Better Box Score Metrics: The Possessions Problem and WUL Weeks 5 & 6

Evolving our per 100 possession stats to the next level.

April 25, 2026 by Paul Würtztack in Analysis

Welcome to Better Box Score Metrics, our analytics column focused primarily on semi-pro statistics and discovering new ways to use data to explore ultimate. It is anchored by EDGE, which we can summarize as converting goals, assists, blocks, turnovers, and yards gained into one metric, while accounting for the scoring and turnover environment.

Introducing xE, xEO, and xEB

Before we get to the WUL leaderboards and full player data, we need to introduce yet another modification to how we’re doing things around here. For six years, we have been using EO100 and EB100 as our per-possession rate stats. It has worked pretty well in terms of its main function of pooling offensive and defensive opportunities across O-lines and D-lines so that we can calculate separate offensive-production and block-production rates.

But there has been a tension in equating one possession to one opportunity, particularly in the EDGE framework that allocates one goal-equivalent per goal and takes as a starting point that the cost of a turnover depends on the context. Given that turnovers and possessions are essentially the same thing, the value of a possession should also be fluctuating to represent the game environment.

Extreme events can bring some of these tensions into sharp relief. In strong sustained winds, Michigan’s 8-7 win over Brown at the 2026 Florida Warm-Up contained approximately 151 possessions. EDGE itself can effectively deal with these conditions because the cost of a turnover becomes extremely low relative to the value of yardage and scoring. But a rate metric that uses possessions as a fixed constant becomes meaningless in this context.

The simplicity of a fixed constant is very appealing, but we don’t think we should continue to defer to simplicity.