Measuring Performance Against the Power Play: Beyond PK%

The conventional measurement of NHL team performance in short-handed situations is the Penalty Kill Percentage (PK%): how often the team’s penalty killers do their job and prevent the other team from scoring on the power play. 

While PK% is the time honored NHL statistic, does it really provide the best measure of how well a team does in preventing power-play goals?  I say not.

First and most obviously, the PK% statistic does not take short-handed goals into consideration. Clearly, if the penalty killers score a goal, that totally offsets what the opponent’s power play is trying to accomplish, not to mention the huge psychological boost it provides.

There are significant differences among NHL teams in short-handed goal scoring. In the first half of this season, nine teams have scored 5 or more times, while six have 1 or none. This performance spread clearly reflects on the quality of the penalty-killing units, but it's ignored in calculating the PK%.  

Unfortunately, the PK% only considers how frequently the opponent fails to score on the power play. Scoring a short-handed goal often has a bigger impact on the result of a game than a successful penalty kill (there are exceptions, of course).

For all these reasons, at a minimum, the NHL should enhance the meaning of the PK% by incorporating short-handed goals into the statistic.

Secondly and more subtly, PK% doesn’t take team discipline into consideration.  The surest way to prevent power plays goals is not to take unnecessary penalties. If you don’t give your opponent the opportunity, they can’t score on a power play.

While the NHL does not maintain statistics on “good” penalties (e.g., obstructing an opponent to prevent a shot on a partial breakaway) versus “bad” penalties; I think it’s fair to say that the vast majority of short-handed situations are caused by bad penalties: hooks and holds when players have been beaten, careless high sticks and trips, delays of game, too many men on the ice, etc.

Teams that minimize the penalties they take (the New Jersey Devils immediately jump to mind) enjoy a distinct advantage in several respects.

By facing fewer short-handed situations, disciplined teams are at lower risk of giving up power play goals. They are able to follow their game plans more closely, playing more 5-5 hockey in the offensive zone and spreading out ice time in accordance with the coaches dictates.

The PK% is intended to indicate how well a team does against the power play. However, in my opinion, a better overall measurement of team performance in this aspect of the game would be who gives up the fewest power play goals against. 

In other words, instead of focusing strictly on how frequently a team stops the power play, let’s focus on which team surrenders the least to its opponents through the power play. 

We need a statistic that tells more about which team plays the best overall game against the power play. My proposal is "Net Power Play Goals Against Per Game." Admittedly that title is a real mouthful, so instead lets call it "Penalty Kill Yield" (PKY).  The calculation is simple:

PKY = (Power Play Goals Against – Short Handed Goals For) ÷ Games Played

The table below shows the PKY statistic based on games played through January 8, 2010. The data provides some interesting insight into team performance. There is definitely not a one to one relationship between the PK% and PKY stats. 

Several teams with relatively high PK% have actually yielded more net goals on power plays than teams who have much lower PK% rankings.

For example, which team is better at preventing power play goals: New Jersey or St. Louis?  The Blues rank fourth in PK% but are eighth in PKY (0.488). By contrast, the Devils are third in PKY (0.390) although only 10th in PK%.

This difference in PKY stats (0.098) means that even though the Blues have a better PK%, the Devils will give up eight fewer net goals during power plays over the full 82 games season...and as a result will probably earn more than a few extra victories over opponents. 

A few other comparisons are equally revealing. 

Phoenix has a much higher ranking in PK% (ninth) than Detroit (15th). Yet when you compare the two teams based on PKY, the Red Wings (0.558) are 11th while the Coyotes (0.644) are 16th. 

That difference (0.086) equates to seven fewer net goals against on power plays during the course of the season. Advantage Detroit.

Dallas ranks only 25th in PK% but they are 15th in PKY (0.636). In the first half of the season, they have already yielded a net of 5 fewer goals during power plays when compared to Tampa Bay, which ranks 18th in PK% but is 26th in PKY (0.786) and has played two fewer games. 

Even at the top, there is a difference between the two statistics. Chicago (No. 2 on PK%) outperforms Boston (No. 1 on PK%) on the PKY measure by 0.054…while that’s not a big difference, it still means 4-5 fewer goals given up by the end of 82 games. 

These comparisons are not just academic. Over the course of the season, how many times will not yielding a goal mean a regulation victory instead of overtime?  Or taking a game to overtime instead of a one-goal loss?

Special teams are a critical part of the game; both on the power play and the penalty kill. From the standpoint of evaluating overall team performance, what really matters most is how few goals you yield to the opponent’s power play rather than simply what percentage of the time you stop them. 

          Penalty Kill Yield (games thru 1/08/10)

Team	GP	PPG-A	SHG-F	Penalty Kill Yield	PKY Rank	PK%	PK% Rank
CHI	44	20	6	0.318	1	86.8	2
BOS	43	19	3	0.372	2	87.6	1
NJD	41	20	4	0.390	3	83.3	10
BUF	43	21	3	0.419	4	85.5	6
SJS	44	25	6	0.432	5	86.2	3
NYR	44	25	4	0.477	6	86.1	5
CGY	44	27	6	0.477	7	84.2	7
STL	43	25	4	0.488	8	86.2	4
PIT	45	28	5	0.511	9	83.3	11
MIN	44	28	4	0.545	10	82.2	12
DET	43	28	4	0.558	11	81.5	15
ATL	43	31	6	0.581	12	80.9	17
MTL	46	31	2	0.630	13	84.0	8
OTT	44	33	5	0.636	14	81.5	16
DAL	44	31	3	0.636	15	78.3	25
PHX	45	30	1	0.644	16	84.0	9
CBJ	46	34	4	0.652	17	81.9	13
VAN	44	31	2	0.659	18	81.8	14
FLA	44	35	6	0.659	19	79.3	24
NSH	44	33	4	0.659	20	77.2	27
COL	45	34	4	0.667	21	80.2	19
LAK	44	34	3	0.705	22	80.1	20
WSH	43	32	1	0.721	23	80.1	21
NYI	45	38	4	0.756	24	76.8	28
ANA	44	39	5	0.773	25	79.5	23
TBL	42	34	1	0.786	26	80.4	18
PHI	43	38	3	0.814	27	79.7	22
CAR	43	41	5	0.837	28	78.3	26
EDM	44	37	0	0.841	29	76.1	29
TOR	45	50	3	1.044	30	68.4	30