What a power play actually looks like: three assumptions, measured
Three things get said about power plays so often that nobody checks them: that a first unit is three forwards and two defencemen, that a club has a first unit and a second unit, and that the first unit is five particular players. Measured against shift-chart membership across 2024-25 and 2025-26, all three are wrong.
Across two seasons and 32 clubs, the number whose most-used five-man first unit carried two defencemen is 0. The first unit takes 77.2% of five-man power-play ice time rather than half. And the median club ran 18 different five-man groups through that first-unit role, its most-used one accounting for 30.1% of the minutes.
1. Nobody runs two defencemen
The phrase describes a formation that has left the league. In 2025-26 31 of 32 clubs built their most-used first unit around exactly one defenceman and 1 used none at all; 2024-25 reads the same way. Four forwards and a point is not a wrinkle, it is the default, and an argument about which two defencemen belong on a top unit is an argument about a unit nobody ices.
| season | no D | one D | two D | clubs |
|---|---|---|---|---|
| 2024-25 | 2 | 30 | 0 | 32 |
| 2025-26 | 1 | 31 | 0 | 32 |
2. The two units are not halves
Because the first unit is a single job rather than a shift rotation, it keeps most of the ice. League-wide it took 77.2% of five-man power-play time, and the club that shared most evenly still kept 63.5% for its first group. That is what makes a second-unit promise smaller than it sounds: on a club with average power-play time, the second unit's quarterback is playing about a minute a night.
| club | PP1 | PP2 |
|---|---|---|
| WSH | 94.2% | 5.8% |
| EDM | 93.7% | 6.3% |
| TBL | 88.1% | 11.9% |
| VGK | 88.1% | 11.9% |
| DAL | 85.5% | 14.5% |
| OTT | 69.8% | 30.2% |
| SEA | 68.8% | 31.2% |
| CGY | 67.2% | 32.8% |
| PHI | 65.9% | 34.1% |
| STL | 63.5% | 36.5% |
3. “PP1” is a role, not a lineup
The last assumption is the one that survives longest, because a depth chart has to print five names. Over a season those five names change constantly — injuries, matchups, a coach chasing a goal. The median club sent 18 distinct five-man groups out on its first unit, and the group it used most held only 30.1% of the minutes. MTL was the league's most settled at 77.1%; STL the least, at 12.6%.
| club | top five | groups |
|---|---|---|
| MTL | 77.1% | 7 |
| BOS | 64.2% | 15 |
| VGK | 64.0% | 8 |
| DAL | 54.8% | 8 |
| EDM | 52.6% | 15 |
| CHI | 16.9% | 27 |
| COL | 15.6% | 17 |
| PHI | 15.4% | 35 |
| FLA | 13.7% | 32 |
| STL | 12.6% | 39 |
Window: 2024-25 and 2025-26, regular season, five-man units only — rows for three- and four-man units are overlapping-penalty situations and are excluded. This is a two-season finding, not a ten-season one, and the reason is worth stating: unit membership is this engine's own derivation from shift charts, second by second, because the league publishes power-play time but not who was on the ice for it. Where the rest of this site can reach back ten seasons on official totals, nothing outside these two carries who played with whom. A club's "most-used first unit" is the five-man group with the most first-unit ice time, not a coach's stated lineup. The build fails if any player on any top unit cannot be resolved to a position, because an unresolved id would count as "not a defenceman" and manufacture the first finding here. See how this site measures.
All research · This season’s power-play leaders · Ten seasons of PP goals
Know the morning it changes
This article is the record. The daily report is the change: who moved onto PP1 or off it, the morning after it happens. The daily report is free through Tuesday 13 October: every power-play change in the league, in your inbox the next morning. No card, no password. After that, $27 keeps it daily for the whole season, or you drop to the free Monday email.
8 games measured · 0 quarantined · every threshold published