Here’s a variation on the sorting algorithm I use to sort the graph into power rankings.
Find all the teams that don’t have beatlosses. For Week 8, that is DEN, IND, SEA, CAR, NYG, and ATL. Then out of those, find the team that was ranked highest last week. Add that team to the power rankings. Then remove that team (and its beatwins) from the graph. Repeat.
So, after IND and DEN are added, JAC becomes a team with no beatlosses, and gets evaluated against SEA, CAR, NYG, and ATL. This continues on down the line.
It’s a simpler algorithm than the one I’ve been using, where I start with the teams with the longest beatpath segments, but it ends up with similar results. For instance in Week 8, DAL and ATL would be flip-flopped, while NYJ would fall from #18 to #26, but the rest of the ordering remains identical.
In some circumstances, these algorithms can still tend to inappropriate reward teams that have played erratically enough that they have no beatlosses even when they have hardly any beatwins – in other words, teams that have entered a lot of beatloops (like MIA). That may be a good tiebreaker to add into the algorithm, examining the teams’ respective beatloops.