An increase in publicly available data can often help our understanding of the sport. The rollout of Statcast data has been fascinating. Learning how hard Giancarlo Stanton hits a ball, how fast baserunners and fielders move to steal bases and make catches, and how hard outfielders and catchers throw the ball is all very interesting information. Up to this point, it can be tough to determine if the information is useful or if it is more akin to trivia knowledge, like batting average on Wednesdays or pitcher wins. An examination of the batted ball velocity against pitchers provides some hope of providing potentially important information, but until we have more data — and more accurate data — conclusions will be difficult.
Looking at the top of the leaderboard in exit velocity, it is easy to see why linking a low exit velocity with good performance is enticing. I looked at all pitchers with at least 150 batted balls in the first half and 100 batted balls in the second half. Here are the top-five pitchers in batted-ball exit velocity this season, per Baseball Savant, along with their ERA and FIP.
|Exit Velocity (MPH)||Batted Balls||FIP||ERA|
Looks like a pretty good list with some of the very best pitchers in baseball on it. Here are the next five.
|Exit Velocity (MPH)||Batted Balls||FIP||ERA|
|Jorge De La Rosa||86.07||335||4.20||4.17|
The above charts only contain ten pitchers, and Madison Bumgarner, Johnny Cueto, Michael Wacha, and Shelby Miller are all among the next group of pitchers, but Chris Archer and Gerrit Cole are both in the bottom 20 of exit velocity. That does not make the data useless, but making conclusions is going to be difficult when trying to determine if there are outliers or if the data is simply more random than we would like to see.
To get a better sense of what’s going, I separated this season’s starting pitchers into three groups: those whose average exit velocity on batted balls is under 87.5 mph (21 of 91 pitchers), those whose exit velicoty sits somewhere between 87.5 mph and 89.2 mph (constituting 51 of 91 pitchers), and a final group of 19 pitchers who’ve allowed an average batted-ball exit velocity greater than 89.2 mph this season. The initial analysis appears to show some effect on ERA and FIP depending on the batted ball velocities.
There is roughly a half-run difference between the high and low groups, with the middle group falling squarely in between. There looks to be a bit of a difference between the two groups. We, or at least I, would assume that there would not be a great relationship between exit velocity and individual pitcher ERA or FIP given how many other variables go into determining those statistics. Given that, an r of .33 for ERA and an r of .40 for FIP seem surprisingly high.
Given FIP’s slightly stronger relationship with exit velocity, a look at the components of FIP could shed some light on their relationship. The chart below shows the same groups from the graph above using proxies for the FIP components in strikeouts, walks, and home runs per nine innings (K/9, BB/9, and HR/9, respectively) depending on the average exit velocity.
|Exit Velocity <87.5 (n=21)||7.9||2.5||0.92|
|Exit Velocity 87.5-89.2 (n=51)||7.6||2.7||0.99|
|Exit Velocity >89.2 (n=19)||7.4||2.5||1.19|
While walks and strikeouts never happen at the same time as a batted ball, it is possible that exit velocity might be some proxy for a pitcher’s stuff that would show up in strikeout and walk rates. Given the size of the sample, it does not look like exit velocity plays much of a role in the number of walks given up as it fluctuates by group and the r was a very small .15 between walks and exit velocity. Looking at the raw numbers, it looks like K/9 might have some relationship with exit velocity. The r is a little stronger at -.25, but when calculating FIP, that .5 difference in K/9 accounts for only a one-tenth difference in FIP.
The strongest relationship among the FIP components is HR/9: the r is .34 and the data moves in a linear fashion. The .27 difference between the extreme groups makes a big difference in FIP, as it moves the FIP by roughly four-tenths. These relationships mirror the one between the FIP components and FIP itself, but with a much stronger relationship with FIP, as would be expected. Surprisingly, the correlation coefficient between exit velocity and FIP at .40 is roughly the same (.37) as walks and FIP.
FIP is a very good statistic to use because it takes BABIP and defense out of the value equation. While pitchers do exhibit some BABIP skill, it requires a large sample to ascertain. In the case of defense, that’s simply not the pitcher’s responsibility. Attempting to tie a pitcher’s skill to BABIP suppression is incredibly difficult. We know that there is little to no relationship between BABIP and the Hard, Medium, Soft designations used on FanGraphs, and this data supports the notion that average exit velocity for pitchers and BABIP have almost no relationship. There is virtually no relationship between BABIP and home runs for pitchers this season, and there is also no relationship between BABIP and exit velocity for pitchers. If we were hoping to get closer to BABIP as skill, this data gets us no closer, and if we believe that suppressing exit velocity is a skill (which is up for debate), this actually lends support for the argument that pitchers have little control over BABIP.
At this point — or perhaps it would have been better to do it earlier — it seems appropriate to acknowledge the limitations of the data we have available. Tony Blengino discussed the issues earlier in the summer, determining that roughly one-quarter of all the batted ball data is unavailable, and that the numbers which are available are the product more often of positive outcomes (because Statcast has trouble recording weakly hit balls). Blengino concluded:
Don’t get me wrong… Statcast is a great thing, and we are only scratching the surface of what it can eventually become. The sample generated for this article yielded some benchmarks which will serve as the foundation for some analysis you will see here in the coming weeks. Still, when one is faced with a data set, one must put it into some sort of context, while acknowledging its limitations. In many of my previous articles here, I have warned readers to never take pure average velocity data at face value; launch angles, BIP type frequencies, pull percentages, etc., significantly affect hitter and pitcher performance, and can be easily be overlooked. For this year, at least, we should additionally be aware of the Statcast data set’s unique shortcomings, which adjust the context within which analysis takes place.
So it seems like there might be some sort of relationship with production (FIP) and exit velocity, and most of this relationship is due to home runs, but this is where it starts to get difficult between what we have and what we want. We know that home runs are hit very hard. We know that limiting home runs leads to a lower ERA and FIP. In a small sample, it appears that pitchers with lower-than-average exit velocities succeed in limiting home runs as comparted to their brethren on the other end of the spectrum — and removing home runs from the seasonal exit velocities results in essentially the same exit velocities overall. What we really want to know is if lower exit velocities are a repeatable skill. If they are — and if we were able to solidify the link between those velocities and home runs — we could identify pitchers we might expect to allow fewer home runs in the future. We are not there yet. There is some indication that batted ball velocities are at least a portion of the pitcher’s responsibility and that they stabilize, but more research, more data, and more accurate data is necessary before we can get to the bottom of this.
Craig Edwards can be found on twitter @craigjedwards.