The main thing about the Beatpaths Power Ranking calculation method is that when in doubt, they defer to the previous week’s power rankings. The reasoning behind that should be obvious. Assuming the previous week’s power rankings are good, the new power rankings have just as reliable a foundation to base its rankings on.
The question is, what if the previous week’s power rankings are flawed? If that happened, some or all of those flaws would be inherited. However, there is one principle to the power rankings that is key:
The Beatpath Rule: If a team A has a beatpath to team B, team A is ranked somewhere ahead of team B. If team B has a beatpath to team A, team A is ranked somewhere behind team B.
So say that for some reason, team B being ranked ahead of team A is a flaw in prior power rankings. If team A then develops a beatpath over team B, then due to the Beatpath Rule, that flaw would be removed from the power rankings.
This means that the power rankings will heal themselves over time.
The question that remains is, what was the first power ranking, how valid was it, and how much of a bearing does it have on the present-day power rankings?
Well, given enough time, due to the self-healing aspect, the validity of the first power ranking is irrelevant. The question is, are we fully healed yet? Are the present-day power rankings completely unaffected yet by the first “seed” rankings?
And it appears the answer is: Not yet, but we’re so close. I tried to come up with a pretty valid seed power ranking by ordering the teams by their beatpaths. That implied a bunch of subjective (non-beatpath) rankings, which I keep track of from week to week. These subjective rankings become either confirmed or denied by beatpaths over time, and so over time there are less and less of them.
So, I tested the rankings by having them start with a seed that was in the reverse order of the seed I have been using. That seems about as destructive as possible, since it had the super bowl champions in last place and the worst team in the league in first. This was an extremely flawed power ranking.
So of course, over the ensuing weeks, the power rankings were very different than they were from the original rankings. But as the beatpaths asserted themselves, and the (flawed) “subjective” rankings slowly disappeared, the rankings slowly started to match up to the original rankings.
And while we still have one game to go for Week 8, it appears that the power rankings will be almost completely healed. The only difference is that Dallas and Atlanta, who are ranked right next to each other, will be flip-flopped.
So we’re close. Soon, the power rankings will be completely stable, which will mean that the rankings will be in the same order no matter what seed rankings they started with.
There are a couple of other possible ways that the power ranking order can be affected and it should still be considered beta until I get those last couple of issues nailed down. But that’s for a later entry.