Glossary

TERM/CONCEPT DESCRIPTION
“Corsi” A measure of the differential for all shot events during a certain game state (with 5v5 carrying the most analytical weight given that the bulk of every game is played in this state) and period of play within a match or over the course of any number of games. The formula, in short, is shot attempts for minus shot attempts against. Spelled out, it is as follows:
[(goals scored by a team – goals scored by the opponent) + (attempts by the team saved by the opponent’s goalie – opponent attempts saved by the team’s goalie) + (attempts by the team that missed the opponent’s net – opponent attempts that missed the team’s net) + (attempts by the team blocked by the opponent – opponent attempts blocked by the team)]
A team or player-specific (team performance with the player on or off the ice) Corsi number can be expressed either as a plus-minus, percentage or rate (per minutes played). For example, consider the following shot event and time on ice data:
PLAYER
(0 goals for – 0 goals against) + (10 attempts for saved – 2 attempts against saved) + (2 attempts for missed – 4 attempts against missed) + (6 attempts for blocked – 2 attempts against blocked) during 15:00 5v5 play
= 18 shot events for to 8 shot events against
= +10 Corsi, 69.23% Corsi%, 40.00 Corsi per 60:00 5v5 TOI
TEAM
(0 goals for – 2 goals against) + (24 attempts for saved – 17 attempts against saved) + (5 attempts for missed – 10 attempts against missed) + (12 attempts for blocked – 11 attempts against blocked) during 50:00 of 5v5 play
= 41 shot events for to 40 shot events against
= +1 Corsi, 50.62% Corsi%, 1.2 Corsi per 60:00 5v5 TOI
The Corsi metric is named for the current goalie coach of the Buffalo Sabres who initiated the idea of tracking all shot attempts and serves as a proxy for possession (the ability to “drive play”), among other things. With an ample sample size and byaccounting for other factors (e.g. usage, score effects), it also possesses significant evaluative and predictive value. Particularly at the team level, Corsi data has proven to be a reliable, albeit imperfect, indicator of regular-season and playoff success.
“Event” A term primarily used in reference to any shot attempt (one that scored, was saved, missed the net or was blocked) or particular actions such as draws (also called a “start”), zone entries and zone exits but can be used to describe any activity that can be observed and counted for statistical purposes.
“Fenwick” A measure of the differential for most shot events during a certain game state (with 5v5 carrying the most analytical weight given that the bulk of every game is played in this state) and period of play within a match or over the course of any number of games. The formula, in short, is shot attempts for excluding those blocked minus shot attempts against excluding those that were blocked. Spelled out, it is as follows:
[(goals scored by a team – goals scored by the opponent) + (attempts by the team saved by the opponent’s goalie – opponent attempts saved by the team’s goalie) + (attempts by the team that missed the opponent’s net – opponent attempts that missed the team’s net)]
A team or player-specific (team performance with the player on or off the ice) Fenwick number can be expressed either as a plus-minus, percentage or rate (per minutes played). For example, consider the following shot event and time on ice data:
TEAM
(0 goals for – 2 goals against) + (24 attempts for saved – 17 attempts against saved) + (5 attempts for missed – 10 attempts against missed) during 50:00 of 5v5 play
= 39 shot events for to 39 shot events against
= +0 Fenwick, 50.00% Fenwick%, +0.00 Fenwick per 60:00 5v5 TOI
PLAYER
(0 goals for – 0 goals against) + (10 attempts for saved – 2 attempts against saved) + (2 attempts for missed – 4 attempts against missed) during 15:00 5v5 play
= 12 shot events for to 6 shot events against
= +6 Fenwick, 66.67% Fenwick%, +24.00 Fenwick per 60:00 5v5 TOI
The Fenwick metric bears the last name of the man who introduced the revised Corsi figure. It similarly serves as a proxy for possession (the ability to “drive play”), among other things. With an ample sample size and by accounting for other factors (e.g. usage, score effects), it also possesses significant evaluative and predictive value. Particularly at the team level, Fenwick data has proven to be a reliable, albeit imperfect, indicator of regular-season and playoff success.
“Possession” A term that (obviously) refers to a player or team in the act of controlling a puck but also is used to describe the ability of a team or player to “drive play”. In the absence of officially tracked zone time, possession is primarily gauged with shot attempt data. Corsi and Fenwick figures, in effect, serve as proxies.
“PDO” The sum of both on-ice shooting and save percentage for a team or player (while he was on the ice); a metric, so named for the internet commenter that devised it, that regresses heavily in the long-run (to a combined 1000, 100% or 1.000) given the year-to-year variance in shooting and save percentages and is widely considered an indicator of how relatively lucky (PDO > 1000) or unlucky (PDO < 1000) a team or player has been.
“Quality of Competition”
“Regression”
“Sample Size” The number of observations used for some evaluative purpose. For any metric, larger samples of events (typically data sets that run over multiple seasons) reduce the “noise” (unexplained variation, randomness) that often plague small samples and can lead to erroneous conclusions, especially when it comes to power play numbers, traditional +/-, save percentage and shooting percentage, given its year-to-year volatility. Simply put, larger sample sizes allow for greater accuracy, and therefore more confidence, in dependent analysis.
“Scoring Chance” A clear play directed towards the opposing net from a dangerous scoring area loosely defined as the top of the circle in and inside the faceoff dots (the so-called ‘Home Plate’, here’s a visual). Missed attempts count but only shots blocked shots where a defenseman is “acting like a goaltender” are registered. Under some circumstances, shots from just outside the dangerous area may also be counted as chances, such as when there is a screened goaltender or quick puck movement. By tracking how many or few were generated by a team or while a player was on the ice, scoring chances, which are known to have been tracked at the NHL level since 1980, provide additional context beyond shot attempt metrics (which strongly correlate) and serve as a highly valued indicator of performance for a given game or season. This number is most often shown as a +/- figure with both the + and – figures individually rather than adding them together like traditional goal-based +/-. (For example, Victor Hedman was +7/-2 in scoring chance differential last night.)
“Score Effects” A term used to describe what happens mostly to possession data when the score of a hockey game is not tied, particularly late in the game. A team with a lead tends to play risk-averse hockey or go into a defensive shell, since they are trying to protect a lead. Conversely, the team playing from behind is more likely to take risks in an all-out attempt to tie the game. For this reason, teams facing a deficit late in the game tend to have an inordinate share of Corsi and Fenwick events, which can skew proper analysis. For this reason, data samples are often filtered down to “Close” data — when a game is tied, or with a one-goal deficit in the 1st or 2nd periods — to eliminate score effects.
“Usage”
“WOWY”
“Zone Entry”
“Zone Exit”
“Zone Start”