Every Revolution Has Casualties
On Wednesday Jeff Sullivan wrote a piece about the ‘flyball revolution’, which I am sure many of you have already read. Today I want to expand upon what he wrote, and tie in a theory for what may be going on. My theory is probably wrong, but it ties together a few of the articles I’ve written recently (1, 2, 3, 4,) and expounds on my current thinking on the Statcast data.
I know I’ve been touching on the topics of exit velocity and launch angle a lot recently, but I think it is important to understand and begin integrating into everyday analysis. Before anyone can use that analysis, we as a community need to build up an intuition for how the stats relate to one another. This is new for everyone.
Firstly, I want to present two images. These images represent the frequency of batted balls on given launch angles and exit velocities. One is for the 2015 season, and the other is for 2017. I am omitting 2016 since it isn’t particularly interesting for this analysis, just know that it is about half way between the other two. The vertical axis represents vertical launch angle, and the horizontal axis shows exit velocity. Blue shows low frequency, and green is high frequency. The black bar shows zero degrees of vertical angle.
There is a lot we can unpack from these charts. Notice how in 2015 the higher frequency green area is larger, especially towards the lower launch angles. It also extends into lower exit velocities. In 2017, the high frequency batted balls are ever so slightly shifted up in angle and velocity, but not to any great degree.
Notice how deep the blue area dives into the very low launch angles in 2015 compared to 2017. In 2017 there are far more balls in the very high launch angle regions, where very few balls were hit in 2015.
This ‘flyball revolution’ is much more complicated than we first assumed, and I do believe it is better categorized as a change in contact point, not a change in launch angle. Launch angle is only an unintended bi-product of a different contact point.
My Theory (it might be wrong)
I believe you can divide the launch angle and velocity chart in a manner that roughly corresponds to the contact point in front of home plate. I have done so in the chart below. There are four zones, three of which are labeled. You have batted balls that are struck too far in front of home plate and those that are not struck far enough in front of the plate. Next you have batted balls that are poorly struck, which I labeled as a “miss.” Perhaps it is hit off the end of the bat, the handle, etc. Finally, you have balls that are in the Goldilocks Zone.
The boundaries in this chart are roughly sketched and aren’t meant to be taken too seriously, but they should serve as a general guide for where I believe they may lie. These regions are based upon the following pieces of information:
- Contact point correlates with home run rate and exit velocity
- In a pattern that matches the relationships between home run rate, exit velocity, and launch angle.
- Launch angle should increase very rapidly above the peak home run contact point.
- Exit Velocity and launch angle are intimately related.
- Players are openly stating they are trying to hit the ball further in front of home plate.
If all of this analysis is correct, there appears to be a certain point ahead of home plate where launch angle will steeply rise. This appears to be around 30-32 inches in front of home plate. Home run rate peaks around 29-30 inches. These numbers might sound large, keep in mind we’re measuring from the back of home plate, which is 17 inches deep.
There is very little margin of error. If a batter hits the ball even 3% further from home plate than he would like, his launch angle could rise 5 or more degrees. And overshooting by 8% would increase launch angle by as much as 10 or more degrees.
We’re talking about average launch angles here, so you can hit ground balls as well, but it appears the chances of a pop up increases with distance in front of home plate. So, if you are trying to maximize the number of batted balls hit around that peak home run area, you are very likely to also significantly increase your pop up rate.
With all of that said, obviously more batters are electing to ride this razor’s edge than even two or three years ago. The pop up rate has increased substantially. The ground ball rate has decreased. These are easy to measure, but I don’t believe they are the key to what is going on in baseball right now.
Earlier I noted how the frequency of batted balls in the mid launch angles (-20 to 30 or so) and mid/low exit velocities (70-80 mph) has dropped in 2017 relative to 2015. This gif should sum it up pretty well. The square shows batted balls hit between -10 to 12 degrees and 70-92 mph. There are fewer balls hit in both the lower exit velocity and lower launch angle regions. Instead, you see increases in other areas, including the extreme high launch angles.
