Athleticism Metric: Setting the Ground Work

With so much sabermetric work already completed, I’m creating a ton of work for myself to see if a hitter’s athleticism influences how they age? Additionally, do these “athletes” age better? I tried to jump the gun a few nights ago with an ill-fated Twitter thread where I thought about reverse-engineering the stats. Instead, I’m going to put a value on a hitter’s athleticism using some readily available metrics.

I began my search by using some advice from Bill James who commented on my Tweet.

He just rattled the traits off and since he’s likely forgotten more about baseball than I’ve ever learned, I’ll just focus on them. I’m guessing he’s already investigated the subject.

Compiling this information even a few years ago would have included more guesstimating but the available StatCast data simplifies the process. The first two factors, power and speed, are easy to measure using max exit velocity and sprint speed. Agility and quickness are a little tougher to nail down, but I settled on the defensive component of WAR.

The data is/are far from perfect. The first adjustment I made was to remove the pitch-framing element from catchers. I kept all the other elements including the positional and fielding adjustment. If a player can only play first base (one hit to his athleticism), he better do a great job of it. Additionally, I adjusted the fielding value to a per-game rate so all hitters are on a level playing surface since defensive WAR is effectively a counting stat.

I did consider adding a couple of factors. One was height. I read up on the ideal height for any athlete and there seems to be a link between height and injury. I hoping to use height, but in the next step. It’s going to be one of several factors.

The other input that I’m still unsettled on is a hitter’s ability to swing a bat and make contact (Contact%). After a few private debates, I think this trait falls more in the hand-eye coordination basket. I can be talked into being wrong and flip-flop on my decision.

So to get my athletic value, I took all the hitters from the past five seasons (length of available StatCast data) with a minimum of 200 PA and calculated the z-score for the three traits. I weighted each trait evenly for now. And again, I could be talked into weighing one trait more or less than the others.

For a final value, I add the three values together. At some later date, I might set these values on a 100 scale (with 100 is league average) or possibly the 20-80 scouting scale. I’ll nail down the formula first before moving on.

It’s time to start doing the idiot check. Here are the top-10 players from the past two seasons (full list).

Top-10 in Athleticism
Rank Name Season
1 Francisco Lindor 2018
2 Mike Trout 2018
3 Harrison Bader 2018
4 Aaron Judge 2018
5 Matt Chapman 2018
6 Jorge Alfaro 2018
7 Trea Turner 2018
8 Giancarlo Stanton 2018
9 Kevin Kiermaier 2018
10 JaCoby Jones 2018
1 Byron Buxton 2019
2 Aaron Judge 2019
3 Javier Baez 2019
4 Harrison Bader 2019
5 Matt Chapman 2019
6 Trevor Story 2019
7 Jorge Alfaro 2019
8 Adalberto Mondesi 2019
9 Ronald Acuna Jr. 2019
10 Mike Trout 2019

The two names that seem out of place are Stanton and Judge. Both are highly rated based on three traits. These two examples are why I want to jump ahead to include height showing how big bodies breakdown. I just need to tap the breaks and get athleticism down first.

Here are the bottom ten from the two seasons (worst is listed first).

Bottom-10 in Athleticism
Rank Name Season
1 Victor Martinez 2018
2 Albert Pujols 2018
3 Kendrys Morales 2018
4 Brian McCann 2018
5 Luis Valbuena 2018
6 Justin Smoak 2018
7 Tony Kemp 2018
8 Daniel Murphy 2018
9 Mark Reynolds 2018
10 Yonder Alonso 2018
1 Albert Pujols 2019
2 Brandon Belt 2019
3 Yonder Alonso 2019
4 Justin Smoak 2019
5 Miguel Cabrera 2019
6 Domingo Santana 2019
7 Isan Diaz 2019
8 Luis Arraez 2019
9 Kendrys Morales 2019
10 Neil Walker 2019

While not a household name, Tony Kemp seems out of place. Just digging into his inputs, he has almost no power with average to below-average speed and defense. Maybe Kemp is the type of player this process is hoping to find.

So readers, what seems off? Please scrutinize the full list and nitpick everything. I’ll remain active in the comments and can quickly create new rankings if needed for comparison. Once I get athleticism done, I can then start investigating if athletic hitters age better or not.

We hoped you liked reading Athleticism Metric: Setting the Ground Work by Jeff Zimmerman!

Please support FanGraphs by becoming a member. We publish thousands of articles a year, host multiple podcasts, and have an ever growing database of baseball stats.

FanGraphs does not have a paywall. With your membership, we can continue to offer the content you've come to rely on and add to our unique baseball coverage.

Support FanGraphs




Jeff, one of the authors of the fantasy baseball guide,The Process, writes for RotoGraphs, The Hardball Times, Rotowire, Baseball America, and BaseballHQ. He has been nominated for two SABR Analytics Research Award for Contemporary Analysis and won it in 2013 in tandem with Bill Petti. He has won three FSWA Awards including on for his MASH series. In his first two seasons in Tout Wars, he's won the H2H league and mixed auction league. Follow him on Twitter @jeffwzimmerman.

newest oldest most voted
Werthless
Member
Member
Werthless

I don’t know if Tony Kemp is the type of player you’re looking to find. You might just be finding weak or injured hitters! Your model would be most useful if it identifies good players who are high risk. For example, how does Freddie Freeman compare to Anthony Rizzo? Why is the projected 2022 WAR differential between the two 30 year old first basemen higher than the projected 2020 WAR differential, and is this a decline-related prediction for Rizzo?

My suggestion:
1. Perhaps change defensive value to an aspect of the defense that you think matters, like range.
2. We’re getting ahead of ourselves perhaps, but have you looked at which factors are correlated with injury or decline? That may help you select features for further investigation (as in my 1st suggestion).
3. I know you’re looking for predictors of decline, but perhaps the best predictor of future decline is a recent decline in a subset of these metrics. It may answer a different question (Which player has started his decline?) than what you originally set out to answer (How will X player decline based on what we know about him?), so you may not decide to pursue it. Or, does the recent Trout speed score decline portend bad things for him? That’s a potentially very interesting question!

brentdaily
Member
Member
brentdaily

I like the range recommendation a lot.

My initial ‘yeah but’ while reading is that the distinction between agility and quickness seems awfully slight. But range (or the full OAA) could address both.