Better Box Score Metrics: The Cost of Turning it Over Just Went Up [Pt. 2]

Using EDGE to describe a player's individual performance.

April 15, 2021 by Paul Würtztack in Analysis with 0 comments

In Part 1 of this series, I described how one could characterize game outcomes with simple math functions based only on yardage and turnovers, a happy consequence of ultimate’s structural lack of complexity. Those functions became the basis of a new metric: efficiency-derived goal equivalents (EDGE). Now we start adapting it for the assessment of player performance.

First, I need to offer several caveats. The first is that there is nothing about EDGE that is going to solve the problem of having no defensive stats other than blocks.¹ In addition, we have no qualitative information. For example, a completion is a completion, regardless of whether it broke the mark or required a massive layout. For the moment, all of those qualitative attributes are assumed to be proportionally distributed in the countable stats, but we don’t know if that’s actually true. With additional qualitative coding of game play, we can try and figure out to what extent more precise allocation of these attributes changes the story.

The other main caveat is that no person’s individual stats are independent—your ability to hit open receivers on a point may be enhanced not just by how good your cutters are, but also by your teammates’ missed deep shots on previous possessions that has loosened up the defense for unders. Individual stats assume that these indirect effects are reasonably distributed, but it’s always good to keep in mind that player stats fit in an ecology of overall team performance.

Let’s begin looking at how we can get this new EDGE statistic to assess individual players’ game performances.

1. I use blocks instead of “Ds” to avoid confusion with defense and D-lines ↩