Analysis of divisional imbalance in 2012-2013 schedule

Chojin

Tiny Panger...
Apr 6, 2011
4,301
573
The 2012-2013 NHL schedule is quite different from the original schedule structure. The biggest change, aside from the elimination of inter-conference games, is that teams play an un-even number of games against divisional opponents. For example, the Blues play the Wings and Hawks five times apiece, while they only play the Predators an Jackets four times. Since the Blues have one less game against the worst team in the league last season, the Blue Jackets, one could say that they have a slightly more difficult schedule than those that do. The question is how to quantify this effect.

I settled on comparing the difficulty of the current season as compared to the previous one via mean opponent points percentage, or MOPP. This is the average points percentage for a team's opponents in a season, weighted by the number of games against each team. To get a better idea of how the shortened schedule affects each time, I compared the relative change in MOPP between the 2012-2013 season and a regular 82-game season, minus any inter-conference games. Obviously, this is not an ideal measure of team quality; it ignores head-to-head matchups and any offseason improvements, among other factors. But I still think that tracking the change in MOPP provides some insight into how the shortened season affects competitive balance.

I wrote a python script to scrape schedule and standings data from the www.nhl.com, then calulate the change in MOPP. Here are the results (note that the more positive the change, the more difficult the schedule):

Code:
2012-2013 NHL Schedule Difficulty by Team (easiest to hardest)

Team                 Relative Change in MOPP (lower is easier)
-------------------------------------------------------------------------------
Montréal               -0.434 %
New Jersey           -0.378 %
Washington          -0.191 %
Pittsburgh             -0.160 %
Toronto                 -0.130 %
Florida                 -0.103 %
NY Islanders         -0.099 %
Carolina               -0.072 %
Tampa Bay           -0.043 %
Buffalo                 +0.058 %
Ottawa                +0.177 %
NY Rangers         +0.250 %
Boston                +0.341 %
Philadelphia         +0.392 %
Winnipeg             +0.407 %

Colorado             -0.565 %
Nashville             -0.533 %
Detroit                -0.442 %
Anaheim             -0.202 %
Dallas                -0.175 %
San Jose            -0.132 %
Minnesota          -0.132 %
Vancouver          +0.000 %
Edmonton          +0.129 %
Columbus           +0.184 %
Los Angeles       +0.250 %
Phoenix              +0.265 %
St. Louis             +0.294 %
Chicago              +0.481 %
Calgary               +0.574 %

Winners
Colorado, Nashville, Detroit, Montreal, and New Jersey are the biggest beneficiaries of the new schedule. This is typically due to easy divisional matchups; for example, Colorado plays Minnesota and Edmonton five times each, both of whom were among the worst teams in the West.

Losers
Calgary and Chicago are the recipients of the biggest increases in difficulty, followed by Winnipeg, Philadelphia, and Boston. Calgary has to play Vancouver five times, and poor Chicago has to play the top teams in the Central Division five times as well.

EDIT: I didn't hear anything from a mod on this yet, so I'll assume it's OK if I link to the code I used for this analysis: https://github.com/thesquelched/nhl-season-judge
 
Last edited:

Chojin

Tiny Panger...
Apr 6, 2011
4,301
573
I'd like to add that this is my first shot at any kind of hockey analysis, so I'd appreciate any criticism or advice.
 

saskganesh

Registered User
Jun 19, 2006
2,368
12
the Annex
It's cool you did this.

As you said, one obvious flaw is that it is based on last year's results.

It would be very interesting to compare final season standings with this early analysis.
 

Chojin

Tiny Panger...
Apr 6, 2011
4,301
573
Its a cool idea but did you use the whole season last year or just the results from the teams playing in their own conference?

Originally, I made the mistake of using the whole year, but I revised the result to use only in-conference points percentage before I posted it here.

EDIT: Added link to code in the OP.
 

MNNumbers

HFBoards Sponsor
Sponsor
Nov 17, 2011
7,658
2,536
There is another statistical method used to evaluate competitions with uneven scheduling. As an example, it's used in chess tournaments.

It called the Bradley-Terry Method. It yields what could be thought of as an odds/power rating for each team. Based totally on results W/L/T (I would exclude shootouts). The ratings end up as an odds scale. So, if LA had 200, and Minn 67, then La would be likely to earn 75% of the points in games vs Minnesota (that is, 200/(200+67). I haven't done a full work up of last year ever. But, I think that is the place to start.

After you have each teams rating from last year, it takes a little more math to figure out strength of schedule for this year.

Has anyone seen that?
 

Ad

Upcoming events

Ad

Ad