Team power
How many goals per game better or worse is this team than an average team?
How it works
Start with the chances
At five on five, we look at a team's share of expected goals: the quality of its chances compared with the chances it allows. We adjust for the score so the game situation is part of the comparison.
Use the lineup that is here now
Early in the season, a few games cannot tell us much. We blend this season's play with last season's player ratings for the current lineup, weighted by minutes and adjusted for age. That track record counts as 20 games of evidence: about 80% of the five-on-five part after 5 games, and 50% after 20.
Add special teams, leave out hot streaks
Power-play goals minus goals allowed on the penalty kill add another piece, per game. We give that piece more weight as games accumulate. At five on five, finishing and goaltending luck are left out: the rating uses chances there, not goals. Special teams use actual goals, so a hot power play does count.
The formula
The formula
Rating = blended five-on-five contribution + weighted special-teams contribution, in goals per game versus average.
Lineup share of the five-on-five blend = 20 ÷ (games + 20). Current-season play supplies the rest. The lineup uses last season's player ratings, weighted by minutes, with age adjustment k = 0.01.
Special teams = [(PP goals − PK goals against) ÷ games] × [games ÷ (games + 35)].
How we tested it
| Check | Finding |
|---|---|
| After 5 games | Model 0.64 · points % 0.27 |
| After 10 games | Model 0.68 · points % 0.35 |
| After 20 games | Model 0.69 · points % 0.45 |
| After 41 games | Model 0.69 · points % 0.64 |
| Separate weighting comparison, 5 games | Fixed 80/20 this-season weighting: 0.49 versus 0.62 |
The early advantage is why the lineup's track record fades gradually. The separate weighting comparison is a different test from the checkpoint rows.
What does correlation mean?
Correlation, written as r, describes how closely two sets of results move together. A stronger positive relationship means better results in one tend to go with better results in the other. A value near zero means little relationship. It is not a percentage of predictions that were correct.
What it can't tell you
This is an estimate of team quality, not a promise about the next result. At five on five it deliberately leaves out the hot shooting or goaltending that can decide games and move the standings.