Skip to content

Glossary

A plain-language guide to the terms wRink uses.

23 terms
01

Player contribution

The wins a skater adds, and where those wins come from.

Wins above replacement

WAR

A player’s estimated contribution beyond a depth replacement, expressed in wins.

On wRink
Compare overall contribution over the games measured; read the components to see where it comes from.
Keep separate
Replacement is a lower bar than average NHL performance; WAR and wins above average use different baselines.

Wins above average

WAA

Estimated contribution relative to average performance, expressed in wins.

On wRink
Read the skater components and their sum before the replacement adjustment.
Keep separate
Average is a higher bar than replacement. Adding one replacement adjustment to total WAA gives overall WAR.

Replacement level

The performance of depth players used as the starting point for WAR.

On wRink
Understand what a regular contributes beyond the depth a team turns to when it runs short.
Keep separate
wRink measures forwards and defencemen separately; replacement is below average.

WAR per 82 games

WAR/82

WAR divided by the games covered, multiplied by 82.

On wRink
Pool WAR and covered games across the last three seasons, then multiply by 82. A skater needs at least 60 covered games for a listed recent rate.
Keep separate
This is a contribution rate in completed games, not a forecast. Goalies use WAR per 60 covered games, with a 30-game minimum.

Covered games

The games in which our records can measure a player’s contribution.

On wRink
WAR per 82 and goalie WAR per 60 divide by covered games, and season rows count only those games.
Keep separate
Covered games can be fewer than games played. An uncovered game is unmeasured, not a game with zero contribution.

Regular forwards and defencemen

The group an article ranks a skater against: forwards or defencemen with at least 500 five-on-five minutes in a season, or 1,500 across a three-season window.

On wRink
Read article standings in steps of 5%, such as top 10% or bottom 25%, against regular players at the same position.
Keep separate
Player-page ratings use current NHL skaters at the same position with at least 60 covered games across three seasons. Different groups and windows can give different standings.

Component value

The contribution from one part of a skater’s game, measured in wins above average or converted into dollars.

On wRink
See how five-on-five play, finishing, special teams and penalties contribute to the total.
Keep separate
Components are measured against average. One replacement adjustment is added to their sum to obtain WAR; component dollars use the same current cap scale as overall WAR $.
02

Contracts and cap dollars

How past contribution becomes dollar context for contracts.

WAR $

A skater’s past WAR converted into salary-cap dollars at one league-wide rate per win.

On wRink
The rate gives a player the same share of teams’ skater cap spending above minimum salaries as his share of league WAR, so WAR $ ranks players exactly as WAR does. Season tables show WAR $; player listings show WAR $/82.
Keep separate
WAR $ prices past play on today’s cap scale. It is not a salary, contract estimate or forecast; cap hit remains a separate contract fact.

Cap hit and AAV

AAV

Cap hit is the annual salary-cap charge. AAV means average annual value: the contract’s total value divided by its number of years.

On wRink
Establish the annual price, which team carries it and when the deal begins.
Keep separate
A player’s cash salary in one year can differ from his cap hit.

Price-implied WAR

The WAR corresponding to a price on a declared dollar scale.

On wRink
Divide the price by the stated dollars per WAR; identify the benchmark and its purpose.
Keep separate
This is arithmetic context for a price, not a prediction or a whole-contract valuation.

Uncapped WAR $

WAR $ is not limited by an individual salary ceiling.

On wRink
A player’s WAR keeps its full signed dollar equivalent, even above any salary he could be paid.
Keep separate
It prices past contribution. It is not an amount a team can pay.

Current cap dollars

Past performance expressed on one dollar scale for today’s salary-cap environment.

On wRink
Compare seasons from different years without mixing their original spending environments.
Keep separate
Repricing a past season changes its dollar value, not the player’s WAR or what he was paid.
03

On-ice performance

The shot, chance and special-teams language behind the model.

Expected goals

xG

The estimated scoring chances of a set of attempts, found by adding each attempt’s probability of becoming a goal.

On wRink
Compare the danger of chances rather than relying only on the shot count.
Keep separate
Expected goals describes opportunity, not goals owed. Compare actual goals and xG over the same attempts.

Process

A skater’s estimated contribution to creating and limiting dangerous chances at five-on-five.

On wRink
Read offence and defence separately, then their net contribution, with adjustments for teammates, opponents and game context.
Keep separate
Rates describe contribution per unit of ice time; totals also reflect minutes played. Finishing and special teams are separate components.

Finishing

A skater’s estimated contribution from scoring above or below the expectation for his chances, after adjusting for uncertain shooting results.

On wRink
Separate converting chances from creating the opportunities in the first place.
Keep separate
It is not simply goals minus expected goals. The adjustment does not prove that the removed part was luck or that the retained part will repeat.

Goals above expected

G−xG

Actual goals minus expected goals from the same attempts.

On wRink
Describe how far scoring finished above or below the expectation for those chances.
Keep separate
This raw gap differs from the adjusted finishing contribution. Matching the goal and xG samples is essential.

Goals saved above expected

GSAx

Expected goals against minus goals allowed on matching shots; positive means the goalie allowed fewer goals than expected.

On wRink
wRink adjusts the estimated goalie effect toward average before calculating the reported goal contribution. Its GSAx estimate can therefore differ from the raw subtraction.
Keep separate
GSAx is measured in goals, not wins or dollars. Read it alongside the shots faced and period covered; the total alone is not a ranking of ability.

Special-teams value

Estimated contribution relative to average on the power play or penalty kill over the minutes a player played.

On wRink
wRink measures five-on-four and four-on-five play using the balance of expected goals at both ends. Totals reflect performance and workload.
Keep separate
Power-play and penalty-kill percentiles compare accumulated value within the same position group among players meeting that unit's deployment threshold. Little deployment is different from average performance or missing data.

Penalty differential

The estimated value of the penalty opportunities a player earns for his team, minus those he concedes to the opposition.

On wRink
Drawing calls adds value; taking them subtracts it. We account for assessed duration and offsetting penalties, then convert the balance to goals and wins.
Keep separate
Earning a power play is separate from performing well on it. This uses an average opportunity value, not the outcome of each ensuing power play.

Shrinkage

Pulling an uncertain estimate toward a reference level, with less pull when individual evidence is stronger.

On wRink
Explain why wRink’s adjusted finishing contribution differs from raw goals above expected.
Keep separate
Shrinkage changes the estimate, not the recorded goals. It does not identify exactly which part of a result was luck.
04

Uncertainty

Limits of estimates, and why unusual results may not repeat.

Central estimate

A single number summarising an estimated contribution.

On wRink
Read it with the player, period and measure it describes.
Keep separate
A point estimate does not establish that uncertainty is small or supply a supported range.

Measurement range

An interval describing uncertainty about a player’s estimated contribution in completed games.

On wRink
Distinguish uncertainty about past contribution from the spread of possible future outcomes.
Keep separate
It does not include everything that could change next season. We do not currently publish supported measurement or forecast ranges.

Regression toward average

The tendency for unusually high or low results to be followed by less extreme ones when temporary influences do not repeat.

On wRink
Explain why an unusually strong or weak result may not repeat in full.
Keep separate
It does not mean an opposite result is owed or that a player cannot improve. Deliberately blending a record with a reference is a modelling choice, not the definition of the tendency itself.