I wrote: <<But, combining them with per-tossup statistics does allow teams that lost one close match to move ahead in the rankings--reflecting the notion that teams should not be punished for a small statistical fluke.">> while Shaun responded: "Also agreed - in the case of UPenn, we statistically may have been the inferior team, but I still think when we beat UPenn, we have shown we are the better team. However, a difference between 9-4 and 6-7 (in the case of Princeton B) is more than a statistical fluke. Unlike in baseball, there IS such a thing in quizbowl as "performing in the clutch". I'm not sure that you understood what I meant. I *support* a system which allows teams to move up a place (or two) in the ranking against a statistically weaker team. The reason for this, as I argued in a similar post, is that two teams which are of essentially even strength--such as Yale, Williams, and MIT--should, in theory, have roughly equal numbers of wins and losses if they play each other often enough. As a result, just because team A beat team B on packet X, doesn't necessarily mean team B won't beat team A on packet Y. If one assumes that those teams will have roughly even statistics for that match, the "better team" will be decided by comparing not one team to the other (in the context of a single game), but by comparing them to the rest of the field. On that account, you need something beyond win-loss record to fine-tune the rankings. --AEI
This archive was generated by hypermail 2.4.0: Sat 12 Feb 2022 12:30:43 AM EST EST