Batters do not appear to be intentionally hitting the ball into the air. Rather, they are attempting to maximize the value of their batted balls. In order to do so, they are trying to hit the ball far out in front of the plate, which is an ideal location for producing high value line drives and fly balls. However, there is very little margin for error, and overshooting your target by even a very small amount dramatically increases the odds of a high launch angle, low exit velocity batted ball. I believe this explains the dramatic increase in weak fly balls and pop ups (40+ degrees) with only a modest increase in the lower angle, more valuable flyballs (20-30 degrees).
Andrew Perpetua is the creator of CitiFieldHR.com and xStats.org, and plays around with Statcast data for fun. Follow him on Twitter @AndrewPerpetua.
Is there any way to correlate this info with the batter’s position in the box? In other words measuring the distance from the front of the box as a z-coordinate?
I don’t think I have batter position data, but maybe I’m overlooking something.
PS, if you want to nerd out over this stuff, the Y coordinate is from home plate to the pitcher, X is from first to third (or vice versa, this is the one that always confuses me), and Z is from ground to sky. This is why pitch location is in px and pz, for example. Although that Y coordinate is the magic that I seek!
This seems like chicken versus egg argument. Batter’s want more value. Did they want to accomplish this by getting out in front which causes more balls in air or did they want to get more balls in the air and getting in front is how to do it. Personally, I’m guessing that people figured out that getting it in air is important first and the getting out in front is the solution on how to do it (but could go either way).
Well, what is truly valuable in a baseball game isn’t what is traditionally thought of as ‘in the air’ by most people. If you were a perfect robot, you’d likely want to hit the ball about 19-26 degrees every single time.
While that is ‘in the air’, it isn’t necessarily a ‘fly ball’ in the sense that you might otherwise consider. These are somewhere between a flyball and a line drive. A very shallow fly ball. The sort of flyball that would make you say ‘wow, I didn’t think that was high enough’ when it goes over the fence for a home run.
So, in that sense, yeah, batters want to hit the ball in the air. But when you talk about the uptick in flyballs, you aren’t talking about *those* hits. You’re talking about the traditional flyballs, the ones that go high up in the air. The ones an outfielder can run to and camp under to make an easy catch. These hits can be doubles, homers, etc too, obviously. They often are. Probably 50-70% of homers are hit like this. Maybe more. But these aren’t the balls that maximize value.
I am not talking about the “traditional flies” as you assume when I say in the air with regards to the fly ball revolution, but the more valuable fly balls. I think most people that when talking about getting the ball in the air in regards to the fly ball revolution, they are talking about flies that are closer to liners instead of the ones that are easy to run under.
I stand by thinking contact point is more the how do we achieve fly balls (valuable fly balls that is if you need the clarification).
I want to see if the response to this batting trend will be to value pitchers (even more) than induce groundballs and possibly stopping the progression toward seeking arms that miss bats (and/or allow hard hit flyballs when they don’t miss those bats). iow, is the current trend of the 3 true outcomes coming to a head?
Related, we are beginning to see teams value contact hitters more, if for the moment just for supply/demand reasons. These evolutions in the game are interesting to follow/observe.
Great work! I think you made a really good point that the adjustments in swing have resulted in a sharp increases in grounders and pop ups, which tend to cancel each other out in terms of average launch angles.
I like to think about this in terms of attack angle (swing plane). Players have increased their attack angle, which allows them to generate larger EV at higher angles. That increase in attack angle also results in more grounders and popups when the timing is slightly off. If you’re interested I just did a community post about getting attack angles from Statcast: https://www.fangraphs.com/community/reverse-engineering-swing-mechanics-from-statcast-data/
I don’t know exactly how players have made this adjustment to attack angles. You might right that they are just thinking about making contact out in front of the plate, which improves the timing in such a way that it increases the attack angle without having to make any mechanical adjustments to the swing.
Yes, I read that yesterday, good job. You confirmed a suspicion I had regarding the statcast swing speed. I was nearly positive that it was based on Alan Nathan’s formula, which he wrote about shortly prior to those swing speeds going up on Savant.
Anywho, the saying is “what goes up must come down.” In a baseball swing maybe it ought to be ‘what goes down must come up.’ At some point in the swing, the bat must be on an upward path.