clock menu more-arrow no yes mobile

Filed under:

Micro stats: Tracking possession using individual events

While everyone talks about possession's effect on the game, we rarely delve too deep into what creates possession. Micro stats are perhaps the best way to break down the game at a deeper level.

Andy Marlin-USA TODAY Sports

Tracking individual puck-possession events is both time-consuming and tedious. That said, the payoff is exclusive data that can be applied to the accurate evaluation of multiple aspects of player performance. I have been tracking passes, dekes, shots, blocked shots, blocked passes, stick-checks, dump-ins, and many, many other events for the better part of the last four years. I track the majority of these events by documenting them as either successful or unsuccessful, while events such as zone exits, line carries, and zone-entries are simply tracked as they happen. This system provides us with upwards of 1200 points of data per-team per-game, and upwards of 5500 points of usable data per-player per-season.

As a result of this work, I have data that can be used similarly to the data being tracked in the NBA using SportsVU. I have borrowed some of my hockey metrics from the NBA; including among others; expected goal value, expected shot value, and scoring-chances per-100 possession-events.

Another metric that I developed uses the events tracked to determine how much time each player is spending either with possession of the puck in the offensive-zone, or defending in the defensive-zone. Offensive-zone possession time is calculated by determining the percentage of total events (both offensive and defensive) that are spent with possession of the puck in the offensive-zone. Simply put, the calculation looks like:

= (events with possession in o-zone/total offensive and defensive events (all 3 zones))

The higher the percentage, the more opportunity each player has had to produce offense in the offensive-zone.

Defensive-zone defending time is calculated by determining the percentage of total events (both offensive and defensive) that are spent attempting to remove puck-possession from the opposition in the defensive-zone. Simply put, the calculation looks like:

= (defensive-events in the defensive-zone/total offensive and defensive events (all 3 zones))

The higher the percentage, the more opportunity the other team has had to create offense while that player was on the ice.

This metric is similar to Corsi and Fenwick; except in the fact that it tracks individual players rather than teams. Corsi and Fenwick do a good job of quantifying team possession. Counting shot attempts, especially over the course of entire season can help us recognize which teams are driving possession, and which teams are likely to struggle down the road.

That said, the advantage to the metrics I'm working on is that the results are specific to each player, and based on their performance; not necessarily on the performance of their teammates.

These two metrics have shown to accurately reflect both territorial advantage and scoring-chance dominance. Numbers taken from the nearly 1000 games I have tracked show that teams that out-chance the opposition at even-strength contribute over 29.5% of their total events with possession of the puck in the offensive-zone, while teams that are out-chanced contribute 26.5% of their events with offensive-zone possession. Similarly, teams that out-chance the opposition at ES use up only 15% of their total events attempting to remove possession from the opposition in the defensive-zone, while teams that are out-chanced contribute 16.6% of their events defending in the d-zone.

In terms of individual players, these two metrics also help as quantify both offensive and defensive-players. Offensive-players contribute a substantially higher percentage of their events with possession in the offensive-zone, while defensive-players (particularly defensive-defensemen who struggle moving the puck out of the defensive-zone) contribute a higher percentage of their overall events attempting to remove possession from the opposition in the defensive-zone.

Among the teams who I've tracked a minimum of 500 even-strength minutes, players with a high percentage of offensive-zone possession events (relative to an average player at their position) include Max Pacioretty, David Desharnais, P.K. Subban, Phil Kessel, Nazem Kadri, Dion Phaneuf, Bobby Ryan, Mats Zuccarello, Viktor Stalberg, Tyler Seguin, Jordan Eberle, and Taylor Hall.

Players with a high percentage of defensive-zone defending events (relative to an average player at their position) include Douglas Murray, Dion Phaneuf(!), Jay McClement, Jarred Cowan, Mika Zibanajed, Joe Girardi, and Brian Boyle.
An advantage will always exist in tracking individual puck-possession events to accurately evaluate multiple aspects of player performance. A true competitive advantage will develop for those teams that are able to evaluate players before the competition; that advantage will show up in player procurement, player trades, and drafting.