Hockey: Fighting to Win?

FIGHTS AND WINNING

We had an interesting question posed to us last week.  How does fighting in hockey affect win percentage?  Well, we thought it was interesting too and now we have an answer for you.  In fact, we decided to answer a couple questions about fighting in hockey.  Our data was taken from ESPN and DropYourGloves.com.  We first analyzed the correlation between fights per game and win percentage.  The relationship is shown in this scatter plot.

The plot shows that there might be a positive trend in win percentage as fights per game increase, but the trend would be small and there likely isn’t any real trend or correlation at all.  We decided to try the same thing using points instead of win percentage.  In hockey, the standings are not determined by win percentage, but by points.  In an NHL game, two points are awarded to the winning team, one point to a team that loses in overtime, and zero points for a team that loses in regulation.  Here is the scatter plot of fights per game against standings points.

As with win percentage, there might be a slight positive trend, but it would only be slight and the trend or correlation is again, not probable.  We ran a multiple linear regression with random effects for teams and years to solidify our understanding of the correlation.  Basically, we ran a statistical test to see the effect of fights per game on standings points for a team, controlling for teams and years.  We found that there is not enough evidence to claim any kind of correlation between fights per game and points.

FIGHTS AND ATTENDANCE

Then, we analyzed what the correlation is between fights and attendance.  A lot of people claim that fights bring in the fans and we’ve thought the same thing, hockey fights can be extremely entertaining!  The analysis showed some interesting and surprising results.  The statistics that we used were home attendance percentage (attendance/capacity) and again, fights per game.  The plot for that is shown below.

The trend is positive in the beginning, but then slopes down just over halfway through the observations.  This is not the trend that we expected, but there are a few stray points (outliers) that make us question the strength of the trend.  So, we looked at last year’s data as well.

We have the same trend.  We ran another multiple linear regression and found a possible negative trend, but we don’t have enough evidence to claim it with much confidence.  The next step was to add standings points into the regression and see if the trend strengthens after we account for points.  The trend stayed faint and weak.

WINNING AND ATTENDANCE

However, we noticed something weird in our analysis, there doesn’t seem to be a strong trend between winning and home attendance.  The obvious assumption is that as a team is more successful, they draw a larger crowd, but some teams with terrible records still sellout.

Teams that are highly successful always seem to draw a large crowd, but so do some teams that are awful.  We thought this was counter intuitive and hard to believe, so we contrasted it with the NBA.

Same thing.  One interesting thing to note is that if teams have a win percentage (NBA) or points percentage (NHL) of 60% or higher, then they sell out.  However, we’ll need more data to be certain, a project for another day and another post.

During this project, we discovered some interesting things.  Fights in hockey don’t seem to have any real, direct, significant effect on a team’s success or home attendance.  An interesting question for the future “Do fights in hockey prevent injuries?” The idea that they do prevent injuries seems to be widely believed by hockey players.  The data for this has been hard to find at a reasonable price (free or close to it), but we’ll continue to search for it.  As for the hockey fight questions that we discussed today, we have found no real evidence that fights have any significant, direct effect.

These charts and another chart, along with more information on the data can be found here.