Normalizing to Full Round Robins What this means is that teams' winning percentage multiplier as established by preliminary play cannot be altered by subsequent bracketed play or playoffs insofar as teams continue to win or lose as "expected." To illustrate, consider a classic round robin outcome among 12 teams, with no upsets. Team A goes 11-0, B is 10-1, C is 9-2, D is 8-3, etc. Say that this tournament then moves to playoff brackets after division into the top four, middle four, and bottom four. D loses three more matches, as expected, to A, B, and C. Their record has now dropped to 8-6, and their winning percentage has dropped from .727 to .571. Meanwhile, Team E, which started at 7-4 (.636), was the best of the middle bracket, and finishes 10-4 (.714) as expected. For our purposes of using winning percentage as a major factor in ordering teams, clearly it would be unfair to rank E ahead of D. What we do then, is assume a second full round robin, even though only three games of it are actually played. In addition to its three additional losses to A, B, and C, Team D is given additional paper victories over each of the teams in the middle and lower brackets, and thus its adjusted winning percentage remains just as it was before bracketing, at .727. Likewise, E is given paper losses to A, B, C, and D, and paper wins over the bottom bracket teams, and it's adjusted winning percentage also remains where it began, at .571. Going into those final three matches, .727 has to be the lowest winning percentage that Team D can emerge with, since they are not "supposed" to win any of the three. They can, however, go up. If D wins one of those last three matches, their final record is 9-5, and their actual winning percentage has dropped from .727 to .643, though what they've done is score an upset. So our response is to "normalize for another full round robin," reflecting that result. They went 1-2 against the top three, but for our purposes we assume they were also capable of going 8-0 against the bottom eight that they would have also played in another full round robin. So for our multiplier purposes, we use the winning percentage of their presumed second round robin record of 17-5 (8-3 in the first one, and 9-2 in the second, with 8 of those games not actually being played -- and for our purposes, their adjusted winning percentage becomes .773. The percentage has gone up from their original .727 thanks to the upset win. Had they followed form and lost all three, we'd still assume that the unplayed games would follow form, and that they'd be 8-0 against the rest, so they'd be 16-6; the same percentage as they started with. If it was Team C that D beat, which went 0-3 in the bracket, since they also lost to A and B, they would also now be presumed to have finished with a record of 17-5 and a .773 winning percentage for our calculations purposes. This all requires an intelligent look at playoff structures and matches played in each tournament, but the approach extrapolates on the same principles to cover any situation. The point is that we use must use winning percentage in an intelligent manner that hurts no team because they had to add extra playoff or finals matches versus stronger teams that they were supposed to lose to, and also helps no team that gets extra matches versus weaker teams that they are supposed to beat.
This archive was generated by hypermail 2.4.0: Sat 12 Feb 2022 12:30:45 AM EST EST