Saturday, December 25, 2010

The East might be the better conference

The Western Conference has had the better record in interconference games in each of the last ten seasons. That trend has continued into the current season, with the West putting together an impressive 67-50-20 against their Eastern counterparts thus far (games that ended in a shootout are counted as ties). However, upon closer inspection, it would appear that the two conferences are much closer to one another in relation to ability than the results of the interconference games would suggest.

[Note: The data in the table was calculated after removing empty net goals]

The table pretty much says it all. Firstly, while the West has 12 more non-shootout wins than the East, they're only +11 in terms of goal differential. Generally speaking, a win is worth 5 or 6 goals with respect to net goal differential, so the West should only have about four more wins than the East on merit.

Secondly, the underlying numbers are revealing. The West has done better in terms of the percentages, particularly at even strength, whereas the East has done better virtually everywhere else. Part of the shot differential gap is surely attributable to the fact that Eastern teams have (presumably, given their record) spent more time playing from behind, but it's worth noting the East has still done better with the score close. The East has also been better at generating shots on the powerplay - their six extra PP opportunities can only account for about 15-25% of the shot gap.


Norm Albertson said...

So the East maybe better because the goal differential isn't as big as it should be? Where does the win = 5-6 goals in goal differential data come from? You've removed EN goals from the data, do you expect that 5-6 goal differential per win to stand up if you do that? No snarkiness intended, just trying to understand.

JLikens said...

"So the East maybe better because the goal differential isn't as big as it should be?"

The fact that the Western conference has outperformed its goal differential merely shows that the gap between the two conferences is smaller than the head-to-head results indicate.

The fact that the East has outshot the West in head-to-head contests suggests that it might be better.

"Where does the win = 5-6 goals in goal differential data come from?"

The slope of the linear relationship between goal differential and wins is about 5-6 (closer to 5 since the advent of the three-point game).

"You've removed EN goals from the data, do you expect that 5-6 goal differential per win to stand up if you do that?"

Good point - the removal should lower the value of the slope.

But even taking that into account, the original point remains, that being that the West has still outperformed its goal differential to some degree.

"No snarkiness intended, just trying to understand."

Yeah, no problem. I'm happy to answer any questions.

Norm Albertson said...

Thanks for the explanation. My non-statistical impression is that the Western conference is better but the difference between the two is smaller now than the last two years. Mostly, I think this is because of the upgrade of the Southeast Division. I was surprised by the 5-6 goal differential/win. Being a Predators fan, I'm used to win being +1-2 goals and a loss being -3 or more so the 5-6 range seemed high. Thanks for putting the data together.

Anonymous said...

The problem with this analysis is that it treats all teams in the East and West equally. I would be much more interested in a comparason of say the top 4 or top 8 teams of the east and west. That is what really matters for the SC Finals. Who cares how the Devils and Islanders match up against the Flames and Oilers?? Its all about the Red Wings, Flyers Pens Canucks etc.

JLikens said...


Yeah, the gap is definitely smaller now. I was looking at detailed East vs. West head-to-head data from 2007-08 to 2009-10, and the West was just comprehensively dominant during those seasons, both in terms of goal and shot differential, at both even strength and on special teams, and even in terms of drawing penalties.

I agree that, intuitively, 5-6 goals seems like an overestimate when it comes to quantifying the relationship between goal differential and wins.

I find that it's easier to conceptualize if you think about the result of a single hypothetical game. Suppose Team A defeats Team B by three goals (the average margin of victory at the NHL level is between 2 and 3 goals). Team A will emerge with one win and will be +3 in goal differential. Team B remains winless and has a goal differential of -3.

Now further suppose that Team A and Team B are the only two teams in the league, with the above game being the only game played thus far. If one set out to determine the relationship between wins and goal differential, he or she would find a slope of exactly 6.

JLikens said...


How is that problematic? The purpose of the post was to determine the strength of the two conferences vis-a-vis one another. In order to properly do that, you need to include all teams in the analysis.

I understand if you're more interested in how the elite teams in each conference stack up against one another, but that's a separate matter.

Host Pay Per Head said...

But they have the best record it makes them more interesting but I have to admit hat east is more competitive.