Advanced Stats Glossary

By Data Cockpit Page

Game Debriefing

Advanced Stats>Games

Used Metrics: Win Probability, Strength States, Expected Goals xG, Goals, Shots on Goal, Unblocked Shot Attempts (Fenwick), Shot Attempts (Corsi)

Time Based Metrics

GP

games played

TOI

Time on ice

Description

How many minutes is the player on the ice per game?

Shift.Length

median (Median) shift length

Description

• What is the typical duration of a player's shift?

• Using the mean would include distortions from unusually short and unusually long shifts, the median is a more robust value facing atypical outliers. 

Shot Based Metrics

Corsi

All shot attempts, goal/on goal/missed/iron/blocked, are called Corsi by most advanced stats users. It can be looked at 

• as a measure of offensive or defensive volume, the sum of Corsi for or Corsi against in a certain setting

• as a differential of two teams, e.g. in a game, Corsi for:Corsi against, much like the often used shot differential (Schussverhältnis).

• as a percentage in a game, over a time span, against a certain opponent, … Typically the Corsi for percentage is calculated.

Description

If 

• Team A had 62 shot attempts in a game against Team B, 

• and Team B had 38 shot attempts,

• then Team A had a Corsi for percentage (CF%, often only "Corsi") of 62%,

• while Team B had a Corsi for percentage of 38%.

But why use Corsi (instead of shots [on goal] e.g.)? As a basic principle in statistics, the more data considered, the more reliable the information. Contrarily, the less data considered, the more uncertainty in the information, the more hazard, the more so-called randomness, the more likely a question is not answered based on evidence (correlation) but on coincidence.

Using Corsi gives a better indication of puck possession (Spielanteile) than using shots on goal. 

CF60 (offensive volume)

Corsi for per 60 minutes

Description

The value is standardized to 60minutes in order to enable comparison.

Shot attempts for generated on average per 60 minutes of play.

While player X is on the ice, shot attempts for generated by his team per 60 minutes.

Typical information on:

• Approximation on how well does a team’s 5v4 unit/while player X is on the ice manage to penetrate the OZ in puck possession?

• How much offensvie volume does a team generate/while player X is on the ice per 60minutes?

CA60 (defensive volume)

Corsi against per 60 minutes

Description

The value is standardized to 60minutes in order to enable comparison.

Shot attempts against generated on average per 60 minutes.

While player X is on the ice, shot attempts against generated by opposition team per 60 minutes.

Typical information on:

• Approximation on how well does a team’s 4v5 unit/while player X is on the ice deny opposition to penetrate the DZ in puck possession?

• How much opposition offensvie volume does a team allow/while player X is on the ice per 60minutes?

CF%

Corsi for percentage

Description

If Team A had 31 shot attempts in a game against Team B, and Team B had 19 shot attempts, then Team A had a Corsi for percentage (CF%, often only Corsi) of 62% [31/(31+19)%], while Team B had a Corsi for percentage of 38% [19/(31+19)%].

If Team A with Player X had 16 shot attempts in a game against Team B, while Player X was on the ice, and Team B had 9 shot attempts, while Player X was on the ice, then Player X had a Corsi for percentage (CF%, often only Corsi) of 64% [16/(16+9)%].

Approximation at 5v5, on how much puck possession did Team A have? Approximation on how much puck possession did Team A have while its Player X was on the ice (=X's Corsi value)?

Fenwick (unblocked shot attempts, Corsi without blocked shots)

All unblocked shot attempts, goal/on goal/missed/iron, are called Fenwick by most advanced stats users. It can be looked at 

• as a measure of offensive or defensive volume, the sum of Fenwick for or Fenwick against in a certain setting

• as a differential of two teams, e.g. in a game, Fenwick for:Fenwick against, much like the often used shot differential (Schussverhältnis).

• as a percentage in a game, over a time span, against a certain opponent, …

FF60 (offensive volume)

Fenwick for per 60 minutes

Description

The value is standardized to 60minutes in order to enable comparison.

Unblocked shot attempts for generated on average per 60 minutes of play.

While player X is on the ice, unblocked shot attempts for generated by his team per 60 minutes.

FA60 (defensive volume)

Fenwick against per 60 minutes

Description

The value is standardized to 60minutes in order to enable comparison.

Unblocked shot attempts against generated on average per 60 minutes.

While player X is on the ice, unblocked shot attempts against generated by opposition team per 60 minutes.

FF%

Fenwick for percentage

Description

If Team A had 31 unblocked shot attempts in a game against Team B, and Team B had 19 unblocked shot attempts, then Team A had a Fenwick for percentage (FF%) of 62% [31/(31+19)%], while Team B had a Fenwick for percentage of 38% [19/(31+19)%].

