By The NumbersHockey Analytics... the Final Frontier. Explore strange new worlds, to seek out new algorithms, to boldly go where no one has gone before.

This is a fairly theoretical question, but I'm interested in hearing if anybody has any thoughts.

What is the value of (say) 1 minute of penalty killing, expressed in terms of points?

In other words: let's say Player 1 average 1:00 per game killing penalties. Assume he's an average penalty killer (we're not talking about Ed Westfall or Jere Lehtinen here). How many points (goals/assists) is this worth?

Another way of looking at it: Player A scores 80 points and doesn't kill penalties. Player B scores X points and spends 1:00 on the PK per game. How many points does Player B need to score to bring equivalent value to his team?

This is a fairly theoretical question, but I'm interested in hearing if anybody has any thoughts.

What is the value of (say) 1 minute of penalty killing, expressed in terms of points?

In other words: let's say Player 1 average 1:00 per game killing penalties. Assume he's an average penalty killer (we're not talking about Ed Westfall or Jere Lehtinen here). How many points (goals/assists) is this worth?

Another way of looking at it: Player A scores 80 points and doesn't kill penalties. Player B scores X points and spends 1:00 on the PK per game. How many points does Player B need to score to bring equivalent value to his team?

Depends on other factors that have to be considered in the answer.
Specifically how many penalties the team is required to kill per game? How successful the team is killing penalties, not only actual opportunities but penalty time killed as well? How B killing penalties allows A to have an advantage at even strength or on the PP.

This is a fairly theoretical question, but I'm interested in hearing if anybody has any thoughts.

What is the value of (say) 1 minute of penalty killing, expressed in terms of points?

In other words: let's say Player 1 average 1:00 per game killing penalties. Assume he's an average penalty killer (we're not talking about Ed Westfall or Jere Lehtinen here). How many points (goals/assists) is this worth?

Another way of looking at it: Player A scores 80 points and doesn't kill penalties. Player B scores X points and spends 1:00 on the PK per game. How many points does Player B need to score to bring equivalent value to his team?

Canadiens1958 is right that more specifics are needed for any particular case. But I'll take a quick and dirty run at this, using league averages.

For the 2010 season, the spread of team penalty killing in preventing goals is about 30 goals. So lets say that the average team is about 15 goals above replacement. In points, that's about 40 points. (almost 3 points per goal, keeping it Q&D).

The average team spent over 500 minutes killing penalties last year. That's about 2500 player-minutes total, including goaltenders. Assuming all positions and players contribute equally, we have about 40 points equal to 2500 minutes played. Therefore we have about 60 shorthanded minutes per point. Your hypothetical 1 SHTOI/G forward deserves 1.3 points in credit for his PK time, on an average penalty killing team.

Another way to look at it would be to consider the trade-off between playing SH minutes and ES minutes. If a forward plays 4 minutes a game on the penalty kill, how many fewer ES minutes will he play? How many points is that worth? How much did the coach value his shorthanded contributions to sacrifice those potential ES minutes and points? Unfortunately while I like the theory on this, as NHL coaches should know the value of a player's penalty killing as well as anyone, it's probably not possible to say how many "scoring minutes" a player lost by killing penalties.

This is a fairly theoretical question, but I'm interested in hearing if anybody has any thoughts.

What is the value of (say) 1 minute of penalty killing, expressed in terms of points?

In other words: let's say Player 1 average 1:00 per game killing penalties. Assume he's an average penalty killer (we're not talking about Ed Westfall or Jere Lehtinen here). How many points (goals/assists) is this worth?

Another way of looking at it: Player A scores 80 points and doesn't kill penalties. Player B scores X points and spends 1:00 on the PK per game. How many points does Player B need to score to bring equivalent value to his team?

One way of looking at it would be to assume that Player 1 is giving up even strength ice time to kill penalties. Therefore, just take the difference between Player 1's Even Strength points per minute and his Short Handed points per minute, and then multiply by the number of games played. If you want to add something else, you could try to factor in how many more goals would be allowed with a "below average" replacement penalty killer as well as how many less goals for.

That's a nice and simple way to do it. Unfortunately, what C1958 said is a much more realistic answer.

Last edited by Hawkey Town 18: 08-19-2010 at 10:26 PM.

Canadiens1958 is right that more specifics are needed for any particular case. But I'll take a quick and dirty run at this, using league averages.

For the 2010 season, the spread of team penalty killing in preventing goals is about 30 goals. So lets say that the average team is about 15 goals above replacement. In points, that's about 40 points. (almost 3 points per goal, keeping it Q&D).

The average team spent over 500 minutes killing penalties last year. That's about 2500 player-minutes total, including goaltenders. Assuming all positions and players contribute equally, we have about 40 points equal to 2500 minutes played. Therefore we have about 60 shorthanded minutes per point. Your hypothetical 1 SHTOI/G forward deserves 1.3 points in credit for his PK time, on an average penalty killing team.

