A Batter’s Hard Hit Angle: Introduction

I had no idea who Dylan Moser was. In all respects to Dylan and his family, I still don’t. When I saw an article about him come through my feed, I was interested in how Tanner Stokey described Driveline Baseball’s evaluation of Moser. While Driveline has its own advocates and critics, it pushes the research bounds so I wanted to see what they considered important about Mr. Moser. Immediately, I saw this little nugget.

The “Average Hit Angle of Hard Hit Balls” caught my eye and I’ve been investigating its implications ever since.

Determining and finding the effectiveness of a hitter’s launch angle spread has been investigated several times in the past. Andrew Perpetua pointed out the importance of High Line Drives (a close cousin to Barrels) and how too much of an uppercut can hurt a player’s production. Alex Chamberlain and Brock Hammit both found a link between the standard deviation in launch angle and increased production.

It was because of the work from the last two that Stokely’s article piqued my interest. Moser had an Average Launch Angle of 13.36 degrees while his Hard Hit Launch Angle was -5.67 degrees. If a hitter only makes solid contact on groundballs, his upside is limited. I just assumed that most hitters would have batted balls evenly distributed on both sides of their ideal swing has Perpetua shows here:

I needed to dive in to find out more.

First, I came up with a standard for a Hard Hit ball. Instead of using the 95-mph standard from Baseball Savant, I moved the value up to 98 mph which catches 20% of all batted balls and has an xwOBA over .400.

I'm trying to define an MLB hard-hit ball. One chart is the xwoba on all hard-hit balls at different velos.

For reference, Trouts wOBA is in the .440 range.

I'm leaning between …
90: wOBA over .300, ~25% of BIP
98: wOBA over .400, ~20% of BIP
100: Easy to remember, 15% of BIP pic.twitter.com/JsBxKBU9GE

— Jeff Zimmerman (@jeffwzimmerman) May 10, 2020

With the threshold, I determined I needed to use the difference between Average Hard Hit Launch Angle and the Average Launch Angle for all other batted balls (Average Hard Hit Difference or AHHD). This decision was filled with trial-and-error and I sure some other more predictive and usable variants could be found. It’s the best that a man, a weekend, and some High Life could find for now.

Determining the stat’s repeatability was next. If the stat shows no sign of repeating, it’s useless for evaluating players. Here are the year-to-year values for the match pairs with a match harmonic mean of 500, 300, and 100 balls in play.

The season-to-season values have some predictive value and with the stabilization point somewhere around 100 to 200 balls in play.

Since AHHD is predictive, I went down several wormholes and came out finding that AHHD pairs best with the Average Launch Angle and the Maximum Exit Velo for being predictive. The Average Launch Angle provides a value that the AHHD is centered around. If two hitters have the same AHHD, the one with the Average Launch Angle of 15 degrees is more productive than the one at 0 degrees. The Max Exit Velo is needed because a hitter can have the perfect launch angle but if his Max Exit Velo is 90 mph, he’s going to be unproductive.

I’ve been trying to use several methods to evaluate data to see where the final differ. I used three models to help predict several outcomes using Hit% (i.e. BABIP plus home runs). Simple. Batting Average contains strikeouts and BABIP removes home runs so I need a compromise. I took the three values and found the linear regression equation. Also, I ran them through the TREE and CTREE functions in R to determine certain groupings.

Linear Regression

TREE Function

CTree

There is a lot of information, so it’s time to simplify. The biggest takeaway for me with the linear regression model, the final three variables are highly significant … but I knew that since I had already done some trimming to just these three. While the launch angle is significant in this equation, it barely shows up in the two trees.

The TREE function provides some simple benchmarks. I hitter with a Max EV at 109 mph or more and an AHHD over 4.4 puts their Average on All Batted Balls at least 20 points higher (i.e .318 vs .336) but could be 70 points. I’m coming back here soon.

The CTree output is a little harder to take in. The underlying function found the 109 mph break. The next node (#9) divides AHHD almost at 0 with those with a positive AHHD having a higher Hit Rate. Max Exit velocity shows up three more times on the right side with dividing values at 111.5, 113, and 114.5. Also, at node 10, the -4.4 AHHD value shows up again.

Using the above values, I grouped the hitters into several categories (min 30 balls in play).

First and foremost, the AHHD variable takes a major backseat to Max Exit Velo in determining any hit rate. At the major league level, it takes a certain power baseline to be an above-average hitter. The difference between just the two groups split at 109 mph is stark. 26 points in Hit%, 8 points in BABIP and AVG. The difference only goes up from there as the data is split into smaller groups. The other change is that a decent AHHD adds about 5 points to a hitters BABIP. One use for this data for is to regress BABIP and similar measures to these individual batted-ball measures instead of assuming one league-average rate,

With preceding information known, here is how hitters performed in 2019 with the various values (min 100 PA).

There is a ton to digest but the list passes the eye test for a batter’s ability to make hard solid contact.

I’m calling it quits for today. There’s a ton that can be be done with these measures and I have a few more articles planned. One option is to determine if a hitter’s struggle or breakout is from a swing adjustment. Also, instead of all batted ball, the batted balls off fastballs can be examined to find out how aging vets are adjusting to keep up with their declining bodies. All of this and much more from a profile on Dylan Moser. Thanks to Tanner Stokey and everyone else at Driveline for the little nugget.