A few days back there was a discussion on Discord focused on pitcher stamina, and what it was good for. This, of course, is like chum in the water to me. I started asking myself questions and I started thinking about the data I take on a plate-appearance-by-plate appearance basis. Each of those identifies the pitcher and hitter, I thought. And if I could go and grab the pitcher’s stamin ratings from the game, then patch them in…well…maybe I could do something to study it.
So that’s what I did.
Bottom line, of course, is that what I’m posting here is interesting but mostly doesn’t confirm one thing of the other. But it was fun to look at, and in the end, for better or worse, I feel like I’ve come away with a more informed view of how Stamina works in the game. I should note that the data I have is a tiny bit noisy because using the method I used, it is difficult to fully remove a few low stamina pitchers who appeared early in games. I don’t think it’s a horribly big source of error, but it’s in there somewhere (meaning I didn’t care enough to take the hours it would have taken to pick the few bad apples from the bunch).
Anway…at the end of the process I grabbed the PA-by-PA data from the 278 pitchers who started games this year, and I mapped in the pitcher’s current Stamina Ratings (note, that’s another source of error. If the Samina rating changed from the time the stat line was created to now, then I’m bucketing the data wrong). Once complete, I began compiling information.
First up, let’s look at how many batters were faced by pitchers of each stamina level cross the innings. I wasn’t really interested in guys who got rocked early, so I compiled BF data for innings 4, 5, 6, 7, and 8. Then I looked at how they fell off. The assumption is that in those innings, low stamina pitchers would face more rapidly than high stamina guys.
Here is my result for raw batters faced:
| STM | 4 | 5 | 6 | 7 | 8 |
|---|---|---|---|---|---|
| 3 | 383 | 362 | 161 | 195 | 138 |
| 4 | 820 | 757 | 412 | 200 | 125 |
| 5 | 1529 | 1538 | 1113 | 728 | 665 |
| 6 | 5225 | 4923 | 3873 | 2459 | 1503 |
| 7 | 4353 | 3981 | 3258 | 2144 | 1399 |
| 8 | 4166 | 4035 | 3430 | 2097 | 1181 |
| 9 | 2604 | 2448 | 2110 | 1541 | 955 |
| 10 | 1312 | 1217 | 1197 | 855 | 573 |
That’s interesting enough, but I really want to look at fall-off rates (or really, I guess, retirement rates—how fast pitchers get removed by inning).
So here’s the same data, but shown as a %Falloff. In other words, Pitchers with three stamina faces 383 batters in 4th innings. They faced only 362 in 5th innings, which is a reduction of 5.48%. Here is that data:
| STM | 4-5 | 5-6 | 6-7 | 7-8 |
|---|---|---|---|---|
| 3 | 5.48% | 55.52% | -21.12% | 29.23% |
| 4 | 7.68% | 45.57% | 51.46% | 37.50% |
| 5 | -0.59% | 27.63% | 34.59% | 8.65% |
| 6 | 5.78% | 21.33% | 36.51% | 38.88% |
| 7 | 8.55% | 18.16% | 34.19% | 34.75% |
| 8 | 3.14% | 14.99% | 38.86% | 43.68% |
| 9 | 5.99% | 13.81% | 26.97% | 38.03% |
| 10 | 7.24% | 1.64% | 28.57% | 32.98% |
The most interesting number of all (for me, anyway) in this chart is the -5.78% that 5 STM pitchers saw between the 4th and 5th innings. This means that either my data is wrong (and we had “5” Stamina pitchers who started a game relieving a bunch in that span), or maybe “5” STM pitchers just gave up the ghost in that inning more often than others. I dunno. Like I said…Data noisy. Regardless, I think the data in interesting to scan both down the columns and across the rows. For example, adding the first two columns together, we see that the 4th-6th inning fall off was about 50% for “4” STM, 25% for 5, 6, and 7 STM, maybe 18% for 8 and 9 STM, and 10% for 10 STM.
Those are not surprising numbers, I suppose. Or at least they fit the model we have of high stamina pitchers lasting longer into games.
Note, though, that there is no real quality metric here.
But, of course I’ve got them.
Why, yes, I do.
And I’ll add them here in a few minutes…or whenever I get my next patch of time.