Another way to look at it would be to consider the trade-off between playing SH minutes and ES minutes. If a forward plays 4 minutes a game on the penalty kill, how many fewer ES minutes will he play? How many points is that worth? How much did the coach value his shorthanded contributions to sacrifice those potential ES minutes and points? Unfortunately while I like the theory on this, as NHL coaches should know the value of a player's penalty killing as well as anyone, it's probably not possible to say how many "scoring minutes" a player lost by killing penalties.

Should have been more specific. PK numbers very often are insufficient when it comes to looking at how effective certain teams and players are killing penalties.

Biggest weakness is that "time killed" is not expressed in the stats.
Example two teams. Each is assessed 250 single minors generating 500 minutes of potential penalty time. Each gives up 35 goals on the PK. Stats tend to reflect them as being equally efficient. However if the first team on average killed 1:30 of the penalty before giving up a goal while the second team on average killed 1:00 of the penalty before giving up a goal then the first team is more efficient.

Situational and detailed analysis are also rather weak. 2-men short, reducing a 2-man to a 1-man before giving up a goal, first goal of the game, just a number of areas that could use deeper analysis.

Canadiens1958 is right that more specifics are needed for any particular case. But I'll take a quick and dirty run at this, using league averages.

For the 2010 season, the spread of team penalty killing in preventing goals is about 30 goals. So lets say that the average team is about 15 goals above replacement. In points, that's about 40 points. (almost 3 points per goal, keeping it Q&D).

The average team spent over 500 minutes killing penalties last year. That's about 2500 player-minutes total, including goaltenders. Assuming all positions and players contribute equally, we have about 40 points equal to 2500 minutes played. Therefore we have about 60 shorthanded minutes per point. Your hypothetical 1 SHTOI/G forward deserves 1.3 points in credit for his PK time, on an average penalty killing team.

Very nice work for a Q&D analysis. I did some very rough calculations too and am getting around 2 points per PK time, on average. Initially I suspected that penalty killing would be a bit more valuable but that's probably my preconceptions.

This wouldn't work at the extremes for elite penalty killers (who contribute more per minute than average penalty killers) but it's a reasonable starting point.

Quote:

Originally Posted by overpass

Another way to look at it would be to consider the trade-off between playing SH minutes and ES minutes. If a forward plays 4 minutes a game on the penalty kill, how many fewer ES minutes will he play? How many points is that worth? How much did the coach value his shorthanded contributions to sacrifice those potential ES minutes and points? Unfortunately while I like the theory on this, as NHL coaches should know the value of a player's penalty killing as well as anyone, it's probably not possible to say how many "scoring minutes" a player lost by killing penalties.

I should have specified if I meant a player getting say 3 minutes of PK time instead of ES/PP time, or in addition to the ice time he already received.

For example, if we assume that players generally trade away ES/PP time for PK time, it would probably be detrimental to put elite scorers on the PK unless they were exceptional penalty killers.

Very nice work for a Q&D analysis. I did some very rough calculations too and am getting around 2 points per PK time, on average. Initially I suspected that penalty killing would be a bit more valuable but that's probably my preconceptions.

This wouldn't work at the extremes for elite penalty killers (who contribute more per minute than average penalty killers) but it's a reasonable starting point.

I should have specified if I meant a player getting say 3 minutes of PK time instead of ES/PP time, or in addition to the ice time he already received.

For example, if we assume that players generally trade away ES/PP time for PK time, it would probably be detrimental to put elite scorers on the PK unless they were exceptional penalty killers.

Penalty killing is the most physically demanding of the various game situations which may be reduced to ES/PP/PK.

A team's PP usually features the elite scorers who tend towards the higher TOI minutes. The optimum from a PK standpoint is having the lower TOI players neutralize the higher TOI players.

During the 2009-10 season Atlanta did slightly worse using Chris Thorburn amongst other secondary players on the PK - allowing 57 goals in 320 PK situations than Pittsburgh did using Crosby, Malkin and Staal allowing 52 goals in 327 PK situations. At 2 minutes PKTOI per game per elite player over a full season close to 500 minutes of elite time that should definitely generate better results.

Very nice work for a Q&D analysis. I did some very rough calculations too and am getting around 2 points per PK time, on average. Initially I suspected that penalty killing would be a bit more valuable but that's probably my preconceptions.

This wouldn't work at the extremes for elite penalty killers (who contribute more per minute than average penalty killers) but it's a reasonable starting point.

I should have specified if I meant a player getting say 3 minutes of PK time instead of ES/PP time, or in addition to the ice time he already received.

For example, if we assume that players generally trade away ES/PP time for PK time, it would probably be detrimental to put elite scorers on the PK unless they were exceptional penalty killers.

