39
When you look at something through the same lens thousands of times you come to accept any astigmatism as if were part of the real world. Such is the situation with the NHL's standings — two points for a win and one point for an overtime loss. The only material change to this lens in decades was the introduction of a bonus point for winning a game after a regulation-time tie.
What is wrong with the standings? The first problem is that not all teams have played the same number of games at any point in time. A second problem is that teams may have played more home than road games (or vice versa). And a similar problem is that the schedule is unbalanced, especially part way through the season. The final problem is that these games represent only a small number of contests. The 'statistical sample' is small and luck can play an important part.
So-called "power rankings" are not the solution. These are, almost always, an opinion, or a committee view, and they tend to mix up the questions of team abilities and "who is hot and who is not".
There are, however, scientific approaches to rating teams. Jeff Sagarin is probably the most famous sports team rater on the planet, but there are a couple of others out there worth of note — Ken Massey and Andy Dolphin .
Here is how the NHL and these scientific methods rank the top NHL teams today:
| Rank | NHL (points) | Sagarin | Massey | Dolphin |
|---|---|---|---|---|
| 1 | Detroit | Detroit | Detroit | Detroit |
| 2 | Ottawa | Ottawa | Ottawa | Ottawa |
| 3 | San Jose | San Jose | San Jose | Colorado |
| 4 | Dallas | Colorado | Colorado | Boston |
| 5 | Boston | Boston | St. Louis | Minnesota |
| 6 | New Jersey | Dallas | Minnesota | San Jose |
These scientific ratings rely on complex interpretations of game match-ups and results. Importantly, these methods rebalance the schedule and enable the comparison of teams that have not met (a real issue given the NHL's shortage of inter-conference play). They rely on mathematics like the simultaneous solution of, for the NHL, 30 equations involving 30 unknown variables — something not to be attempted without proper safety equipment.
It is always important to read the disclaimers on product packaging and the fine print on these methods says that they were designed to rank college football teams. In that world mismatches are common and the possibility of blowout games is quite real. To compensate, each of these ratings methods finds a way to discount large margins of victory and, in effect, the ability to win games is given greater weight than the ability to dominate an opponent.
That would seem to be just plain wrong in the NHL. As hockey's pre-eminent league is quite well balanced, a blowout is a meaningful event. Every piece of research I have seen suggests that winning is very highly (about 93%) correlated with goal differential. Teams that can win big are better than teams that eke out victories in close games. Expressed differently, good teams win more than half of their close games but dominate the games that are not close. Furthermore, goal differential provides for a 'statistical sample' is much, much larger than for wins and losses. There is less statistical noise associated with luck.
So goals for and against are a better way to measure a team than points in the standings. Here is what that says about the NHL's standings today:
| East | ||||
|---|---|---|---|---|
| Rank | Goal Differential per Game |
Pts | Pts Rank | |
| 1 | Ottawa | 0.90 | 45 | 1 |
| 2 | Buffalo | 0.35 | 33 | 10 |
| 3 | Philadelphia | 0.23 | 35 | 6 |
| 4 | Boston | 0.22 | 39 | 2 |
| 5 | New Jersey | 0.15 | 39 | 2 |
| 6 | Montreal | 0.09 | 37 | 4 |
| 7 | Carolina | -0.06 | 37 | 4 |
| 8 | Toronto | -0.06 | 34 | 8 |
| 9 | Pittsburgh | -0.06 | 34 | 8 |
| 10 | NY Rangers | -0.09 | 35 | 6 |
| 11 | Florida | -0.18 | 32 | 11 |
| 12 | Tampa Bay | -0.18 | 31 | 13 |
| 13 | Washington | -0.39 | 28 | 15 |
| 14 | NY Islanders | -0.52 | 32 | 11 |
| 15 | Atlanta | -0.81 | 29 | 14 |
| West | ||||
|---|---|---|---|---|
| Rank | Goal Differential per Game |
Pts | Pts Rank | |
| 1 | Detroit | 1.22 | 49 | 1 |
| 2 | Dallas | 0.39 | 40 | 2 |
| 3 | San Jose | 0.34 | 40 | 2 |
| 4 | Vancouver | 0.15 | 38 | 4 |
| 5 | Colorado | 0.13 | 37 | 6 |
| 6 | Columbus | 0.03 | 34 | 9 |
| 7 | Chicago | 0.03 | 32 | 12 |
| 8 | St. Louis | 0.00 | 34 | 9 |
| 9 | Minnesota | -0.03 | 38 | 4 |
| 10 | Calgary | -0.06 | 37 | 6 |
| 11 | Phoenix | -0.19 | 30 | 13 |
| 12 | Nashville | -0.20 | 30 | 13 |
| 13 | Edmonton | -0.32 | 34 | 9 |
| 14 | Anaheim | -0.43 | 35 | 8 |
| 15 | Los Angeles | -0.55 | 26 | 15 |
It's unanimous. Detroit and Ottawa are tops in the league, in a class by themselves. But no rocket science was required to reach that conclusion. My other top six teams include Dallas, Buffalo, San Jose and Philadelphia. Boston and New Jersey slide out of the NHL's top six, although not far. I conclude that the three rating pros think too highly of Colorado.
Buffalo has been awfully unlucky — generally outscoring opponents but coming up short on victories. Chicago has a roughly break-even goal differential yet lags in the standings. Bet on these teams, especially the Sabres, to climb in the standings as the season progresses.
The Ducks (especially), Flames, Wild (notwithstanding the rating pro's methods), Hurricanes and Rangers may not realize it but they are in danger of missing the playoffs. These teams have been fortunate to win as much as they have been. Look for these teams to sag in the standings unless they pull up their stockings.
It looks like it is going to be a long season in Atlanta. And nearly as long for the Islanders, Kings, Ducks, Capitals and Oilers.
This approach does not rebalance the schedule, which is an important thing to do but a lot of hard work. However, it is dead simple. You can play along at home.
