Last time, I took a look at one of the bigger enigmas in baseball, John Lannan of the Nationals. To rehash, since 2007, he has thrown nearly 400 innings with a 3.90 ERA. However, based on his peripheral statistics, you would expect his ERA to be in the mid to high fours. The main reason that Lannan has been able to do this is by means of a .275 batting average on balls in play, or BABIP. For those unfamiliar with the concept, hit rate on balls in play is generally considered to be out of a pitchers control and tends to settle at around .300 for most major leaguers. Hence, it is often used as a proxy for when a pitcher has been lucky or unlucky. However, it’s also been accepted that some pitchers have an ability to force a lower than average BABIP. Just look at Johan Santana.

Given that Lannan has only pitched 388 innings, the most likely scenario is that he has simply gotten lucky and has no special ability to induce a lower batting average on balls in play. A further examination of his location by pitch type, did nothing to assuage that notion. However, I didn’t want to leave it like that. Something about Lannan makes me feel funny. Maybe it’s the consistency (ERA’s from 3.88-3.91 the past two years), or maybe it’s the fact that he’s had so much success on an awful Nationals team. Either way, I had a feeling that Lannan was doing something right. Today, I wanted to check out a few more possible factors to find out what that is.

First, let’s start with pitch sequencing. For this, I took all possible combinations of two different consecutive pitches in the same at bat, 16 in all, and measured the frequency at which he threw them against both righties and lefties. I then looked at his BABIP on the second pitch of each combination, and compared it to the league average BABIP by a LHP for the second pitch of that combination. Here are the results:

*For Pitch 2 only

For the most part, there is a lot variation on BABIP based off of pitch sequencing. From a left handed pitcher to a left handed hitter, the range is .129 points of BABIP, which is huge, and the standard deviation is .036. From a left handed pitcher to a right hander hitter, the range is .080 and the standard deviation is .023. The best combinations appear to be back to back offspeed pitchers. Fastball-slider to a lefties and fastball-change to righties also do well.

Like we did with location, if we multiply the league average hit rate on each combination by the frequency that Lannan induces a ball in play and sum the results, we can get an expected BABIP based off of pitch sequencing (or at least back to back pitches). In this case, his expected BABIP comes out at .295, exactly league average. Again, we see no evidence that Lannan is doing anything special.

So it looks like we have to dig even a little deeper to see cause of Lannan’s low BABIP. If you’ll look at the above chart, you can see that he has allowed a very low BABIP on the changeup, especially when it follows a fastball. While this can certainly may just be luck, the fastball-changeup combo has been the weapon for soft throwing lefties for as long as I can remember. It seems just too convenient that Lannan has allowed such a low BABIP on the tail end of that combo without some skill. So let’s take a look at both parts of that sequence on balls in play to righties for Lannan:

His changeup generally mirrors the fastball location. More specifically, a little under 60% of his changeups put in play that followed a fastball were within 1.5 total feet (horizontally and vertically) of the fastball. So what does that mean? Well, to find out, I took a look at all changeups put in play by a RHH against a LHP that followed a fastball. I then found the total distance between each each changeup and the fastball that preceded it, and order them into 16 groups. I then plotted that against the BABIP for each group:

As you can see, the results were… inconclusive. The trend line “peaks” at a BABIP of around .275 for changeups with 1-1.5 feet of separation from the fastball; however, the correlation is weak, although not terrible, and each data point had under 400 observations. Considering that, when we plot out the separation on each of Lannans changeups to lefties following a fastball that were put in play and model expected BABIP using the equation of the line shown above, we get… .285, which **exactly league average** for a changeup following a fastball.

The other pitch combination that had a big effect on his BABIP was the fastball-fastball combination to righties. If we do the same thing as above, we get a relationship among all major leaguers that looks like this:

This time, the correlation is a little stronger and the trend line is more pronounced. BABIP peaks at about .285 for fastballs with 1.5 – 2 feet of separation between them. If you use the equation of that line on each of Lannan’s balls in play that came on a fastball-fastball sequence, you can an overall expected BABIP of .295, which, once again, is almost exactly the league average mark for that situation.

Now, I think I’m done. I’ve tried looking at the location of each pitch, the sequencing, and the relative location of the two different pitch combinations that were most responsible for his low BABIP. I’ll still root for Lannan, and I may keep looking for something else, but all signs point towards Lannan not doing anything special to be able to control his BABIP.

Will said...

Fantastic article. Thanks for doing this.

I’m anxious to see how Lannan fares next year.