Nick Papagiorgio
Holding court
Re: 2012 NCAA Tournament: Bracketology
Excellent breakdown. Much appreciated. All the best to your scarf (which really should check in more often on elf).
That's right. The pairwise at TBRW?/slack.net also uses average winning percentages, although it's not spelled out as nicely in terms of team-by-team breakdowns as on CHN. For example, Denver wins the comparison with Michigan State thanks to common opponents, which are:
Air Force: both 1-0
MTU: Denver is 0-1-1, MSU is 1-0
BC: Denver is 1-0, MSU is 0-1
Under the old system, Denver would be 2-1-1 (.625) and lose to Michigan State's 2-1 (.667). But in the new system, Denver's averaged winning percentage is (1.000+.250+1.000)/3 = .750, while MSU's is still .667, so they win the criterion and with it the comparison. I didn't have the time to make the TBRW/slack version pretty, so it says "2-1-1 .7500 COp .6667 2-1" which is somewhat confusing, since it implies we're comparing 2-1-1 to 2-1, rather than .7500 to .6667. Similarly, USCHO's version says "2-1-1 2.250 1 COp 0 2.000 2-1-0" which is the same thing multiplied by 3 (the number of opponents). The CHN version leaves out the (irrelevant) overall record and includes an opponent-by-opponent breakdown. But all of them compare equivalent quantities, whether it's 2.25 vs 2.00 or .750 vs .667.
Excellent breakdown. Much appreciated. All the best to your scarf (which really should check in more often on elf).