Recently, the Man Behind the RotoViz Curtain posted this article and also this article about the relationship between scouting and advanced statistics, noting that, in the end, all analysts end up “talking about some kind of objective measure, and it will probably be a simple counting stat like yards or touchdowns.”
Here is the part of the first article that I want to use as a starting point: “I think that the advanced stats crowd would find a lot of common ground with the tape grinding crowd if we could all just figure out how to record what’s being watched. [. . .] Today I just wanted to establish that at least on a very important part of the player evaluation discussion, it’s likely that in the end we’ll be talking about numbers.”
I agree. I think that the tape grinders know that numbers are important, and they merely process them in a different way. For instance, Matt Waldman, in his excellent yearly publication, The Rookie Scouting Portfolio, assigns numerical significance to qualities that have often seemed unquantifiable. In fact, I imagine that for many tape grinders, the process of quantifying what they know they can quantify seems both 1) needless—why should they have to calculate and put down as a number what they can so obviously see?—and 2) laborious—why should they have to spend the extra time to calculate and record numbers that they themselves do not even need? In effect, like people who close their eyes at a musical performance so they can all the more intently experience the music, most tape grinders understandably let the performance of a player on tape wash over them, and the numbers that they could process and report they instead internalize as a composite impression—not “That running back gets X degrees of leverage when blocking and so he is able to drive back blitzers Y yards Z% of the time,” but “That guy is good at picking up the blitz.”
Still, I believe that for the field of scouting ultimately to progress significantly, the thorough quantification of what scouts see will be necessary, especially because many of the methods of measurement we use now are imprecisely predictive and representative.
For example, two RBs identical in height, weight, build, Speed Score, Agility Score, vertical and broad jump, collegiate statistics, collegiate conference, and collegiate offensive system could still be vastly different runners—and collectively the scouting community can explain those differences, at present, by relying primarily upon qualitative statements: “This guy is a pounder; that guy is a slasher.” What do these statements mean numerically? How can they be used to project a player’s future performances? When we can start using quantifiable statements, we will all of us be better at what we do.
For RBs, what are the qualifications that need to be quantified?
Let me start with this—to me, RBs are like pitchers. Two pitchers can be identical in all their physical attributes, and what will constitutively distinguish them from each other are these two traits: 1) the ability to command one’s location of the ball and 2) one’s decision process in determining when to throw certain pitches. In baseball, we can speak quantifiably about a pitcher’s ability to throw with accuracy into certain areas in (and out of) the strike zone, and we can make certain predictive statements about the pitcher’s potential based on how often he hits and misses his spots and where the ball goes when he misses his spots.
Further, we can speak quantifiably about a pitcher’s decision process in approaching hitters, some of whom do well against some types of pitches and some of whom don’t—and we can make these statements because we track both what kind of pitch a pitcher delivers on each offering and what a hitter does each time he receives a pitch. Because of the existence of these numbers, we can make value judgments about a pitcher’s performances and predictive declarations about his future performances when certain conditions apply.
We need to be able to do the same for RBs. We need to know, quantifiably, how good an RB is at location—how good he is at consistently finding and hitting with speed the crease, the right crease, or the cutback lane. Also, we need to know, quantifiably, how good he is at making decisions, not just taken in isolation but sequentially structured across a series of plays—how good he is at determining, at optimal moments, when to follow the design of a play and when (and how) to deviate, when to run up the middle and when to bounce a run to the outside. Does he sometimes not bounce a run outside when he should? Or, rather, does he throw too many curveballs and not enough (well-located) heaters?
Without the answers to these questions and without the numbers that alone can provide these answers—and without the many other numbers that we don’t even know yet that we need—scouting will be, for all its progress, a pseudo-science of unfortunate inexactitude.
In pushing scouting toward its future, the tape grinders and stat nerds must complement each other. In a way, they exist to serve each other in this endeavor. Without the knowledge and skills of both parties, neither will achieve its potential, nor will the field of scouting as a whole.
To improve, we must seek to quantify what most people take to be unquantifiable—a player’s quality. A running style, just like the color blue, can indeed by described with numbers. Although those numbers mean nothing to most people, in the hands of some people those numbers represent a world of knowledge waiting to be acquired and applied. Until we can quantifiably explain in a more nuanced way both a running back’s decision to reverse field on a run designed to the left and the wisdom of that decision, our analysis of RBs, their skill, and their probable futures will be—at best—problematically incomplete.