51 SB Kluber-noPerez (71%)
28 opps, 19 SB Shaw-noPerez (68%)
As you can see, the Bauer-Perez pairing was far better than the Bauer-nonPerez pairing, while the Carrasco-Perez pairing was, in comparison, slightly better than Carrasco-nonPerez. And suddenly, Shaw is better off without Perez.
Indeed, Perez without his big 3 was close to average, and the big 3 without Perez were close to average. But, when Perez is paired with the big 3, suddenly, each of these batteries was well-above average.
The question is how much to rely on the actual battery combination, and how much to rely on the performance of these players when they are not a battery.
In studying all the battery combinations in MLB since 2012, we can break down the pairings as follows:
- the battery in question
- the performance of the pitcher without that catcher
- the performance of the catcher without that pitcher, but with the same pitching hand
- the performance of the catcher without that pitcher, but with the opposite pitching hand
If everything was random, then all the data would get weighted proportionate to the number of observations. But, we know that not everything is random.
And in fact, the weightings, for each observation, comes in at:
100%: the battery in question
40%: the performance of the pitcher without that catcher
13%: the performance of the catcher without that pitcher, but with the same pitching hand
13%: the performance of the catcher without that pitcher, but with the opposite pitching hand
My prior expectation was 100/30/20/10. So, these numbers are, kinda, inline with my prior. The biggest surprise to me was that the pitching hand, once you account for the 5% points of bias, had no additional information.
In other words, how Bauer-Perez truly interact is going to be based, disproportionately, on our observation of the two as a battery. And the battery performance generates well over twice the signal as their non-battery performance (pound for pound).
The end result, the "true talent" of the batteries are:
54%: Perez with Bauer
55%: Perez with Carrasco
58%: Perez with Kluber
70%: Perez with Shaw
66%: Bauer with average catcher
63%: Carrasco with average catcher
67%: Kluber with average catcher
74%: Shaw with average catcher
62%: Perez with average RHP
In the end, we can see that there is the most synergy with Bauer, and the least with Shaw. The Bauer-Perez combo is 12 points better than the Bauer-average-catcher, while the Shaw-Perez is only 4 points better.
Further research is required to determine if we want to apply a "familiarity factor". After all, maybe all these popular batteries has a synergy factor that allowed the pairings to shine when you have a familiar battery, but not shine, when not familiar.
***
It's alot of effort to go through this process. A simple regression toward the mean will get you almost all the way there. For catchers, you add 58 opportunities at league average, and for pitchers, you add 38 opps. And then you apply a standard Odds Ratio Method to get the true-talent for the battery in question. That is, we apply no synergy effect and no familiarity effect. That's the Marcel-like method, and it works almost as well as WOWY Battery.
***
This process actually opens the door to a whole host of possibilities. Do you want to know if the observations we witness of a direct Kluber-Betts pairing as pitcher-batter gives us any additional insights beyond their individual talents? Well, WOWY Battery will get you there. How about Felix-Safeco? Yes, that works too.
And we're just scratching the surface with possibilities. I'll detail more in the future, though you may be able to guess at a few already.
Recent comments
Older comments
Page 1 of 152 pages 1 2 3 > Last ›Complete Archive – By Category
Complete Archive – By Date
FORUM TOPICS
Jul 12 15:22 MarcelsApr 16 14:31 Pitch Count Estimators
Mar 12 16:30 Appendix to THE BOOK - THE GORY DETAILS
Jan 29 09:41 NFL Overtime Idea
Jan 22 14:48 Weighting Years for NFL Player Projections
Jan 21 09:18 positional runs in pythagenpat
Oct 20 15:57 DRS: FG vs. BB-Ref
Apr 12 09:43 What if baseball was like survivor? You are eliminated ...
Nov 24 09:57 Win Attribution to offense, pitching, and fielding at the game level (prototype method)
Jul 13 10:20 How to watch great past games without spoilers