The Batted Ball Project: Launched Flies (5 of 6)

Previously on the Batted Ball Project 

We identified three distinct “batted-ball” components of offense:

  • BABIP (the familiar “hits per balls-in-play” which does not encompass home runs);
  • 2b3b / Bttd, which is doubles & triples per all batted balls (which does encompass home runs); and
  • HR / Bttd, which is homers per all batted balls.

We then used those concepts to develop the “batted ball” versions of the two components of OPS:

  • BACON (“batting average on contact” or ‘hits per batted ball”); and
  • BtISO (“batted-ball ISO” or ISO based solely on batted balls).

We have been applying these stats to our database of 537 qualified MLB hitters with at least 1,500 career plate appearances since 2002 (when batted-ball data is first available).  All of which accounts for almost 1.9 million cumulative PAs.


Now: fly balls.

Except that fly balls include pop-ups.

Fortunately, the batted-ball data reports pop-ups (infield flies) as a subset of flies (although only as a percentage of all fly balls).  With a bit of math, we can exclude pop-ups from the fly-ball totals.  The resulting stat I call “launched flies”  — fly balls excluding pop-ups.

Since the analysis confirms the obvious — that pop-ups do not lead to base hits period — we will limit the analysis we report here to “launched flies.”  And the shorthand will be “LF%” (launched fly percentage).

[Note: I have retained the same scale for each chart across all the articles so that they are visually comparable.]


First, launched flies and BABIP:


Here we find a negative relationship between launched flies and BABIP.  It’s actually pretty much the mirror image of the chart for ground balls:


Even the slope of the trendline and the “goodness of fit” (R-squared) are mirror images.  All of which makes sense, since launched flies are not conducive to singles.  As we concluded with ground balls, the relationship is “real” but fairly mild and only a relatively small piece of the big picture.


So what about doubles + triples?

In the prior article, we were surprised that line-drive rate had little discernible impact on doubles + triples.  Well, it turns out that launched flies are a stronger factor here:


The slope is not steep, but the scatterplot is relatively tight, which shows up in the “goodness of fit” (R-squared of 0.13).  In other words, we can find a meaningful explanatory relationship between launched flies and doubles + triples, in a way which we could not for line drives.  It’s kind of mild, but it’s there.


And here’s launched flies and homers:


Now here we have a “fit” that’s stronger than what we’ve seen before.  An R-squared of 0.44, indicating a lot of explanatory power.  In other words, we learn a lot about home runs from launched flies.

But wait.

Home runs are launched flies.

Aren’t we just saying that guys who hit a lot of home runs hit a lot of home runs?

Yeah, kinda.

[Technically, if I remember right, that’s called an “endogeneity” problem.]

So let’s run a couple more tests to see if our data holds up.

First, let’s exclude home runs from “launched flies” total, and see how it affects the HR rate:


It’s obviously not as strong — the scatterplot gets “blobby-er” and the R-squared drops below 0.1 — but it is true that hitters who launch more balls in the air hit more homers.

And what about the other way around?

Do guys who hit more homers also hit more non-HR extra-base hits?  Yes.  (Be aware that this chart is on a different scale.)

hr-2b3bThose relationships make me comfortable that our reasoning is not circular.

In other words, launching the ball hard in the air seems to have a direct effect on offensive production.  And we still see it when we look at the relationship of non-HR batted balls to HR, and the relationship of HR to non-HR hits.


Now to our two batted-ball components of OPS.  First, getting on base (not making outs) with batted balls, measured by “BACON.”

lf-baconNow, if you recall, the net effect of ground balls on “BACON” was totally neutral with a flat trendline and miniscule “goodness of fit” (R-squared).  And here we have the exact same thing.

We don’t find any relationship between launched-fly rate and a hitter’s chance of successfully reaching base on a batted ball.  The only time we found a positive relationship there was in line-drive rate.


But offensive production from batted balls is a different story.

lf-btisoThis is the strongest stuff we’ve seen.  Production of offense from batted balls (ISO measured with batted-balls only, such that K% doesn’t affect it) is most affected by the hitter’s ability to launch the ball hard in the air.

There’s no two ways about it.


So to summarize:

  • Launched-fly rate is neutral with respect to reaching base on batted balls
  • Launched-fly rate is positive with respect to producing offense

Here’s part 6.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s