If Team A with Player X had 16 unblocked shot attempts in a game against Team B, while Player X was on the ice, and Team B had 9 unblocked shot attempts, while Player X was on the ice, then Player X had a Fenwick for percentage (FF%) of 64% [16/(16+9)%].

Approximation at 5v5, on how much puck possession did Team A have? Approximation on how much puck possession did Team A have while its Player X was on the ice (=X's Fenwick value)?

xFSh% ("Expected Fenwick Shooting Percentage=offensive danger)
xFSv% ("Expected Fenwick Save Percentage=defensive danger)

Danger of unblocked shots

Description

xFSh% expected Fenwick shooting percentage. Information on the danger of the unblocked shot for, on average.

The expected shooting (scoring) percentage of unblocked shots. The more dangerous the position/setting, the higher the xFSh% Value.

• is a team’s offensive output (xGF60) more based on volume (FF60) or on danger (xFSh%) or evenly distributed among the two?

xFSv% expected Fenwick save percentage. Information on the danger of the single shot against, on average.

The expected save percentage of all unblocked shots against. The less dangerous the position/setting, the higher the xFSv% Value.

• is a team’s defensive output (xGA60) more based on volume (FA60) or on danger (xFSv%) or evenly distributed among the two?

• Calculation: In a game Team A had 31 Fenwick For (unblocked shot attempts for) and 3.1 xGF  while conceding 45 Fenwick Against (unblocked shot attempts against). It follows that Team A had xFSh% 10% (3.1 xGF/31=0.1=10%) and an XFSv% of 94.7% (1-(2.4/45)=.947=94.7%). It follows that Team B had the respective inverse numbers: xFSh% 5.3%/xFSv% 90%

ZSR

Zone Start Ratio

Description

How often does a player start his shift with an OZ face off?

Over half of all player shifts start with a face off. Let’s say Player X was on the ice for 18 shifts in a game. 8 started on the fly (OTF), 3 started with a face off in his DZ (defensive zone start, DZS), 2 started with a face off in the NZ (NZS) and 5 started with an OZ face off (OZS).

• So that’s 8 OTF, 3 DZS, 2 NZS, 5 OZS.

• ZSR excludes all OTF and NZS, what remains is 3 DZS, 5 OZS.

• ZSR defines the percentage of shift starts with OZ face off compared to shifts starting with DZ face off.

• In this example: 5 OZS / (5 OZS + 3 DZS) = 5 of 8 = ZSR = 62.5 [%]

Attack Scenarios

Rush

off the rush, entering OZ with shot attempt within 5s. Rebounds belong to antecedent attack scenario, even in sequence.

OZ FC

Offensive zone fore-check, gaining puck possession in OZ with shot attempt within 5s. Rebounds belong to antecedent attack scenario, even in sequence.

OZ FO

offensive zone face-off with ensuing puck possession and shot attempt within 5s. Rebounds belong to antecedent attack scenario, even in sequence.

Sust Pos

Sustained offensive zone puck possession, any shot attempt after a team has been in puck possession 5s or more in OZ. Rebounds belong to antecedent attack scenario, even in sequence.

Scope

This glossary aims to provide a definition and an applied example for each value used on this site.

Faceoffs

FO%

Percentage of Face Offs won

Description

Percentage of face offs won / with Player X on ice

Categories for Player Metrics

"On Ice"

• Most variables belong to the category on ice.  

• Values without further denomination belong to the on ice category.  E.g. xGF%.

• The on ice category measures the team values while a player is on the ice.

"Relative"

• Relative values compare the team value of a metric when a player is on the ice compared to the team value of a metric when a player is not on the ice. 

• Relative values carry the initial add on "Rel." E.g. Rel. xGF%: 

Strength States

• Unless otherwise specified, metrics are displayed at 5v5 strength state.

• The most frequent strength state in hockey is by far 5v5. It is followed with some distance by 5v4/4v5.

• All other strength states (6v5/5v6, 6v4/4v6, 6v3/3v6, 5v3/3v5, 4v4, 4v3/3v4, 3v3) are comparatively rare. That means little data and relatively unreliable information (small sample size). Therefore, this server focusses on 5v5, 5v4/4v5 and all strength states. 

Win Probability

Calculated by xGF% of the home team, see example below.

Expected Goals 

Description

Goal probability of unblocked shot attempts (Fenwick) as calculated by the xG Model. 

Is at the base metric for most numeric analysis.

Strength: Includes Volume and Danger.

xG, xGF, xGF60, xGA, xGA60, xGF%

Expected Goals, expected Goals For, expected Goals For per 60min, expected Goals Against, expected Goals Against per 60min, expected Goals For percentage