Power play statistics

[dubgeek]

Does anyone track power play effectiveness by goals per minutes (or seconds) on the PP?

It has always seemed odd and ineffective to me that everyone simply tracks it as goals per PP opportunities. By this method a PP that nets a goal after 1:55 of PP time weighs the same as a PP that nets a goal after :30.

When a team gets a man back on the ice after being 4 on 4 and gets a brief :30 or even :15 second power play but fail to score that weighs the same as going a full 2:00 with no goal.

With the move toward more advanced stats, shouldn't we start tracking power plays differently to more accurately evaluate a teams' PP efficiency?

For example, team A and team B have each had 5 minor penalty power plays for a potential maximum PP time of 10:00 minutes. They each scored goals on three of their power plays tying for a 60% PP rating.

However, team A scored their goals at :40, :40, and :50. Team B scored their goals at 1:10, 1:20, and 1:30. Team A has a PP effectiveness of 3 goals per 1:30 or 1 goal for every 30 seconds of PP time. Team B has an effectiveness of 3 goals per 4:00 or 1 goal for every 80 seconds of PP time.

In this analysis team A is the more effective PP team by far. Perhaps over a season things balance out, but at the very least tracking power plays by goals per time on the PP would smooth out the impact of the partial power plays that result from non-coincidental minors.

The same analysis could be applied to penalty kills in which you'd want to have a large number of minutes or seconds for each goal scored against your PK units.

Am I the only one that sees value in this? Is anyone out there tracking it already?

Click to read more...

 

[Bernier the Boats]

Does anyone track power play effectiveness by goals per minutes (or seconds) on the PP?

It has always seemed odd and ineffective to me that everyone simply tracks it as goals per PP opportunities. By this method a PP that nets a goal after 1:55 of PP time weighs the same as a PP that nets a goal after :30.

When a team gets a man back on the ice after being 4 on 4 and gets a brief :30 or even :15 second power play but fail to score that weighs the same as going a full 2:00 with no goal.

With the move toward more advanced stats, shouldn't we start tracking power plays differently to more accurately evaluate a teams' PP efficiency?

For example, team A and team B have each had 5 minor penalty power plays for a potential maximum PP time of 10:00 minutes. They each scored goals on three of their power plays tying for a 60% PP rating.

However, team A scored their goals at :40, :40, and :50. Team B scored their goals at 1:10, 1:20, and 1:30. Team A has a PP effectiveness of 3 goals per 1:30 or 1 goal for every 30 seconds of PP time. Team B has an effectiveness of 3 goals per 4:00 or 1 goal for every 80 seconds of PP time.

In this analysis team A is the more effective PP team by far. Perhaps over a season things balance out, but at the very least tracking power plays by goals per time on the PP would smooth out the impact of the partial power plays that result from non-coincidental minors.

The same analysis could be applied to penalty kills in which you'd want to have a large number of minutes or seconds for each goal scored against your PK units.

Am I the only one that sees value in this? Is anyone out there tracking it already?

NHL.com provides a nice breakdown:

http://www.nhl.com/ice/teamstats.ht...ll&sort=powerPlayGoals&viewName=powerPlayTime

 

[dubgeek]

NHL.com provides a nice breakdown:

http://www.nhl.com/ice/teamstats.ht...ll&sort=powerPlayGoals&viewName=powerPlayTime

Nice that they have those numbers.

So, I plugged that into a spreadsheet and did the calculations to determine PP goals per seconds of PP time which I'm calling PPEff (Power Play Effectiveness). No major differences in the rankings. Most teams are within 1 ranking position in PPEff to where they are in PP%. A couple of teams are 2 or 3 positions better or worse. I still like the idea of tracking PPEff as it feels more accurate.

DET remains tops scoring one PP goal for every 365 seconds of PP time.

BUF remains worst with one PP goal every 881 seconds - that's 14+ minutes of PP time for each PP goal , and it's even worse than that. I did an additional calculation where I accounted for short handed goals against by subtracting them from PP goals scored (maybe there's a better way to account for them, but this was easiest). Factoring in BUF's 8 SHGA puts them at a whopping 1586 seconds (26+ minutes) of power play time per PP goal.

Oh, and don't put NJD on the 5 on 3. They're scoring 1 goal for every 66 seconds of 5 on 3 time. The closest to them are SJS and STL at 102 and 106.

 

[eklunds source]

Nice that they have those numbers.

So, I plugged that into a spreadsheet and did the calculations to determine PP goals per seconds of PP time which I'm calling PPEff (Power Play Effectiveness).

I like where you're headed, because the way the NHL tracks it...

- If you take a penalty 10 seconds into a powerplay, you are 0/1
- If you get trapped in your own zone, while up a man, for 120 seconds, you are 0/1
(these things clearly are not the same)

- If you score one goal on a double minor penalty, you are 1/2
- If you score one goal on a 5-on-3, you are 1/2
(clearly, teams should be expected to convert far more often on 5v3s)

However, one issue you're going to run into is that goal scoring on powerplays is not predictive of future goal scoring on powerplays.

How does a powerplay sustain success?

While many hockey analysts appear to obsess over raw conversion rates, the same above principle – that shot generation is far more predictive of future success (or failure) than shooting percentage – is vital in forecasting future success.

Why? Because shooting percentage, even in a power-play situation where scoring chances are aplenty, is not a very repeatable skill. It’s easy to illustrate this by simply grabbing a team’s shooting percentage over the first half of the season, and finding how well it correlates to the team’s shooting percentage over the second half of the season.

Detroit is a great example: here's their 5v4 performance, comparing this year to last (it's important to separate 5v4 from 5v3, because the conditions on ice in those 2 situations are very different, and teams cannot control how often they get 5v3 advantages):

All stats are adjusted to "per 60 minutes of 5v4"

Year|Goals|Shots|Fenwick|Corsi|Shooting Percentage
'14-'15|9.54|53.7|76.1|100.6|17.76
'13-'14|6.03|52.9|73.2|92.7|11.40

The difference in shots on net between the two seasons is 1 extra shot every 37.5 powerplays (assuming a full 2 minute powerplay) - it's up 1.5%. None of their shot metrics vary by more than 8%, but their goal total is up over 50%.

Detroit clearly has a "pretty good" powerplay - they're top-10 in all of those metrics - but it would be ill-advised to bet on their shooting percentage to continue.

They're definitely the most "effective" powerplay - they're creating goals at a rate faster than any team in the league - but at the same time, that's a descriptive statistic, not a predictive one.

TLDR; powerplay performance is dominated by shooting percentage, which is not a repeatable skill over the minuscule sample sizes that we get.