The rankings are an average of two separate ranking algorithms. Each team starts the season with the same base rating of 7. The rating gets adjusted throughout the season based on a team’s performance against every other team it plays, so strength of schedule builds and is adjusted throughout the season based on each team’s interactions with every other team. The more data that the system intakes, the more accurate it becomes.

From the creator of the algorithms:

There are no weights per se used in the algorithm.  It is purely predictive, and a team's rating is akin to their average game performance, where each game performance is simply opponent rating + goal differential (limited).   Sounds simple, but the magic happens in the recursive nature of the algorithm as every team's rating depends on all of their opponents' ratings...plus goal differential.  After thousands of iterations in which each teams rating is adjusted to minimize the initial error (that resulted from the initial assumption that all teams are equal) convergence is reached and the rankings lock in.

The second rating is done exactly the same except that the goal differential in each game is limited to 1, thereby emphasizing wins and losses rather than pure goal differential.

All ties in a ranking will be broken by determining which team has the highest ranking in the 1 goal differential ranking.

And a little further explanation:

The program starts all teams with an equal arbitrary value and then

adjusts them all recursively based on game scores until the lowest cumulative

error between expected goal differentials and actual goal differentials is

reached (convergence).  

One effect often noticed is that a team's rating can change slightly without

playing due to previous opponents playing which affects their rating and their opponent's

rating, etc.

It should be noted that these rankings are an average performance value

for each team for the entire season, all games are equally weighted, so a team's

latest game may make up only 1/40th of a team’s rating at the end of the season.

Games that go to shootouts for decision are considered ties for ratings purposes.

Actually, there are two different rankings listed below.  The first is as described above with the maximum goal differential used per game set at 5.

The second ranking sets maximum goal differential at 1, thereby just considering

whether the game was won, lost, or tied (and strength of opponent) above their

average opponent (if they were undefeated and untied) and as low as 1.0 below

their average opponent (if they lost all of their games).

The two rankings are then fed into a “merge program” which converts the two into one average rankings which is then posted on the ACHA website each week.