fanlogs.com: Sioux from afar A look at UND's PWR and playoff chances

Sioux from afar

Index

Search this log

About PWR and UND's ranking

College hockey is not plagued with a BCS or opaque tournament selections coming from a smoke-filled room, rather the sport enjoys a well-documented (if not universally apperciated) selection criteria. In the late '90s, USCHO coined the term and method of expressing those criteria as Pairwise Rankings (PWR). For my fellow stats geeks, I've got some pretty nifty detailed calcultations of UND's PWR on SiouxSports.com.

Though the tournament invites 16 teams and UND is currently tied for 12th, it's grip on 12 is tenuous. UND currently wins 15 comparisons against the 27 other teams under consideration (RPI greater than .500), but dropping just to 14 would put it in a 3-way tie with the teams currently below it (thus invoking RPI, the importance of which will become a tiresome theme of this post, as a tie-breaker). Also, while teams in the top 16 of PWR often secure some of the conference autobids, it's quite likely that a few teams out of the top 16 will secure tournament selection via conference championships, preventing teams near 16th in PWR from being selected.

PWR comparisons at high risk

NoDak vs. Colgate (2-1)
Deciding factor - RPI .5559 vs. .5504
UND will keep the COP point, Colgate's lead on TUC is tenuous

NoDak vs. Dartmouth (2-1)
Exact same story as Colgate, but the RPI's are .5559 and .5402

NoDak vs. Ohio St (2-1)
RPI is .5559 to .5497

NoDak vs. Northern Michigan (2-1)
RPI is .5559 to .5340

What will happen to UND's RPI?

Assuming UND's three remaining opponents perfectly maintain their current rankings (minor deviations will have insignificant impacts on UND's RPI), UND's opponents' win percentage will fall from .5742 to .5635 and it's opponents' opponents' win percentage will fall from .5251 to .5224.

UND's control over its destiny is its win percentage, which will be the primary driver of its RPI, as the following table elucidates:

Additional
Wins Final
Win% Approximate
Final RPI

1 .4861 .5339

2 .5139 .5408

3 .5417 .5478

4 .5694 .5547

5 .5972 .5617

6 .6250 .5686

While UND's opponents will also be moving targets, these numbers suggest that UND would need 4 wins in its remaining 6 games to continue to win all the "at risk" comparisons listed above. However, UND has an ace in the hole, RPI bonus points. UND has 2 road quality wins (from a season-opening sweep of Maine that seems almost like a different season). While the exact value of those bonus points is unknown, past results suggest that it's probably between .006 and .010 RPI points (the impact of which is illustrated by the chart above).

How many wins does UND need to make the NCAA tournament?

If you read this far, you should have some idea of that yourself, and know that there isn't a definitive answer. If you decided to skip all the numbers, ceteris paribus, I would speculate that UND needs 3 or 4 wins in its remaining 6 games to have a good shot. Caveat lector -- PWR has some inherent volatility that can make close comparisons take unexpected flips; these numbers assume all other teams will continue to perform exactly as they have to date, an assumption the other 57 D-I hockey teams are unlikely to oblige.

Other things to watch out for

As hinted at above, PWR has an inherent volatility that more formulaic ratings like RPI lack. Purely formulaic ratings are continuous, game results push teams up and down in decreasing increments as more games are played. PWR incorporates a concept of "teams under consideration", those with an RPI over the arbitrary cliff of .500. Because one of the criteria for PWR is record against other teams under consideration, seemingly unrelated contests that push teams over that cliff can create some unusual shifts throughout PWR.

Michigan State has a pocketful of quality wins, 11 PWR comparison wins, and a few PWR comparison losses decided by slim RPI margins. With remaining series at Alaska-Fairbanks, Ohio State, Bowling Green, and Notre Dame, the Spartans could join the battle for one of the last tournament spots.
St. Cloud and Miami are likely already TUCs, given their quality win bonuses and near-.500 RPI's. UND already has 2 wins against St. Cloud and closes their season with an additional series. Michigan Tech, Massachusetts, and Minnesota-Duluth are also knocking on TUCs' door. UND is 1-1 against Duluth and 2-0 against Michigan Tech. Increasing UND's winning percentage against TUCs could help buttress it's comparisons to Dartmouth, Mankato, and Ohio State; give UND the comparison win against Mass. Lowell; and possibly allow the Sioux to gain the comparison win against Michigan State. See USCHO for a hands-on calculator of how much RPI bonuses (or other shifts in RPI that affect TUCs) can affect PWR.