Better Box Score Metrics: Do Players Score Proportionally to their Yardage Gain? [Pt. 3]

Giving scoring its own weight.

April 23, 2021 by Paul Würtztack in Analysis with 0 comments

The EDGE metric introduced in Parts 1 and 2 includes yardage not just as a key measure of offensive production, but as the only measure of offensive production. No goals, no assists. At the team level, scoring is essentially a function of yardage (and turnovers).

However, if one maintains a yardage-only basis for offensive production at the player level, one is assuming that players score in fairly consistent proportion to their yardage.¹ But do they? Are there players who, for one reason or another, consistently score more or less often than their yardage would suggest? Let’s call this effect “nose for the goal,” which is exactly the kind of sports cliché analysts typically like to use.

For some initial evidence, Figure 1 below shows cumulative yardage plotted against cumulative scoring totals, updated after each of the five Women’s U24 Championship games that we’ve been using, for each of the 25 players. So, the sample size grows with each additional game. My assumption was that the correlation would get consistently stronger as sample size increased. Instead, although the relationship becomes fairly good after three games of data—with yardage explaining about 70 percent of the variation in scoring—it doesn’t budge after that. In other words, yardage alone does not appear to sufficiently explain player scoring rates.

1. Statto enhances this relationship by including end zone yardage in its totals, such that scores come complete with “extra” yardage. For example, a throw from the 1-yard line to 10 yards deep in the end zone is recorded as 11 yards, not 1 yard. ↩