Counting Rugby Steps: A New Way to Understand Possession Value

Researchers have come up with a fresh way to look at possession values in rugby league. They used something called a Bayesian Mixture Model to create a smooth pitch map and guess the odds of different results from possession. This study used loads of data from the 2021 Super League season, with 99, 966 observations. Instead of focusing on fixed zones, they picked 33 points on the field (30 in the playing area, 3 near the opposing try line). Each point told a story about five possible outcomes of a possession: a converted or unconverted try, a penalty, a drop goal, or no points at all. These probabilities were then spread across the whole pitch using this special model.The key takeaway? They figured out a way to estimate the expected possession value (EPV) by combining the likelihood of each outcome with the points it could earn. This method gave a smooth surface of the pitch, showing varying odds of outcomes and EPVs in every spot. To make it even more interesting, they compared how different teams' attacking and defensive strategies looked on this map. Plus, they developed a new rating system for players, measuring actual against expected performance. This approach offers way more flexibility than older zone-based methods. And the best part? It can be tweaked for other sports that have similar data.