Correlation Fun With Statcast’s New Bat Tracking Data

You should all know by now that Statcast recently made their bat tracking data public. This is a big deal! Our own Ben Clemens made some early observations and also shared what the data does and doesn’t tell us, which included a correlation table between the new metrics and the familiar.

As a former projectionist, I care most about BABIP and HR/FB rate. So although the sample size probably isn’t big enough to definitively validate my reaction, I was saddened to see that none of the new metrics correlated with BABIP in a meaningful way. Rather than keep digging there, I’ll turn my focus to power.

How do these new metrics correlate with HR/FB rate? If any of them did correlate strongly, did they do a better job of explaining batter variation than existing Statcast metrics such as Barrel% (which I use frequently), maxEV, and others? Let’s dive into the numbers.

Correlations with HR/FB
Metric Correlation with HR/FB
Barrel% 0.737
HardHit% 0.579
EV 0.535
Blast Per Bat Contact 0.528
Avg. Bat Speed (MPH) 0.437
Fast Swing Rate 0.434
maxEV 0.372
Swing Length (feet) 0.233
Squared Up Per Bat Contact 0.000

There’s a reason why I frequently mention a hitter’s Barrel% when investigating their HR/FB rate — it’s the best proxy given the metrics we have available. Actually, Barrel% on fly balls is even better, but it’s not available on FanGraphs and would require additional data filtering on Baseball Savant. It’s simply good enough for a quick way to determine whether a hitter’s HR/FB rate is “real” or not.

Barrel% is by far the best metric, and then we see a dip to two more existing Statcast metrics, HardHit% and EV (average exit velocity). Again, correlations are better if filtering for just fly balls, but requires some extra work to get to.

Finally, we find the first new bat tracking metric, Blast Per Bat Contact, settling in at a meaningful 0.528 correlation with HR/FB. That’s not surprising, as blasts are defined as “when a batter squares up a ball and does so with a minimum amount of bat speed”. That’s essentially the bat tracking equivalent of the Barrel% metric, which combines launch angle with exit velocity. What perhaps is surprising is that the correlation is almost identical to average exit velocity, which includes all batted ball types, and a bit weaker than HardHit%. It’s therefore well below Barrel%, whereas I would have guessed it would have been stronger, ranking a solid second in correlation.

Two more of the new bat tracking metrics follow the Blast Per Bat Contact correlation, Avg. Bat Speed and Fast Swing Rate, which are almost identical. These make for pretty obvious strong correlations, as we know that higher bat speeds generally result in greater exit velocities, which in turn produce higher HR/FB rates. I calculate a correlation between Avg Bat Speed and average exit velocity at 0.575.

Next up is maxEV, which I also reference often when looking at HR/FB rates. A 0.372 correlation is meaningful, but it ain’t great, and certainly not as useful when projecting HR/FB rate than a bunch of other metrics.

We round out our correlation table with the final two new bat tracking metrics — Swing Length and Squared Up Per Bat Contact. Sure enough, Swing Length does have a meaningful positive correlation with HR/FB rate, but it’s significantly weaker than the rest of the metrics. Perhaps because a long swing doesn’t necessarily mean you possess enough power to record a high HR/FB rate, though the positive correlation does suggest that the longer the swing, the higher the mark. Interestingly, the Squared Up metric has absolutely no correlation with power. It also has essentially zero correlation with BABIP, so hopefully we figure out some uses for it, because it certainly seems valuable!

While Squared Up rate has no correlation with power, Squared Up Per Swing (I used Per Bat Contact for this article) does had a strong negative correlation with strikeout rate at -0.669. So it does have value, which is great, right? Wellllll, except that SwStk% has a 0.708 correlation with strikeout rate this season so far, so if you’re going to use one metric as a proxy for strikeout rate, it might as well be SwStk%, and not this new bat tracking metric.

So now that we have determined which metrics best correlate with HR/FB rate, I was curious how correlated with each other the metrics were. Maybe we can combine the best metric, Barrel%, with another, to come up with a strong xHR/FB rate equation. If you recall, I have come up with numerous xHR/FB rate equations in the past, the latest one incorporating Statcast data three years ago. But with new metrics on hand, it’s always smart to revisit, especially if it could be simplified.

However, I don’t want to combine metrics if they are highly correlated with each other. So let’s find out how Barrel% correlates with the others.

Correlations with Barrel%
Metric Correlation with Barrel%
HardHit% 0.702
EV 0.679
Blast Per Bat Contact 0.631
Avg. Bat Speed (MPH) 0.574
Fast Swing Rate 0.553
maxEV 0.552
Swing Length (feet) 0.267
Squared Up Per Bat Contact -0.066

My initial hope was to be able to combine Barrel% with Blast Per Bat Contact, but this table gives me some pause. The two metrics are highly correlated and the metrics actually correlate in the same exact order as they do with HR/FB rate. A regression equation using these two metrics to predict HR/FB rate resulted in a nearly identical correlation as just using Barrel%, meaning Blast rate added no additional value. The same result was produced when swapping Blast rate with Swing Length, the weakest non-zero correlation with both HR/FB rate and Barrel%.

With only about a month and a half’s worth of data, we’re still teetering on small sample size territory here, so it would be silly to give up so quickly. That said, the early returns suggest that the five new bat tracking metrics won’t help us project HR/FB rate (or strikeout rate or BABIP) any better than we’re already able to with existing metrics.

Hopefully, we do identify ways these metrics could provide additional value for predicting HR/FB rate and power (ISO). Perhaps Avg. Bat Speed hints at future power potential or Swing Length helps identify a group of hitters that are just an optimal swing length away from increased power. It might take a while to collect enough data to perform these studies, but my imagination is running wild with the possibilities right now.





Mike Podhorzer is the 2015 Fantasy Sports Writers Association Baseball Writer of the Year and three-time Tout Wars champion. He is the author of the eBook Projecting X 2.0: How to Forecast Baseball Player Performance, which teaches you how to project players yourself. Follow Mike on X@MikePodhorzer and contact him via email.

1 Comment
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Ryan DCMember since 2016
8 months ago

This type of work is so helpful; it’s easy to get pumped up about new metrics without actually examining whether they give us actionable fantasy info.