Elite scorers don't just bring a defensive contribution to the PK - they can also be an offensive threat, and coaches may value them as such. But if you are working within the framework of (points + PK credit) you wouldn't include that offensive value in PK credit, because it's already counted in points.

Quote:

Originally Posted by Canadiens1958

Penalty killing is the most physically demanding of the various game situations which may be reduced to ES/PP/PK.

A team's PP usually features the elite scorers who tend towards the higher TOI minutes. The optimum from a PK standpoint is having the lower TOI players neutralize the higher TOI players.

During the 2009-10 season Atlanta did slightly worse using Chris Thorburn amongst other secondary players on the PK - allowing 57 goals in 320 PK situations than Pittsburgh did using Crosby, Malkin and Staal allowing 52 goals in 327 PK situations. At 2 minutes PKTOI per game per elite player over a full season close to 500 minutes of elite time that should definitely generate better results.

Pittsburgh's elite forwards are all centres. Centres often get penalty killing duties, in part because it's extremely important to win the faceoff (even more so since all power plays start in the zone now). The Penguins have three faceoff men who can be trusted to take that faceoff - Crosby, Staal, and Craig Adams. As a result, two of their three best forwards have to see significant minutes on the penalty kill.

It's easier for offensive wingers to avoid killing penalties and focus on scoring.

Biggest weakness is that "time killed" is not expressed in the stats.
Example two teams. Each is assessed 250 single minors generating 500 minutes of potential penalty time. Each gives up 35 goals on the PK. Stats tend to reflect them as being equally efficient. However if the first team on average killed 1:30 of the penalty before giving up a goal while the second team on average killed 1:00 of the penalty before giving up a goal then the first team is more efficient.

I am all for more detailed stats that tell more of the story. But this is just plain insignificant. We are talking about 483 minutes versus 465 - a difference of 4%, and in an extreme example to boot - over the long run there is not much to suggest that some teams will be anomalies. Sometimes you get scored on after 5 seconds; sometimes after 115, not always one or the other.

I am all for more detailed stats that tell more of the story. But this is just plain insignificant. We are talking about 483 minutes versus 465 - a difference of 4%, and in an extreme example to boot - over the long run there is not much to suggest that some teams will be anomalies. Sometimes you get scored on after 5 seconds; sometimes after 115, not always one or the other.

At the NHL level or any pro sport level low single digit % differences are very significant.

Leaving things to chance is the old adage "Failing to prepare is preparing to fail".

Your analogy bolded is very interesting. Up 3-1 facing a 2 minute PK at the 18:00 mark of the 3rd period. Giving up a goal in 5 seconds is very problematic, giving up a goal after 115 seconds is very acceptable. Let's see them score in 5 seconds after the puck is dropped at center.

At the NHL level or any pro sport level low single digit % differences are very significant.

Leaving things to chance is the old adage "Failing to prepare is preparing to fail".

Your analogy bolded is very interesting. Up 3-1 facing a 2 minute PK at the 18:00 mark of the 3rd period. Giving up a goal in 5 seconds is very problematic, giving up a goal after 115 seconds is very acceptable. Let's see them score in 5 seconds after the puck is dropped at center.

Your very specific and specialized situation proves nothing. Boiling it down to one PP does not illustrate the potential inaccuracy in the PK efficiency statistic. Showing the effect that such a situation would have on the overall statistic at the end of the season is what would illustrate the stat's limitation. And that would prove to be extremely insignificant.

Pittsburgh's elite forwards are all centres. Centres often get penalty killing duties, in part because it's extremely important to win the faceoff (even more so since all power plays start in the zone now). The Penguins have three faceoff men who can be trusted to take that faceoff - Crosby, Staal, and Craig Adams. As a result, two of their three best forwards have to see significant minutes on the penalty kill.

It's easier for offensive wingers to avoid killing penalties and focus on scoring.

Underlining the attributes that teams look for in their 3rd and 4th line centers or multi position forwards.The PK value of these players has to reflect their ability in such situations and the ripple effect that is generated in game management - example the resulting ability of the coach to play a Crosby immediately post PK when there is an opportunity for a momentum shift or having to rest him.

Your very specific and specialized situation proves nothing. Boiling it down to one PP does not illustrate the potential inaccuracy in the PK efficiency statistic. Showing the effect that such a situation would have on the overall statistic at the end of the season is what would illustrate the stat's limitation. And that would prove to be extremely insignificant.

You fail to realize that no one was trying to prove anything. The discussion is about what has to be considered as value in PK situations.

In the future I will gladly annotate with a QED any of my posts where a proof is intended.

You fail to realize that no one was trying to prove anything. The discussion is about what has to be considered as value in PK situations.

In the future I will gladly annotate with a QED any of my posts where a proof is intended.

You made the point that PK stats are limited because they are based on opportunities and not actual time. I am telling you this is insignificant. You took it off on a tangent as usual.