Playing the tournament percentages

Whether your favorite team used the first half to build a foundation for the rest of the season or only managed to dig itself a hole, plenty of work remains over the next couple of months in order to stock a worthy highlight video.

Assuming that your team of choice has tournament dreams, how realistic are they? The only true measure of whether or not a team will catch the fancy of the NCAA selection committee is the PairWise Rankings (PWR), a set of criteria and rules for their application that is admittedly short of the federal tax code in complexity, but is still likely more convoluted than one may want while nibbling on holiday baking and contemplating placing a wager on an upcoming bowl game.

A far simpler method for determining if a team’s current pace would be sufficient to secure a slot in a national tournament bracket is to merely consider winning percentage and compare it to the historical threshold for receiving tournament bids. Obviously, such an approach is inherently flawed, because if the field as determined by winning percentage mirrored the PWR selections, then any win over any opponent would be just as valuable and the NCAA could jettison the PWR.

For an example of the deficiencies of raw winning percentage as a field predictor, one need look no farther than the current PWR, where Minnesota-Duluth would qualify eighth with a winning percentage of .575, while Northeastern’s .750 has it on the wrong side of the bubble. Robert Morris owns an even better .800 mark, yet its Ratings Percentage Index (RPI) fails to land it in the top 12 or merit consideration.

However, society likes to talk about percentages. Watch your local weather, where the forecaster will confidently point at a graphic and authoritatively pronounce a 20 percent chance of precipitation as the rain falls steadily and shows no sign of stopping. Or the doctor will tell you that your odds of survival are 40 percent in the same tone that he used two months prior to deliver the good news that your disease was in remission. How are these percentages derived, and what do they even mean? By comparison, our hockey winning percentage discussion is downright scientific.

The field expanded to eight teams for the 2005 tournament, meaning seven years worth of data exists to consider. During that time, the lowest winning percentage for the last team into the field was .588, by Harvard in 2006. A couple notes about this value are worth mentioning. The Crimson would have advanced to the NCAAs in any case, because they won the ECAC tournament and the conference autobid, but as I remember it, their postseason run had moved them into the top eight in the PWR as well. The second-lowest cutoff mark, .622 by Boston University, came in 2010. Both 2006 and 2010 were Olympic years, indicating that the absence of elite players for weeks or even an entire season serves to increase parity. By comparison, the eighth team into the tournament has had a maximum winning percentage of .667, a median of .656, and a mean of .646, suggesting that the results are rather consistent beyond the Olympic year outliers.

Those numbers cause one to theorize that UMD’s current .575 pace will not be sufficient for a tournament berth. That makes sense when looking at their schedule, as the Bulldogs played six games, and lost three times, to highly-ranked teams Boston College and North Dakota, but aren’t scheduled to face either in January or February. Suppose that UMD repeated the rest of its results from the first half:  sweeping St. Cloud State and Minnesota State, splitting with Ohio State and Bemidji State, and being swept by Wisconsin and Minnesota. Assuming two wins in the WCHA quarterfinals and a loss in the WCHA semifinals, UMD would come into Selection Sunday with a record of 19-15-1 for a winning percentage of .557. It is very doubtful that would warrant an at-large bid. Let’s assume that the Bulldogs would need to achieve a percentage more in line with the mean of past years. How many losses would they need to eliminate from the seven just projected for the remainder of the season? Rather than losing six more times during the remainder of the WCHA regular season as in the scenario above, if UMD reduced its losses to only three before a WCHA semifinal loss, then the resulting winning percentage increases to .643. The strength of their conference should allow them to qualify with that record, and perhaps even with an extra blemish or two, but that would be the lowest winning percentage to get an at-large berth outside of an Olympic year.

Looking at Boston University next, the Terriers currently sit at 10-9-1 with 14 games remaining on the schedule. Assuming that BU plays two additional games in the Hockey East tournament, 13 more wins would produce a record of 23-12-1 and a winning percentage of .653, likely good enough to advance. Prior to the postseason, the Terriers have a total of four games remaining with ranked teams BC, Northeastern, and Harvard, so their fate may hinge heavily on how they perform in three games versus Maine right out of the break.

Suppose the goal is not merely to qualify for the tournament, but to host an NCAA tournament quarterfinal. Historically, the fourth seed has qualified with a winning percentage as high as New Hampshire’s .847 in 2007, to a low of the .703 turned in by Harvard in 2010. Again, the two lowest values came in Olympic years, the 2010 Harvard record and a .724 by Minnesota in 2006. As with the eighth spot, the outliers impact the mean more than the median, as those values for four seeds come in at .770 and .779 respectively. To date, those values are holding true, as Mercyhurst at .778 would host, while Boston College at .710 would not.

For teams hoping to enter the tournament as the top seed, the bar is set high. Harvard posted the best winning percentage with a lofty .970 in 2008, while Wisconsin’s .882 in 2009 was the lowest. That didn’t prevent the Badgers from taking home the title, while the Crimson fell in the semifinals; ironically, UNH, the owner of the second-best winning percentage at .929 in 2006, also was eliminated in round two. The other three top seeds with winning percentages above .900 — Wisconsin in 2007 and 2011 and Minnesota in 2005 — went on to claim the championship. This season, only the Badger’s mark of 19-1 for .950 is in line with any precedent.

While not a perfect indicator, raw winning percentage can still act as a rough guideline of a team’s postseason potential. If the picture painted is dismal, then depending on your conference affiliation, fix your sights on the automatic bid or adopt the time-honored mantra: there’s always next year!