Loading...
The foundational shot metrics in hockey analytics. Before xG existed, these were how you measured which team was actually controlling the game.
TL;DR
Corsi counts all shot attempts (on goal + missed + blocked). Fenwick counts unblocked shot attempts only. Both are measured at 5v5 and expressed as a percentage — above 50% means you're out-possessing the opponent.
Includes everything — shots on goal, missed shots, and blocked shots. The largest sample size of any shot metric, which means less noise game to game.
Removes blocked shots — since blocking is a defensive skill, Fenwick isolates the attacking team's output more cleanly. Slightly smaller sample but slightly purer signal.
The relationship isn't perfect in any single game — hockey has too much variance for that. But over a full season, teams with sustained CF% above 52-53% almost always finish above .500.
More shot attempts means more zone time, more scoring chances, and less time defending. Shooting and save percentage fluctuate year to year, but CF% tends to be sticky and repeatable.
Example
A team finishes with 54.2% CF% despite a losing record. Their save percentage collapsed due to goalie injuries. The underlying possession says they're better than their record — expect a bounce-back when the goaltending stabilizes.
They're not competing — they answer different questions. The best teams rank well on all three.
Corsi counts all shot attempts — shots on goal, missed shots, and blocked shots — at 5-on-5. CF% above 50% means a team is out-attempting their opponents, which correlates strongly with winning over a full season.
Fenwick removes blocked shots from the equation. Because the defending team influences how many shots get blocked, some analysts consider Fenwick a cleaner measure of offensive zone pressure. In practice, they tell similar stories.
They each answer a different question. Corsi and Fenwick measure shot volume and puck possession. xG measures shot quality. The strongest teams rank well on all three. PuckCast uses all three, weighted by their predictive value.