One of the most important parts of a scouting report is the pro comp. The pro comp helps paint a picture of a prospect to someone who hasn’t seen that player firsthand, in addition to gaining an understanding of the future expectations for that player. I would argue that pro comps have increasingly more importance as the availability of data decreases the further down the level of play one is evaluating.
As I have started to create my own reports, I have found that actually creating a pro comp is one of the most difficult parts. Some of that is because I need to do a better job of watching players from more than just a few teams, and some of it is because I don’t always realize what it is that I’m comparing. We can make physical comparisons, mechanical comparisons, skill-set comparisons, make-up comparisons, and combinations of all of those. Where I have struggled is assigning a certain grade to a player, and having different area scouts I've gotten to know look over my reports and say something like, “you gave Player A a future Hit grade of 60. That means you think he’ll be as good as Kris Bryant” and sometimes I hesitate and reconsider my grade because no, I don’t think he’ll be as good as Kris Bryant.
What I set out to do with this project, is define the underlying stats behind each grade on the 20-80 scale, and assign grades to current MLB players based on those stats. Then, amateur players (actually, more like all non-major leaguers) who get evaluated on the 20-80 scale get a list of the most comparable players at the MLB level. I see this as more of a quality assurance tool than anything. A scout would use this and grade out prospects as normal, then they are asked before turning in a report, “the most comparable MLB players to Prospect A are X, Y and Z. Do you want to stick with your evaluation?” Maybe they do, maybe they don't.
There are a few ways in which I think this can be useful. I took the players on Fangraphs Top 100 Prospects Board and assigned them each a current pro comparison. I have only done this for hitters so far, the pitcher comps are going to be more complex but I will complete them eventually. Below is an example of the results for a few top prospects. These comparisons were made based on future Hit, Power and Run grades. I initially included defense as a combination of the Arm and Field tools, but removed it for various reasons that I will touch on in the end of this post.
As you can see, the pro comp is not based on the position of the player at all. That is something I would like to add in the future. It's also important to note that these comps are based solely on 2019 stats. Comparing Jarred Kelenic to 2019 Ryan Braun is much different than comparing him to 2011 Ryan Braun.
Below is a snapshot of a more broad cheat sheet that can be used, based on future Hit and Power grades. The use for this would be for a scout to give a prospect future grades of, for example, 60 Hit and 50 Power, look at this chart and say something along the lines of, "I actually don't think this guy will be as good as Jorge Polanco," or "Jorge Polanco is a perfect comp."
Criteria for Creating Comps
For those interested (about to get a lil nerdy) in what exactly went into this, I only used three of the five tools. I initially wanted to combine the Arm and Field tools using UZR/150 to create an all encompassing "Defense" tool because it is a stat that embodies characteristics of both fielding ability and arm strength. I removed it for several reasons; 1) in my experience, comps are primarily based around offense and I did not want a perfect defensive match to take away from what is a better offensive comp, 2) catchers are not given UZR/150 values which would have thrown off my scale, and 3) the dataset I had for UZR did not have ID numbers to index defensive data.
The Hit tool grades are based on Batting Averages, which I know might be frowned upon by some reading this. I read a few different articles on previous research to try and determine the best stat to use, and found varying answers with nothing very clear-cut. I didn’t want to include stats that include measures of power ability, and I didn’t know how I should properly weigh plate discipline factors with batted ball performance. Batting average could easily be substituted for, but it’s what I went with.
For the Power tool, I went with ISO as my evaluation metric. Once again, someone could make the case that another stat could be a better measure, but I’m sticking with it.
Lastly, the Run tool is based on runner splits from contact to first base. I toyed with the idea of using stolen bases, or Statcast sprint speed, but liked the idea of splits because typically, scouts grade speed on the home-to-first time.
I assigned the grades to major league position players from the 2019 season that had at least 300 at-bats using a normally distributed bell curve. For those who do not know, that is why the scouting scale is 20-80. 50 represents the MLB average, and every change of 10 “points” represents a standard deviation away from the mean. Additionally, I did not use the grades of “25” or “75” because I view those grades as marginal and they lack conviction. Below is what the scale looks like in mathematical terms.
And here are the "actual" MLB grades from the 2019 season.
Limitations
I think the biggest thing to consider is that the definitions of each tool are not completely cut and dry. Talk to five scouts, and you’ll probably get five different definitions for each tool. The best way to improve this system would be to clearly define an actual statistic or combination of stats that embody each tool. Many scouts also assign grades based on a player’s game power, in addition to his raw power. That would need to be considered in this system, but scouting departments could easily make adjustments as to how they define each tool.
As I mentioned before, a player comp is typically created from a variety of things such as physical appearance and skillset. This system is more specific to how a scout expects a player to perform in the future statistically, as opposed to their physical appearance/mechanics, which is something that would need to be clearly understood by those reading the reports.
There may also be some unrealistic comparisons generated if, for example, an MLB player has a down year. It’s possible that a prospect could compare to 2019 Joey Votto from a numbers standpoint because it was a relatively “down” year for Votto’s standards. Hearing that the comparison you made is similar to Joey Votto would be pretty alarming at first, given that he is a future Hall of Famer. This would mean that 1) the comparisons need to be taken with a grain of salt, and 2) I need to come up with a better system where Votto’s name would not appear as a comparison in a down season.
My comps were based solely on grades, not an overall future value or role. That is yet another area to improve this system.
Lastly, it likely won't be soon, but it is possible that Hawkeye and PitchAI could allow us to make pro comps based on biomechanics reports. Combining all of the comp systems would seem to be the holy grail to me, but then again, not a lot of prep prospects will be moving quite like a big leaguer.
Thanks Camden Kay (@k_camden) for putting some finishing touches on my code to find the player comps. Check out his Pitcher comp tool based on Trackman data.
Comments