The special teams index adds unrelated units
Somewhere in the early analytics era hockey adopted a convenience: add a team’s power play percentage to its penalty kill percentage, call the sum the special teams index, and treat anything over 100 as good. It is a tidy number, it fits in a graphic, and it is used every spring to explain playoff series. It is also an addition of two quantities that share a name and nothing else. The two percentages describe different rosters performing different tasks against different opposition, computed over denominators that neither unit sets, and summing them produces a figure whose units are not meaningful.
Two different teams
The players on a first power play unit and the players on a first penalty kill unit overlap very little. Power play personnel are selected for puck skill, vision, and shooting; penalty killers for positioning, stick detail, and willingness to block shots. Many of the best killers never see the man advantage, and several of the most productive power play specialists are deliberately kept off the ice shorthanded. Adding the two success rates describes an organization, not a group, which is fine as long as nobody then uses the sum to argue about a particular unit — which is the only way it ever gets used.
The denominators are not comparable
Power play percentage is goals scored divided by opportunities. Penalty kill percentage is opportunities survived divided by opportunities faced. One is a scoring rate, the other a prevention rate, and the two are not symmetric in difficulty or in variance. League-wide, power plays convert somewhere around a fifth of the time and kills succeed around four fifths of the time, so a one-point move in each is worth wildly different amounts of information. Adding a number that lives near 20 to a number that lives near 80 means the sum inherits almost all of its movement from the first term.
Neither unit controls how often it plays
Both percentages have opportunity counts in the denominator, and opportunities are handed out by officials. A team that draws few penalties has a small power play sample; one that takes few has a small kill sample. Early in a season these samples are tiny enough that percentages swing enormously, which is why November special teams rankings look nothing like April ones. The index inherits the instability of both components at once.
There is a strategic wrinkle too. A disciplined team that rarely gets shorthanded faces its kill in unusual, often high-leverage circumstances, and a team that takes constant penalties builds a kill with enormous repetition. Neither pattern is captured by a success rate.
The 100 threshold that gives the index its appeal is also arbitrary. There is no reason a combined figure of exactly 100 should mark the boundary between adequate and good; it marks the point where two unrelated percentages happen to sum to a round number in base ten. League-average special teams has sat slightly above 100 in most recent seasons, which means the threshold everyone treats as a bar to clear is roughly the median. Half the league clears it every year by definition.
Shorthanded goals count in the wrong place
A shorthanded goal is scored by the penalty killers and charged against the power play percentage of the team with the advantage. That is defensible bookkeeping and it means the two halves of the index are not independent: an aggressive kill that generates chances lowers opposing power play numbers across the league. Empty-net goals complicate the kill side at the end of games. The sum treats all of this as settled.
Four-on-four and three-on-three situations sit outside the framework entirely. Coincidental minors produce open-ice stretches that belong to neither unit in the bookkeeping, and overtime is now played at a man count that no special teams percentage describes. Teams increasingly staff those situations with dedicated personnel. The index has no slot for any of it.
Quality of opposition disappears
Special teams percentages are rarely adjusted for who the opponent was. Facing an elite power play eight times in a division will depress a kill percentage for reasons unrelated to the kill, and a team in a weak division gets flattered. Across an eighty-two game schedule the effects partially even out; across a seven-game series, which is exactly when the index gets quoted most, they do not even slightly.
Playoff hockey compresses every one of these problems. Series are short, officiating standards tighten, and a team may draw only two or three power plays a night, which means a percentage computed over a round rests on a dozen opportunities. Analysts reach for the index precisely when its sample is smallest, then use it to explain an elimination. A measure that needs a season to mean anything is being asked to describe seven games.
What to use instead
The per-sixty framing fixes most of this. Goals for per sixty minutes of power play time and goals against per sixty of penalty kill time put both units on the same scale, remove the opportunity-count problem, and let you compare a team that gets eight minutes a night with one that gets four. Expected goals per sixty on each unit goes further by addressing shot quality. Both are available in public data, both are more stable, and neither requires adding a scoring rate to a prevention rate.
Goaltending is the hidden third party in the kill number. A penalty kill percentage is substantially a save percentage on shorthanded shots, and a team with an excellent goaltender will field a kill that rates well regardless of how its skaters defend. That is the whole argument of the separate case against penalty kill percentage, and it carries straight into the index: half the sum is partly a goaltending measurement, and the other half is not.
How to read it on StatLine
StatLine’s NHL tables carry ice time alongside production, which is the raw material for the per-sixty view. If a special teams claim is being made about a team, look first at whether the units are even getting meaningful minutes and at which players are absorbing them. A power play that ranks well on a handful of opportunities a night is a small sample wearing a percentage, and the ice-time column is what tells you which case you are in.
One further caution: the index is often compared across seasons, and league-wide special teams rates drift as rule enforcement changes. A season in which officials call more minors produces larger samples, different percentages, and a different league-average index. A team whose number improved while the league’s improved more has gotten relatively worse, and the raw figure says the opposite.
The honest read
The special teams index is arithmetic performed on two numbers that do not belong in the same equation. It combines different personnel, different tasks, different baselines, and denominators set by officials, then presents the total as a property of a team. The underlying units matter enormously — special teams decide playoff series, and the people quoting the index are right about that much. They are simply using a measure that cannot support the argument. Per-sixty rates say the same thing more honestly and in numbers that mean something individually.