Drawing penalties, taking penalties, and delaying the game. 

Travis Yost of TSN wrote on January 27^th, 2015 a piece called "Drawing penalties a valuable NHL skill". As you can imagine it discusses how valuable it is for teams to have players on their roster who can draw, or induce, penalties by the opposition. It also talks about the opposite, players who are goaded into taking more penalties than they cause their opponents. We all agree that being able to generate power play time for your team is a valuable skill set.

However, there are some flaws in his assessment and evaluation. We will identify them as we go through his column.

The Tables

He has four tables within his article, two showing positive effects and two with negative effects. The issue is that each of the paired tables is in fact the same. Take the two positive tables, 'Forward Penalty Differential Leaders Since 2011' and 'Penalty Differential as Goal Value, 2011-Present'. They tell the same story and the second does not add much more value in its conversion to Goals.

His assertion is that if you convert the penalties into goals then you see the positive impact these players have for their respective teams. The fault is that the same factor is used for each player, 18 percent. This 18 percent is the average power play conversion rate. So in essence the two tables are the same just with numbers that look like they tell a different story. Would it not be better to use the actual power play conversion rate of the team the player belongs to though? If Carolina has a worse conversion rate than Los Angeles then it should factor into the second table values. The player did this job to get the team a power play but if the team converts those opportunities into goals is a different matter.

One added difficulty could be if the player is traded sometime during the season, then you would have difficulty figuring out what the team's power play conversion rate would be and then apply it to his penalties stats.

The Range of Data

The statistics for penalties drawn and penalties taken are not easy to find. You can go through the Play By Play logs found on NHL.com (the one in the link is of the Chicago Blackhawks vs Los Angeles Kings game on the 28^th). Yost references an article by Broad Street Hockey where they have done a similar job and done all the compiling for you and then converted the data into 'per 60 minutes' and 'per 82 games' values. I have also been able to find some more data at behindthenet.ca where the two columns that are of interest are 'PTAKE/60' and 'PDRAW/60'. These stats are averages in order to factor out games missed.

Finding the raw data is hard, complicated and time consuming. It is why I believe that his range from 2011 to the present was selected. That is about four seasons worth of data though. If we are to divide the data in the tables by four then his values look a lot less powerful. For example, Jeff Skinner would have a Goal Value Differential of plus-three instead of plus 12.1.

The Assumptions

It is not said explicitly but one assumption is that these values are only when the player gets an opponent to take a penalty, or when they get caught. Nothing is captured when there are off-setting calls.