The quirks of the PairWise Rankings are well-documented. Seemingly-irrelevant games turn out to matter, and wins aren’t always all they’re cracked up to be. Because some explaining will help make the point, this blog entry will be in two parts. Part I presents a little historical perspective, in the form of two well-known PWR head-scratchers.
• “Maybe they should have lost”: This was the situation surrounding Wisconsin’s first-round WCHA playoff series against Alaska-Anchorage in March 2005. At the time, Wisconsin was fighting to stay in position for an at-large bid, and prior to the series UAA was just barely a Team Under Consideration, which at the time was defined to be a team with an RPI of at least .500 (the definition has since been changed to include only the top 25 in RPI).
UW’s conundrum was that by winning the series, the Badgers knocked UAA out of the TUC list and therefore lost credit for six wins that season over the Seawolves that would otherwise have counted as TUC wins (remember that record against TUCs is one of the PWR comparison criteria). Worse yet, by winning the series Wisconsin gave itself a chance to lose one or more additional games at the WCHA Final Five, which the Badgers promptly did by getting beat, 3-2, by North Dakota in the play-in game.
That left observers of the PWR wondering whether the Badgers would come to rue beating the Seawolves. It wasn’t until the dust settled on Championship Saturday that UW was safely in the NCAAs, rendering the discussion moot — but still historically interesting.
• “The Doomsday Scenario”: This is what I called it in private, at least, since it would have resulted in pandemonium among fans and potentially at the NCAA as well. The year was 2006, and the Atlantic Hockey semifinals were set with Holy Cross facing Connecticut and Bentley taking on Mercyhurst. At the time, Holy Cross was a solid No. 19 in the PairWise, but that was five spots off the bubble, meaning the Crusaders would still need to win the AHA tournament title to make the NCAAs.
Or would they? A little random digging uncovered this scenario: if Bentley and Holy Cross were to advance to the AHA finals, and Bentley then beat Holy Cross in the championship game, the rules would have provided a huge and unanticipated boost to Holy Cross’ NCAA credentials.
You see, at that time any conference tournament champion was deemed a TUC regardless of its overall RPI. In one way, this made sense: a team that won its conference tournament was automatically in the NCAAs, so of course it should be called “under consideration.”
But the law of unintended consequences applied; Holy Cross had beaten Bentley four times during the regular season, and with a loss to Bentley in the AHA title game making the Falcons a TUC (they were nowhere close otherwise), Holy Cross would have had an additional 4-1-0 tacked onto its TUC record. Since Holy Cross had otherwise played few TUCs during the season, this would have had a huge impact on its PWR comparisons, potentially putting the Crusaders into the NCAAs as an at-large team.
That exact situation came tantalizingly close to happening. Holy Cross and Bentley did indeed win their semifinal games, but Holy Cross sealed its trip to the NCAAs the old-fashioned way by beating the Falcons in the AHA final, 5-2. Of course, the Crusaders went on to justify their bid by pulling off what most consider the greatest upset in NCAA hockey history, a 4-3 overtime win over Minnesota courtesy of Tyler McGregor’s second goal of the game.
Many more examples of this sort exist, and PairWise weirdness is nothing new. The newest blip might not be on a par with those, but it’s an oddity which gives us a chance to look carefully at how the Rating Percentage Index is calculated. We’ll examine that in Part II.