Creating RAW_Adjusted
My first opportunity in creating and detailing a new way to judge a hitter’s hit tool for college players.
A phenomenal way to gather information and detail a player’s development is by creating and using adjusted stats. Over the past couple of years, Stuff+ has been at the forefront in pitcher evaluations. Without going into too much detail, Stuff+ uses pitch characteristics such as velocity and movement to predict swing and miss, ultimately translating into stuff+. I wanted to figure out a way to do something for hitters, trying to find hitters who are due for positive or negative regression. There is already OPS+ which is an easy way to see how a player performs in the league environment. Inspired by OPS+, I took a similar approach to the stats I mentioned above to create additional statistics to assist in player evaluations. Below, I showcase adjusted stats that I created: HR_Power_Power, QOC (Quality of Contact), Plate_Discipline, Contact_Plus, and RAW_Adjusted_Plus. They are all scaled to 100, so if a player has a 190 in anything then they are 90% better than the league average, if grade out to a 80 in anything then they are 20% worse than the league average. These stats are as of May 17th.
What is RAW_Adjusted
RAW_Adjusted is the dependent variable combining the independent variables HR_Power, QOC, Plate_Discipline_Score, and Contact_Score. I selected statistics such as Z-Contact% (In-Zone Contact%) and EV95+LA10–30% (Exit Velocity 95 MPH or more with a Launch Angle between 10 and 30%) to create the individual variables and found the standard deviations for these individual variables to assist in creating their RAW scores. After acquiring the RAW score, I divided it by the league average RAW score, then multiplied it by 100 to scale it so that 100 is average.
Condon vs Crews
Here is how last year’s 2nd overall pick, Dylan Crews compares to this year’s potential 1st overall pick, Charlie Condon in these 5 statistics along with their OPS+:
It would be disingenuous not to mention that Dylan Crews was a far better fielder than Condon while also controlling a valuable position in centerfield with all of the ability to stay there. We don’t entirely know if Condon will stick in Right field or move to third; the jury is still out. In the batter’s box, Condon clears Crews in all aspects, except for Plate_Discipline_Score.
2024 Prospects
Here are Travis Bazzana, and some other notable first, or borderline-first round prospects and their scores. This is where the “eye-test” becomes important. Potential first overall pick, Charlie Condon, has slugged the most home runs in the BBCOR Era, so it makes sense that he has a 391 HR_Power. Two-way star Jac Caglianone’s Plate_Discipline_Score stands out since he is in contention for a top-5 pick despite the below-average score. Caglianone’s Plate_Discipline_Score suffers from a chase issue, chasing 37.6% of the time, which is actually a career low. It is something I mentioned in late March in my Mock Draft 1.0. After seeing these player’s RAW_Adjusted it’s difficult seeing players who I was “bullish” on like Vance Honeycutt and Kaelen Culpepper, yet they were graded at the bottom two of these 14 position players. I believe Honeycutt can field his way out of his potential struggles at the plate and land himself a home in the middle to late first round. Culpepper might be more suited in the 2nd round. Players like James Tibbs and Christian Moore continue to rise.
What’s Next?
To make this even better, the next step in my process is adjusting it by park factor. Some parks traditionally are tabbed as “Hitter-friendly”, “Pitcher-friendly”, or “Neutral.” We see this in the majors and in the college ranks. Being able to adjust the stat for a hitter who registers more plate appearances at Oregon compared to a hitter who registers more plate appearances at East Carolina would be valuable. Additional aspects I want to look into include seeing what MLB organizations are drafting position players by their RAW_Adjusted and comparing them to previous drafts, as well as seeing if this can be a predictive stat and how well it performs.
Data from TruMedia
Thank you for reading!
The link to the code will be up soon on my Github: Here ; but will be available sooner upon request!
You can also find this on Twitter/X: @EthanMann02