What’s John Lannan’s secret?  Part 1

Jack Moore recently wrote an article at FanGraphs about John Lannan’s ability to defy fielding independent pitching metrics over the past couple of years. Over the past two seasons, Lannan has thrown 388.1 innings, with a 3.90 ERA, making him one of the more productive starters in baseball. However, his FIP is 4.72 during that span, and his tRA is even higher at 5.54 (remember, it’s scaled to RA) so his batted-ball profile doesn’t explain much either.

As one would expect, his BABIP during that span is .275, the lowest mark in the majors. However, this can’t be easily chalked up to good defense; the Nationals have been one of the worst defensive teams in the majors this year, and were only slightly above average last year. That implies something else is in play here. It’s generally acknowledged that BABIP is, for the most part, out of a pitcher’s control; however, some pitchers do have an ability to force a lower mark than others. Today, I wanted to take a look at Pitch f/x data to try to determine if Lannan has “earned” his BABIP, or if it is simply a function of good luck.

First, let’s take a look at what Lannan throws, organized, per usual, by horizontal vs. vertical movement. Remember that each dot (or circle in this case) is being compared to a pitch with zero spin (ignoring knuckling effects of course), and this is from the catcher’s point of view. You’ll notice that since John is a lefty, his fastball moves toward the first base side, while his slider and curveball break the other way.


As you can see, he throws a wide range of pitches. Although you can see some clearly defined clusters, the movement on a lot of his pitches tend to blend. My best guess is that he throws about 10 percent changeups, 10 percent sliders, 65 percent fastballs and 15 percent curveballs. Gameday classified three different fastballs from Lannan; a four-seamer, a two-seamer and a “normal” fastball; however, I couldn’t see enough of a difference in regard to velocity or movement to justify breaking them up like so. He may or may not be actually intending to throw different types of fastballs, but given most of them cluster around the same general attributes, we’ll treat them all the same.

Since the theme of this article is dissecting the cause of Lannan’s low BABIP, let’s check out his BABIP* by pitch type compared to that of a league average player:


*For whatever reason, I get the league average BABIP at .295, compared to the .300 shown at Baseball Reference.

These are definitely interesting results. Lannan’s slider has had more balls than average drop in for hits and his fastball and curveball both have lower-than-average BABIP’s, however, none of those figures are so far out of range that it so surprising. His changeup, on the other hand, has been ridiculous. So much that there is most likely something else in play besides normal luck on balls in play.

Let’s take a look at his changeup location to right handers (he’s only thrown 54 to lefties over the past two seasons), with all non-home run balls in play colored blue and everything else colored orange:


This is from the catcher’s point of view.

Like most lefties, he stays away from right-handed batters with the changeup. Accordingly, most of the balls put in play against him are on the outer half, and a pretty decent portion of those are at the bottom of the strike zone. That leads me to think that those balls would be hit more softly, and thus less likely to be a hit; however, don’t take my word for it. Let’s see what the data tells us. Here the average BABIP by pitch location on changeups to right-handed batters this season:


If it’s not clear, the nine boxes in the middle represent the strikezone, or at least a decent approximation of it, while the four boxes on the edges represent the 8-inch zone out of the strikezone.

As you can see, pitches on the outer third or just oustide of the strikezone generate a lower-than-average BABIP, while pitches in the middle third get clobbered. Surprisingly, changeups that are thrown in the inner third aren’t hit as well as you might expect. I guess that is where the element of surprise comes into play, as most batters don’t expect a changeup in that location. There may also be some selection bias in play, as only pitchers with very good changeups would have the guts to go inside with them.

Anyway, if you organize all of the balls put in play on changeups from righties against Lannan into the same zones as shown above, and multiply the frequency of balls put in play in each zone by the average hit rate, you can get a rough estimate of what Lannan’s BABIP allowed on changeups to righties should be, at least based solely on location. Doing that, we get an expected BABIP of .281 on Lannan’s changeup to right handers. Not bad – slightly above average – but a far cry from what his actual BABIP on that pitch had been.

Using that same process on all balls in play, to both righties and lefties, for each pitch type, we get an expected BABIP of .298, which is actually above the league average (again, I have it at .295). Obviously, this is an oversimplification. Lannan’s pitches are not exactly average, and we probably shouldn’t expect a league-average BABIP, even after adjusting for location. However, we can now reasonably say Lannan has simply gotten lucky on balls in play thus far in his career. You win again, Voros.

I’m not done with this yet though. For some reason, I still think there is something special about Lannan that allows him to have a lower-than-average BABIP. Stay tuned next time for some stuff on batted-ball location and pitch sequencing.

Newest Most Voted
Inline Feedbacks
View all comments
14 years ago

Two questions:
What method did you use to get expected BABIP? And why does the changeup location graph have shadows? Kind of fuzzy looking.

Nick Steiner
14 years ago

The shadow is a default setting in excel, I didn’t bother to change it, but I can see that it obscures the image a little.  I’ll be aware o

To get xBABIP, I took the location distribution each of Lannan’s balls in play to RHH and LHH.  So 8 different batter hand – pitch type states and 13 data points in each of them = 104 total bins.  Then I multiplied the league average hit rate on those balls in play based off of those same 104 bins and summed all of the results.

You can see the spreadsheet I used here: