In one of the fanposts I wrote, a topic that came up in the comments was how to explain the New York Rangers' great 2011 season even though they were an average possession team. One of the reasons I suggested was the fact that Lundqvist's PK save percentage was much higher last year. Another explanation would point to the fact that the Rangers were only giving up 2.22 goals per game (all due to King Henrik standing on his head). But their offense was average and they didn't stick out in any other statistical category, so this was still somewhat of an anomaly.
Turns out I was looking in the wrong place. Where I should have been looking was the Rangers' record in one goal games. The Rangers had the fourth highest winning percentage in one goal games last year, but they're not as good (though still above average) this year. I think this fact is another reason why the Rangers are struggling.
This also prompted me to see if teams could persistently do well in one goal games from year-to-year. My intuition said no but I decided to collect the data and run some correlations to find out. What I did was run three correlations. The first looked at the correlation between team winning percentage in one goal games between the 2011 and 2010 seasons. In other words, I was trying to see how the winning percentage for the 2010 season correlated with the one goal game winning percentage for the same team in 2011. The second looked at the same thing for the 2009 and 2008 seasons. The last examined the same thing again for the 2007 and 2006 seasons. So the output was three correlation coefficients measuring the relationship between team winning percentage in one goal games for the 2010/2011 seasons, 2008/2009 seasons, and the 2006/2007 seasons. All of the data was from NHL.com.
For those who need a primer on correlation coefficients, Wikipedia has a nice summary. In short, the coefficients range from -1 to +1. A value of +1 means a perfect correlation between the two variables (here, that would mean the team winning percentage in one goal games is the same from year-to-year); a value of -1 indicates a perfectly negative correlation; and a value of zero means no correlation at all. Generally speaking, a coefficient of 0.3 or higher means there is something significant, though that could be broadened a bit in the context of hockey games.
With that in mind, the results are below:
I think it's clear that winning one goal games is not a persistent skill (the data from 06-07 is especially damning). Thus, this might be another reason why the Rangers did so well last year but are fighting for a playoff berth this year. However, I still think there are other reasons which explain why the Rangers lit it up last year which have yet to be uncovered.