June 26, 2002
Aim For The Head
More on Lengthy Plate Appearances
The last Aim for the Head, about how offensive production varies with the number of pitches in the plate appearance, generated a lot of questions. Let's dig into the mailbag:
Ted Frank wrote:
Excellent article. One thing missing, though.
Many people brought up the point that strikeouts can only kick in at three pitches, and walks at four pitches. I did mention this in the beginning of the article, but the fact that so many people felt compelled to write in about it indicates that I didn't call it out sufficiently when discussing the results.
Per many requests, I've recalculated the rates of production removing strikeouts, walks and hit batsmen from the totals, which are presented below:
Pitch PA AVG OBP SLG OPS 1 283842 .323 .309 .497 806 2 332025 .319 .310 .491 801 3 288149 .317 .311 .488 799 4 234588 .317 .312 .481 793 5 174877 .318 .314 .491 804 6 102105 .320 .317 .494 811 7 41782 .327 .324 .517 841 8 15910 .331 .327 .530 857 9 5806 .333 .329 .551 880 10 2038 .317 .314 .525 839 11 760 .357 .352 .605 957 12 269 .404 .401 .683 1084
Removing strikeouts and walks accounts for nearly the entire change in production for at-bat lengths three and four pitches, although it's interesting to note that four pitches is also the length with the minimum OPS both with and without strikeouts and walks. Still, the variation across pitches 1-6 is very small, indicating that the length of the plate appearance generally has little to do with how effective the batter is when putting the ball in play in most cases.
However, we also see, as in the overall data, a gradual increase in OPS as the length of the PA increases. And since this takes strikeouts and walks out of the mix, this suggests that batters who have seen a lot of pitches may be able to hold out for a good pitch to hit, or that they are becoming accustomed to the pitcher's repertoire. It's also possible that such plate appearances are skewed towards better overall hitters, and are proportionally overrepresented in the sample.
Excellent work on the number-of-pitches analysis.
I grouped all plate appearances by inning, and considered it a long plate appearance if the batter saw at least seven pitches. I computed the average length of a plate appearance, and the percentage of all PAs that were long.
INNING PA LONG PA %LONG AVG_PIT AVG OBP SLG OPS 1 247473 16264 6.57% 3.85 .273 .346 .420 766 2 238237 13897 5.83% 3.71 .255 .317 .394 711 3 242647 13539 5.58% 3.65 .266 .328 .410 738 4 243417 13260 5.45% 3.59 .269 .331 .425 755 5 243169 12962 5.33% 3.57 .268 .328 .416 744 6 245253 13518 5.51% 3.60 .270 .334 .423 758 7 244368 13614 5.57% 3.63 .263 .330 .406 736 8 243416 14307 5.88% 3.68 .257 .326 .394 720 9 184213 11166 6.06% 3.72 .248 .316 .375 691
We'll look at long plate appearances in a moment, but take a look at the OPS column. While there's not a consistent trend, there are patterns that we can relate back to the way the game is played.
For example, the highest OPS posted is in the first inning, which is the only inning in which a team is guaranteed to have their top of the order batting. Similarly, the ninth inning is quite a bit lower, most likely due to the use of closers. The second-highest OPS is in the sixth inning, and the fourth through sixth innings in general are higher than the surrounding innings. This is probably due to two effects: pitchers tiring and getting hit harder in the middle innings, and mop-up or long-relief pitchers, who tend to be the less capable pitchers on a team, getting work.
But enough digression. C.W.'s question was about whether longer PAs occur towards the end of the game. In fact, the highest average length, and the highest percentage of long plate appearance come in the first inning, which is a manifestation again of the top of the lineup batting. Perhaps some of the effect comes from batters taking more pitches early on to see what the pitcher has working that day. But the ninth inning has the second-highest number in both categories, so it doesn't appear that we have a straight linear effect here. Let's graph %LONG and AVG_PIT by inning to see if that reveals anything:
What we see is not so much that long plate appearances come towards the end of the game, but rather that the middle of the game has decidedly shorter plate appearances than either the beginning or the end. So while C.W.'s intuition is partially true, you're even more likely to see long plate appearances when the game first begins.
You're kidding, right? That was a really, really interesting article—and thanks, by the way, for writing it—but you're really going to leave out the one point of trivia that every BP reader is undoubtedly dying to know? As silly and inconsequential as it is, as unimportant for its implications for your study, you nevertheless determined that in 12 years there were exactly two at-bats with 19 or 20 pitches. Aren't you going to tell us which two batters and which two pitchers were involved?
"Who had the 20-pitch plate appearance?" This was the single most common question I got, and some of the requests were more desperate than B.L.'s plea. Sure, I could have mentioned it in the original article, but it's much more fun to keep you hanging, and wanting more. Well, I won't keep you hanging anymore... the 20-pitch at-bat occurred on June 26, 1998. Bartolo Colon was pitching to Ricky Gutierrez with the Indians leading the Astros 4-2 in the eighth inning with none out in Cleveland.
An 0-2 count, and just 18 pitches later, strike three. It took 13 pitches to make it to a full count. This single plate appearance represents 18% of all the pitches Colon threw that day.
Eleven straight foul balls. That's more consecutive pitches fouled off than 99.8881% of batters see in their entire plate appearance. Bedrosian faced 10 batters and threw just 52 pitches in this relief appearance.
Lastly, here are some other trivial notes I have about long plate appearances (1988-2000):