NHL Edge has changed how teams utilize data. It only recently became available to the public, but NHL teams have had access to it since 2021. The league installed jersey chips that year, but puck tracking wasn’t ready yet, so the first season featured player tracking only.
Now, there’s full tracking across the NHL, and it drives most of the newer analytics.
"We’re not just looking at events," Kruse said. "We’re looking at where everyone is positioned and how each player affects movement across the ice."
For example, in simple terms, if a player has a wide-open net to shoot at from a high-danger area on the power play but misses, that would probably be assigned a 0.95 expected goal. But with a goaltender in the net and square to the shooter, that might be assigned, say, roughly a 25 percent chance, depending on where exactly the shot comes from and who the shooter is.
Essentially, public models may treat situations more generically, but if the goalie is completely out of position or there's a glaring defensive breakdown, that dramatically changes the probability.
"That context matters," Kruse said. "Public models do a good job with event data, but having full spatial awareness of every player on the ice lets us reduce noise and make metrics more predictive year over year. Public models rely on averages; we model the exact scenario. With tracking data, we can better isolate true shooting talent too. It also aligns more closely with what coaches see on the ice."