NHL Team Rankings Adjusted for Schedule Strength

There is no perfect methodology to ranking NHL teams. If we were to rank teams by the number of points, this system would fail to take opponent difficulty into consideration. A Buffalo win over Chicago is worth more than a win over, say, Toronto. One way to correct for the strength of schedule is to follow a procedure called the Colley matrix (ranking) method.

The Colley matrix method, which was developed by Scientist Wesley Colley, was created for ranking college football teams based on each team’s overall wins and losses of the season and the strength of their schedules. The schedule strength takes into consideration the number of times teams play against each other and their team records. Head-to-head records and game scores are not factors here. Colley arrived at his method using “principles of probability, integral calculus, and linear algebra”, but the rankings themselves are determined with only linear algebra.

Besides college football, the Colley ranking method can be applied to other sports such as hockey. Before the unveiling of the ratings and rankings of NHL teams, here are a few notes to consider about the rating method:

• Games determined by the shootout are not removed, although statistical research has shown the shootout is a crapshoot.
• Despite the Western Conference being considered the stronger of the two conferences, there is no bias towards any one conference.
• Team ratings range from 0 to 1, with 1 representing an excellent team and 0 representing a poor team.

With the strength-of-opponents adjustment, the following are team ratings from the recently completed 2014-15 regular season:

Overall, teams do not shift too much. The good teams are still at the top and the bad ones are still at the bottom. Where the rankings differ the most is in the middle from 10th to 20th. In particular, two teams may appear to be better than what the NHL standings indicate. The Calgary Flames and Columbus Blue Jackets had the highest positive ranking differentials:

Calgary Flames

The Alberta playoff-bound team is ranked 11th according to the Colley method, five spots higher than in the NHL overall standings. The Flames faced some tough competition during the regular season, playing against the top Western Conference team, Anaheim Ducks, five times and their first round playoff opponent Vancouver Canucks four times. The Flames’ high differential may also be partly explained by Washington, Detroit, and Ottawa, who are ranked 10th, 12th, and 13th respectively in the overall NHL standings, having more “loser points” (as a reminder, the Colley method recognizes an overtime or shootout loss as any ordinary loss) and playing against much weaker opponents more often. (More on the three teams is explained below.)

Columbus Blue Jackets

The Blue Jackets, with a +4 ranking differential, were labelled as a “dark horse” team to make the playoffs this season by experts. Aside from the number of injuries to key players during the season, another reason they failed to clinch a playoff spot is how frequently they had to face stiff competition. They played against both New York teams four times each and faced the Capitals five times. Considering the strength of those opponents, Columbus did well enough, going on a 5-5-3 record against them.

There were, of course, teams that benefitted from an easier schedule. Specifically, four Eastern Conference teams have a negative ranking differential of -3:

Washington Capitals

Ranking tenth in the league in points, Washington had a great regular season under new coach Barry Trotz. Part of that success may be due to more match-ups against easier opponents. The Caps faced Carolina four times, New Jersey five times, and Philadelphia four times.

Florida Panthers, Detroit Red Wings, Ottawa Senators

These three Atlantic Division teams benefited from playing against weaker opponents, especially ones within their own division, more specifically Toronto and Buffalo, whom they went head-to-head for a total of nine times. The Bruins and Habs also seemingly benefited from playing in the same division since their ranking differentials are -2.

Points in overtime and shootout losses can make a difference in the standings. Detroit, Florida, Ottawa, and Boston have more than the league average number of extra time losses (10.2).  The combination of accumulating “loser points” and playing more often against very weak division rivals appear to be inflating the point totals for these Atlantic teams.

The Central Division, considered to be toughest in the league, has the highest average rating of 5.39. Five out of the seven teams in the division will play in the first round of the playoffs this year. The talent of teams are more evenly spread out between the Metropolitan (4.74), Atlantic (4.65), and Pacific divisions (4.58).

This version of the Colley matrix method is far from perfect. Ratings can be weighted by time and home/away effects. Games can be restricted to those ending in regulation or overtime. The simple, unweighted win-loss method can, however, show how teams are ranked after an adjustment for opponent strength and uncover teams with favorable or unfavorable schedules. It is one way of finding overrated and underrated teams.