After the Rangers recent poor run, I decided to do a little experiment. I believe that Sabremetrics have really improved baseball scouting and I’ve always thought that hockey needed something similar. As an MBA student, I’ve learned various statistical modeling techniques and how to do regression analyses and the like: basically, if you give me a lot of numbers, I can see what is linked with what. So I got to thinking: what makes a team good? Obviously we all have our ideas: from good defense to strong powerplay to youth to experience. We each have an opinion. But if we could get some existing stats and see if they correlated to how many points a team had? That would be pretty cool.
So I’ve done just that.
I got all of the stats that NHL.com, SI.com, and ESPN.com had online from all of the years since the lockout and plugged them into a program through Excel called StatTools. I then ran some regression analyses on number of points and their correlations to various statistics (note: if you’re also stats inclined, I can forward you the file with the actual analysis to check the work out. The file is big and most people don’t care about the method, just about the results).
I chose number of points because it summed up wins and OT losses and both determine whether a team makes the playoffs, not just wins. Then, I compared just about every team stat I could find to see what correlated the best. Once I got the top 5, I coupled those 5 with each other and with others to see what I could get as the highest correlation to points; basically, I wanted the two things that, in combination, were the highest predictor of total points.
The two most important factors were: the ratio of goals for-goals against 5-on-5 and the winning percentage of teams in games in which they did not score the first goal. This seems remarkably obvious right? Score more goals than the other team and win when you’re behind. But it reveals something much more than that.
First, it means that special teams just don’t matter that much. The correlation between powerplay percentage, penalty kill percentage and both together is very small. When watching a game, this might not seem logical because it always seems like powerplays and penalty kills happen at important moments. In individual games that might be true but over the course of a full season, it’s just not that important. 5-on-5 is the bread and butter, followed by a team’s emotional wear-with-all to come back from being down. Similarly, the number of penalties a team takes make almost no difference, nor do the number of fights, hits, or shots.
Practically, what does that mean for our Rangers? Well, right now, the Rangers are 7th in the Conference, barely in the playoffs and 16th in the league. Using the equation that the correlation gave me, the Rangers will end up finishing 6th in the conference, well in the playoffs, but 17th in the league – this is a result of the relative strength of the Western Conference.
What do the Rangers therefore need to do to get better? First, stop worrying about special teams. Instead, work in practice on 5-on-5 breakout plays, defending the neutral zone, and breaking through a “hold the line” team like Philly. It also means not being frustrated by giving up the first goal. This is obviously a leadership and teamwork issue: guys have to believe in each other and in themselves. Group vacation for Olympic break?
Questions? E-mail email@example.com<!--EndFragment-->