Reading hockey stats: plus-minus, Corsi, save percentage and goals against average

TUREFF• •0

Box scores in hockey are full of abbreviations. Some numbers have been around for decades, while others came from the analytics movement and now appear on broadcasts. This guide explains what the most common stats mean, how they are calculated and where each one can mislead you.

The basics: goals, assists and points

Goals (G) are credited to the last player of the scoring team to touch the puck.
Assists (A) go to up to two teammates who touched the puck before the scorer, without the other team gaining control in between.
Points (P) are goals plus assists. A primary assist is the last pass before the goal; a secondary assist is the one before that.

Points are the headline stat for forwards, but secondary assists can be quite random, so many analysts focus on goals and primary assists when comparing players.

Plus-minus

Plus-minus (+/-) tracks the goals scored while a player is on the ice.

A player gets plus one when their team scores at even strength or while shorthanded.
A player gets minus one when the opponent scores at even strength or on its own shorthanded goal.
Power-play goals do not count for the scoring team, and a power-play goal against does not count against the players killing the penalty.

Plus-minus is easy to understand but noisy. It depends heavily on teammates, goaltending and luck, so a good player on a weak team can post a poor number. Use it over a long stretch and alongside other stats, never on its own.

Shot attempts: Corsi and Fenwick

Analysts noticed that teams who control the puck tend to win over time, and that shot attempts are a decent proxy for possession.

Corsi counts all shot attempts: shots on goal, missed shots and blocked shots.
Fenwick is the same but leaves out blocked shots.
Corsi For percentage (CF%) is the share of all attempts taken by a player's team while he is on the ice. Above 50 per cent means his team had more of the attempts.

These figures are usually quoted at five-on-five, because power plays and penalty kills distort the numbers. They say nothing about shot quality, which is why many sites now also publish expected goals (xG), a model estimating how likely each attempt was to score based on things like distance, angle and shot type. Different sites use different models, so their xG figures will not always match.

Save percentage

Save percentage (SV%) is the main measure for goaltenders.

Divide saves by shots on goal faced.
A goaltender who faces 30 shots and allows two goals has made 28 saves, a save percentage of .933.

It is written as a decimal to three places. Save percentage depends partly on the quality of shots allowed by the defence, which is why analysts also look at goals saved above expected, comparing goals allowed with the expected goals of the shots faced.

Goals against average

Goals against average (GAA) is the number of goals a goaltender allows per 60 minutes of play.

Multiply goals allowed by 60 and divide by minutes played.
Empty-net goals do not count against the goaltender, because he was not in the net.

GAA reflects the whole team's defence as much as the goaltender. A goaltender behind a team that gives up few shots can have a low GAA with an ordinary save percentage, so save percentage is usually the better guide to the goaltender himself.

Other numbers worth knowing

TOI (time on ice) shows how much a coach trusts a player, and on which special teams he plays.
Faceoff percentage (FO%) is the share of draws a player wins, mainly relevant for centres.
PIM (penalty minutes) totals the minutes a player spent in the box.
Shooting percentage (SH%) is goals divided by shots on goal. A player far above his usual rate is often riding luck that will not last.

Using stats sensibly

Prefer larger samples; a ten-game stretch can say almost anything.
Compare like with like, such as five-on-five numbers against five-on-five numbers.
Combine traditional and analytical stats rather than picking one side.
Keep watching the games, because numbers miss context such as injuries and linemates.

Which stat do you trust most when judging a player? Share your view in the comments.