I can explain the base premise:
Teams are first assigned a score based on win/loss and severity of win/loss. For the first 4 games, these are then aggregated and sorted by trend and sum. This is how the initial rankings are done.
Going from there, it splits into two formulas. One formula takes the original, then adds a bonus based on the value of the opponent. This is the lesser formula, the K-Score, and is used as the first tiebreaker. I've refined the formula for the upcoming season to be more complex and accurate, but for the previous season K-Score was 4 points for a blowout win, 3 points for a respectable win, 2 for a respectable loss, 1 for a blowout loss, 0 if you were a non-opponent (eg, Malta). You then got bonus points: 4 for a win against an A team, 3 for B, 2 for C, 1 for D, 0 for F or a respectable loss, 3 for A, 2 for B, 1 for C.
The other one, the one used for primary ranking is more complicated. Each team had a value that determined their class (A, B, etc). A respectable win/loss against a team in the same class resulted in negligible change to this value (if two B teams played each other to a 27-24 game, both teams values wouldn't budge enough to cause a class change). Winning big against the same class caused a small, but significant increase in value, conversely losing big to the same class caused a similar decrease. Losing to a team below you or beating one above you resulted in more significant value swings.
