I am on winter break from college, so I find myself with a lot of time on my hands. I decided to channel all of that into a little statistical analysis. I looked at how the devils "special teams" compared to others around the league, and then I tried to see if there was any statistical evidence to support the claims that Cory is getting an absurdly low amount of goal support.
To avoid having to read through all of this, my findings are: The Devils have one of the best special teams in the league, but this has a weak correlation to getting points. They underperform more than most teams, however. Also, Cory's lack of goal support is not statistically significant enough for us to worry, as he faces much tougher competition than Brodeur.
I had a lot of fun making this so if you'd like to see how I got these findings, enjoy.
First comes the invention of a new stat, STR. STR, or special team rank, is just the addition of a teams PP% and PK%. One would expect the average to be about 100%. The chart of STR, from stats at ww.NHL.com, is shown below. (Note the ranking on the left is just the PP rank, my mistake).
I decided to see if there was a correlation between between P% (Team Point Percentage) and STR%. In other words, how much does this STR contribute to teams winning hockey games?
P% = 0.988(STR%) - 42.36
R² = 0.2341
It seems that the correlation is quite weak, as STR accounts for 23% of the total variance in P%. This makes sense as special team time only accounts for a fraction of the 60 minutes. It is interesting to note, however, that this line of fit would predict that the devils would be at P%=0.988(105.5) - 42.36= 61.874%. This is very different from the 50% we are currently at, and at this 61.874%, the Devils would have 46 points: Good for 2nd in the Metropolitan and 5th in the East. Does this point to 5v5 play as a weakness for the Devil's? I think it does. Still, having the 6th best special teams in the league...not too shabby.
Now to look at the infamous Cory vs. Marty Goal support controversy. I'm a little sick of it, so I thought it'd be a good idea to show statistically that people should not worry. All stats are from www.NHL.com. GA/G:Goals Against Per Game, GF: Goals For, Differentiall = GF-GA/G
|Marty Starts||Opponents||GA/G (Op)||GF||Differential|
|Cory Starts||Opponents||GA/G (Op)||GF||Differential|
Now for the important part.
|Goalie||GA/A OP||GF||Differential||Standard D|
So, I compared how much we score for our goalies compared to how much our opponents usually give up. We can clearly see that Cory faces tougher competition, as his opponents GA/G is 2.43 while Marty's is 2.73. Marty receives .3167 more goals/game compared to the league average, while Cory receives .8113 less. This seems very significant, until you take into account the standard deviation. The Devil's offense varies so much on a nightly basis, that statistically the variation in goal support isn't very significant. Marty's additional support occurs 41% of the time by natural variability, and Cory's lack of support would happen 29% of the time by natural variability. Neither of these are significant enough to raise concerns in my eyes. Does Cory receive less goal-support? Definitely, but not so much that we should be overly concerned. 1.65 compared to 3.05 seems like a huge difference, but stronger opponents and the beautiful unpredictability of hockey make these numbers understandable.
I think, as many of you all would agree, that it is just a matter of time before things balance out. It is nice to confirm that Cory IS facing tougher competition and his record is not just a product of less goal support.
I do realize that the creation of STR and some of my other assumptions may not be a totally accurate reflection of hockey, but I think both analyses were interesting. Thanks for reading, and what did you take from these? Is STR a stat of the future? Did I make any mistakes in my analysis? I'd love to discuss my findings below. Go Devils!