Studying MiLB Statcast Exit Velo Data
Sometimes in life, you figure out that you’re being deceived or lied to. Regardless of whether or not the lie or deception was intentional or malicious or accidental, it does make it impossible to trust the source ever again fully.
No, I’m not going through a divorce.
This week we saw Ryan McMahon hit a ball 117 miles per hour in a Spring Training game. Except, no we didn’t. I was immediately skeptical of this (I’ve basically been immediately skeptical of everything I hear my whole life, and even more so since the COVID era), so I did some digging and found that indeed, there was no reason to believe that velocity reading.
So that took away a lot of the appeal of Spring Training data for me. The only things I was really interested in were which pitchers were throwing new pitches, pitcher velocities, and maximum hitter exit velocities. One-third of those things is now gone. It’s okay, I’ll work through the rough patch.
The problems aren’t done there, however. As you may know, last year I started collecting all of the available MiLB Statcast data. I put it in a dashboard that is available to paid subscribers of this blog. It was all extra exciting because in AAA we know that the International League (AAA) will be adding the Hawkeye technology to capture all of that data. That means we’ll have the data for every single AAA game - really exciting stuff.
But now, I don’t know if I can even believe the data. So that’s what I’m here to try to figure out (I don’t think I’m going to be able to come to a firm conclusion, but as I said before - any bit of doubt kind of ruins the whole thing).
MLB Launch Velocity Data
We trust MLB data. We basically have to, or else I guess we should just go back to the 2014 way of watching baseball. I’d rather die.
But when we compare the ballparks, things check out.
I looked at all batted balls hit between 70 and 110 miles per hour (this is most of them, but the filtering was just to get rid of the outliers), and then found the average launch velocity at each park. Top 3, league average, and bottom 3 shown here:
Seattle 86.6
Milwaukee 86.7
Tampa Bay 86.8
~League Average: 87.2~
Boston 87.5
Kansas City 87.5
Colorado 87.6
So the entire range is just one mile per hour, with every park within 0.6 miles per hour of the average. Really consistent stuff.
MiLB Launch Velocity Data
I’m looking at only the Pacific Coast League data from 2021 and 2022. There was one other league (the Florida State League which is A-ball) that used the same hawkeye system as the Majors, and the rest of the stadiums just used Trackman - a different system. Since AAA will be using Hawkeye in 2023, that’s what I want to study here.
Here are the PCL numbers for the same query (batted balls between 70 and 110 miles per hour):
Southwest University Park (Padres): 91.7
Isotopes Park (Rockies): 91.5
Chickasaw Bricktown Ballpark (Dodgers): 91.5
Greater Nevada Field (Diamondbacks): 91.5
Constellation Field (Astros): 91.2
Dell Diamond (Rangers): 91.1
Las Vegas Ballpark (Athletics): 91.0
Sutter Health Park: 90.6
Smith's Ballpark (Angels): 90.5
Cheney Stadium (Mariners): 90.4
We still see consistency in the league, with a difference of just 1.3 miles per hour between the top and bottom. However, you notice that all of these parks hover almost four miles per hour above the highest number we saw in the Major League data.
That’s a problem for me.
I then went and found all the hitters that had 50+ balls in play in the Pacific Coast League and in Majors between 2021 and 2022 and found their average exit velocities in each league on those balls hit between 70 and 110.
There were 80 hitters in the sample, all 80 of them had higher average launch velocities in the minors than in the Majors. It would be expected for hitters to do worse in the Majors than in the minors since Major League pitches are better and harder to square up, but 100% of the sample showing higher marks in the minors really doesn’t sound like reality to me.
Billy Hamilton put 1,867 balls into play in his Major League career from 2015-2022, with a max launch speed of 106.8. But we see above that in just 60 batted balls in the PCL, he hit one 107.4 miles per hour. That’s not impossible, but it’s incredibly unlikely.
Now maybe there is some explanation for this that I’m missing. Maybe someone will reply to this and explain it to me - if so, I’ll bring that to all of your attention.
I’ve heard some speculation about whether or not it’s a different baseball. I compared 2019 MLB launch velo with 2021 and 2022 and found that the launch velo’s actually came up in 2021-2022. However, they were using a different technology (Trackman) back then, so it’s not an apples-to-apples comparison. It’s unlikely that two different technologies would track the exact same numbers. However, I think it’s likely evidence that the juiced ball would not explain such a big different in launch. The main reason the ball performed better for hitters in 2019 was that the seam height created more drag - that has nothing to do with exit velocity. I think most, if not all, of the effects of a changing baseball would show up while the ball is in flight - not reflected in the exit velocity which is captured almost immediately after contact is made.
What I do know for sure right now is that I’m going to be extremely careful with these minor league exit velo’s in 2023. I will continue to investigate this and do checks after a few weeks of data into the season to see if things have “normalized”. If things in 2023 are roughly matching between MiLB and MLB, then we’ll have something and I’ll be able to talk about the numbers confidently. However, for right now - I’m not going to believe the stuff when people cite a guy’s minor league average or max or whatever else exit velocity.
Pitch Velocity
Some good news here is that pitch velocity is quite reliable. I looked at all of the pitchers that threw fastballs in the PCL and the Majors, and their velo’s all look great. The average difference of fastballs velo’s between the two levels was just 0.25 miles per hour.
Good to know that we can still reliably talk about pitch velocity data from the minors. That will be pretty helpful since we’ll have tons of data on tons of fantasy relevant minor leaguers throughout the year who are jumping up to the MLB from the minors.
It will still be a really fun season tracking AAA Statcast data, but it’s really important to be mindful of the things we’ve discovered here. There are plenty of ways we can get around all of this and still glean some insights from launch velo (compare a player to the other players in his park, etc.) - so we’ll figure some good and fun stuff out eventually.
Again, if I’m missing something obvious - please bring it to my attention.