Cliff Lee, the middle of the plate, and BABIP

Warning, long post ahead.  I will be spending 1500 words to argue a very simple point.

If you were poorly adjusted when you were growing up like I was, you might have wondered if the Grinch and Santa Claus were the same person.  After all, why else would they wear the same outfit; why did Santa* deck all of his elves in green regalia.  Maybe he just got tired of bringing Chinese-made toy cars to little kids and sweaters to husbands and wanted to troll.  Maybe Cliff Lee felt the same way.

(*Actually, to paraphrase someone much wiser than myself: what kind of a creepy old man has thousands of little kids sit in his lap during the day, and then breaks into their houses at night and eat their food?  How suspicious is that?  Don't even get me started on Rudolph - I mean he gets used as a headlight just because he has a red nose.  If you ask a bald man to be a lightbulb for you, he'd punch you in the face.)

Need more evidence?

Q.E.D

For whatever reason, Lee has sucked in his latest three outings.  He did not have his curveball in one of those starts, but his stuff seemed otherwise fine.  It was obvious to everyone that the reason for the poor outings was simply too many pitches in the middle of the plate, or is it?

Here are the pitch locations from four of Lee's starts this year, can you guess which ones were good starts, and which ones were poor?  (The caution and caveats in using pitch f/x data applies here - for example, velocity of movements of pitches are sometimes not very consistent.  However, I've only used pitch locations here, and that tends to be pretty accurate.)

(A) is his latest start in KC: 4.2IP, 10 Hits, 7 R (4ER), 0 BB, 5 SO; (B) is his start on July 4th against Detroit, back when he pitched for Mariners: 8IP, 9 Hits, 1 R, 1 BB, 11 SO; (C) is a start on August 6th against Oakland: 8 IP, 7H, 1R, 0 BB, 8 SO, and (D) against Baltimore on August 21st, 5.2 IP, 10 H, 8 R, 1 BB, 4 SO.

Essentially, these charts included two of his worst starts, and two of his best ones.  Were you able to tell which one was which before checking the answers?  Your humble writer here could not.

Phew, that was fun, let's try another one!  (Extra credit: guess which starter on the Rangers these charts belong to)

it's CJ, guys.  (A) is his recent start in Baltimore - one of the best starts this season by anybody on the team: 8.2 IP, 3H, 0R, 1BB, 12 SO.  Seems reasonable, right?  A lot of pitches down in the zone = good results. 

(B) is from his Aug. 4th start against Seattle: 3IP, 6H, 4R, 4BB, 3 SO.  6 hits in 3 innings isn't good at all, but there really isn't that many pitches right down the middle, and (C) the Aug 25. start against Minnesota, 6 IP, 7H, 3R (2ER), 2 BB, 6 SO.  The pitches here seem more elevated compared to A and B, he gave up a few more hits, but it is an overall mediocre outing.  You may have figured that C would have been a bad start, but did you think it would be worse than B (other than the # of pitches giveaway).

The problem is that every starting pitcher throws a large number of pitches right down the middle in every one of his starts.  It's not correct, and lazy, to simply say that the result of a bad start is simply because he left too many pitches in the middle of the plate.  What do you mean "too many"?  Did he throw 25 pitches there instead of 20?  I think we all have a selective bias when it comes to this - if he gets hit hard, we take notice of pitches down the middle.  If he doesn't, we don't really notice them.

"Well Mr. Legraf", you say, "are you trying to tell me that pitch location doesn't really matter?  Do you ever watch the game?". 

To the first question: of course not, and the second: yes, but you still wouldn't believe me.

In fact, we know that the location of pitch matters - consider this study by Jeremy Greenhouse earlier this year.  The first graph is a dependence of BABIP on the location of pitches, divided into handedness of hitters and pitchers.  The details vary a bit, but the general trend is pretty clear: middle of the plate and up, BABIP of above .350.  Down in the outside corners, BABIP of .270 or lower.  The implication is pretty clear, if a pitcher is able to throw every pitch down on the outside corner, he can be expected to maintain a low BABIP.  Similar things apply for release point shown in the second graph, and pitch type in the table at the end.  Of course, there are other studies like pulled fly balls tend to have higher BABIP and those going into the opposite field, and effects of velocity and movement on run values (take into account HR and the type of hit... single vs. double, etc).  You can find dozens of more studies of this nature.

That said, Lee's BABIP as a Ranger is now .310 (with a 18% Line drive rate), a number higher than the .290-.300 league average.  Also, about 41.5% of base runners has scored on him,higher than the average 30%.  Both number would suggest that he has been unlucky.   Speaking as someone who rely a lot of stats, I think there is a tendency to get lazy and over rely on them. For example, if he has indeed been throwing everything down the middle where the BABIP is expected to be inflated as shown in the studies above, perhaps the balls were harder hit and he is not quite so unlucky.   BABIP is essentially an "empirical" stat: we know what happened, but we do not know how or why.  Pitch f/x has allowed people to poke at the "why", and for example, if a pitcher can throw every pitch down in the bottom outside corner, there is no reason why he cannot maintain a BABIP of .250 or lower.  

But the fact is that in the last couple of decades, no  pitcher, over a large enough sample size, has been able to consistently maintain a well-below average BABIP.  This is a fact, you can look it up on fangraphs or B-R, or whatever your favorite website may be.  It may be possible to maintain a low BABIP for one season, but a lot of times, a pitcher who post a .250 one season gets a .330 the next.  It is all really luck?  Why can't someone with good command hit those low-BABIP spots consistently?

Perhaps it is because pitching is hard, and consistently hitting spots, over the course of a season is hard.  The pitcher throws a pitch about 55 feet away from home plate.  Since the strike zone is roughly (but generously) 2 feet wide, the pitcher would ideally want to aim for the edge of the zone, and avoid missing by more than 4 inches, at which point the ball would be at the center third of the plate.  If you do a quick calculation, in order to have the ball miss than less than 4 inches to either side (thus 8 inches total), or from top to bottom, his trajectory of the ball cannot deviate more than 0.62 degrees when he releases the ball.  Pull our your protractors and try to imagine a 0.62° angle - it's pretty small.  This is assuming that everything else is perfect - movement of your pitches are always consistent; no weather effects; no different in mounts, etc.  Being able to throw a baseball exactly where you want it is really hard, and because of these, a lot of pitches end up down the middle.

Finally returning from a long detour to the original point: like TAG said, normally only a small fraction of middle pitches turns into hits.  Certainly, some pitchers will throw more pitches down the middle than others over a large sample, and will get hit more often and harder, but throw down the middle is not so easily avoidable, and should not be used exclusively to explain a poor performance by a pitcher over the sample size of a couple of games.  There are simply too many other factors: pitch sequencing, the hitters, deception, stadium conditions, defense, and luck.

At the same time and for the same reasons, we can't just attribute everything to luck either.  Afterall, there always seem to be games where a pitcher seem to have better command, and throw fewer pitches in the middle of the plate, but people are not machines, and they have periods of physical peaks when they are able to repeat their deliveries, release points and movement of their pitches, and other days when they cannot.  There may be real information behind be noise that gets ignored by the blender of averaging.

Most importantly, no matter which camp you are arguing from, the argument needs to be supported by some type of evidence that an impartial observer can then look and decide for himself.  No matter how correct or obvious something may sound, it never hurts to look it up and double check yourself.