I have a few more things to point out about the A-value calculations.
First, it seems like teams with the same record were ranked in Order of Finish by (playoff) PPG. I thought that it was tradition that teams with the same playoff record are officially recorded as having tied for the position (ex. here
). This affects the A-values of a few teams. For example, Penn State C (3-2, 234 playoff PPG) was determined to have placed above Kenyon A (3-2, 215 playoff PPG). Since Kenyon's prelim ppg was much higher, they had a higher raw A-value. This caused Penn State C, Kenyon, and Wright State's A-values to be averaged. In previous years, and in NAQT, a tie by record was not broken, and teams were simply ranked by A-value (for example, Alabama and Georgia B last year, where Georgia B had higher playoff PPG but lower raw A-value, so was behind Alabama), so there is no possibility of averaging. This affects UNC B, W&M, Duke B, Kenyon A, Penn State C, Wright State, WVU A, Pitt A. Possibly related to this, Pitt B and Michigan C's A-values are averaged together even though Pitt B finished ahead of Michigan C by every possible metric.
Second, at the Carnegie Mellon site, Michigan State A was deemed to have finished ahead of Michigan B. Michigan State (1-4) defeated Michigan B (2-3) in a disadvantaged UG final. NAQT's position on this would be that Michigan still places higher than Michigan State (ex. here
). As far as I know ACF doesn't explicitly have a rule that covers this situation, which leads me to think that it would be the same as NAQT's. If Michigan is placed ahead of Michigan State, their A-values would get averaged, which would pull Michigan B up a couple spots (and Michigan State down a few, but they would receive the UG auto-bid anyway).
Third, the handling of tiebreaker tossups seems a bit counterintuitive. It seems like the extra tossup is counted in TUH, but otherwise no corrections are made (with the exception of the GT site - I couldn't figure that one out). This doesn't take into account the fact that the team that converts TU21 doesn't get a bonus, so you can't use PPTH as a proxy for TUPPTH + BHPTUH * PPB. This is complicated by what to do when a team negs TU21 - does that count as a TUH for the winning team if they don't get a chance to buzz in? This is further complicated by the fact that, at the Yale and GT sites, the winners were allowed to buzz in and claim points, even though the game should have ended on the neg. Depending on how this is handled, teams involved in tiebreaker tossups may see their A-values change by a couple of points in either direction.
In the future, I propose that the score at the end of regulation is used for A-value purposes, with TU21 ignored and not counting as a TUH. (It appears that this is what YellowFruit does when calculating PPG [but not other stats].) This would make things easy to calculate while maintaining fairness.
Fourth, teams that played on half-packets have their SOS calculated with the half-packet opponent counting as a fully-weighted opponent, instead of a half-weighted opponent. (TPPTH and the like are calculated correctly.) This causes MIT, Wesleyan, Penn B, Rutgers B, and Millersville to be a point or two higher than their current A-values.
Ultimately, this probably won't make that huge a difference on the A-value rankings (except for maybe point 1). I think my team's ranking is unaffected by any of this. I just wanted to bring this to everyone's attention.