A look at this weekend’s PairWise Rankings possibilities

UND is currently ranked #5 in the PWR, though the Sioux are definitely looking up the ladder, not down.

Current PWR Rankings

Predicting UND’s PWR as of Feb. 15

Predicting for Tuesday, to include Monday night’s Beanpot games, UND is well-positioned to improve on its current ranking.

A sweep this weekend would almost guarantee a nice climb for UND (leaving only a about a 7% chance of ending the weekend #5 or lower). In that scenario, there would be about an 74% likelihood of UND being ranked #2-#3.

A split would really put UND’s fate in other team’s hands, with anything between #3 and #6 being moderately likely, but leaning toward #5-#6.

Games to watch this weekend

  • SCSU over UMD By far the most important series (other than its own) for UND this weekend. A Duluth loss could give UND the PWR comparison with Duluth (based on RPI and perhaps even TUC). However, SCSU could also become a TUC, giving UND 2 valuable TUC wins. That would be enough to possibly take the TUC comparisons from BC, UMD, and UNH (at least for one week, see “TUC Cliff Revisited” below).
  • UMN over Denver This is important from a defensive perspective, to keep Denver from overtaking UND. Denver wins could help the Pioneers surpass UND in both RPI and COP, depending on UND’s results this weekend.
  • Robert Morris over Mercyhurst This seems to be a pure TUC-cliff play, Robert Morris winning keeps their RPI over .500 and keeps them a TUC.
  • Bemidji State over MTech Another TUC-cliff play. The Beavers aren’t currently a TUC, but with a couple wins they could be, giving UND 2 more wins vs. TUCs.

TUC cliff revisited
The “TUC cliff” is a phenomenon in PWR where a team’s PWR ranking can be quite different depending which teams are “under consideration” (an RPI over .500).

Current RPI rankings

As I noted in this message board post, UND has a winning record vs. the following teams near the TUC cliff:

  • Alaska (1-0)
  • Robert Morris (2-0)

The more of those that become TUCs, the better for UND.

Far more significantly, UND could end with a winning record vs. the following teams near the TUC cliff:

  • Alaska-Anchorage (currently 0-0-1, 2 remaining)
  • St. Cloud (currently 2-0-0, 2 remaining)
  • Bemidji St. (currently 2-0-0, 2 remaining)

UND wants to beat those teams to get the maximum numbers of wins against them, but those losses will make it much harder for those teams to become TUCs. It’s like rain on your wedding day.

End of regular season outlook

With only 8 games left in the regular season, this is starting to shape up.

Edit… the above chart is the percentage share of possible outcomes. If you read those as probabilities, you’re essentially assuming that each team has a 50-50 chance of winning each game. More realistic, if you want to know what outcomes are actually likely, is guessing the winner of each game based on past results. The below chart does that (using KRACH).

The below notes refer to the probabilities of outcomes.

  • Win 4 — Minimum required to stay reasonably safe for an at-large bid.
  • Win 6 — Puts UND about where it is now, likely PWR ranking of #5-#6.
  • Win 8 — Very likely to be top 4 (82% chance).

Current PWR rankings (SiouxSports.com)
Explanation of how PWR mimics NCAA tournament selection (CollegeHockeyNews.com)
USCHO Bracketology Blog (USCHO.com)

5 thoughts on “A look at this weekend’s PairWise Rankings possibilities”

  1. Thanks for doing this Jim – It is really fun to read and think about. You are usually right-on. I was curious how much (historically) WCHA teams have moved up or down in the Pairwise when comparing where they sit at the end of the regular season, and again after the Final Five. Obviously the number of variables goes way up since you have no idea which teams will play in it – but chances are that if the Sioux make it, they would be playing 1 or 2 (or 3?) TUC. So basically, if we ended the regular season at say, #3. Would we have a better chance to move up or move down after the Final Five? Assuming of course that WE make it… Very dangerous to “Ass-u-me” – I know! 🙂

  2. Good observation. I stop at the end of the regular season not because the games after that don’t matter, but because it’s pretty hard to take the set of regular season outcomes and calculate the set of conference tournament games.

    You can browse the historical PWR data here:

    For that season (2009) the WCHA play-in round is the Mar 12-14 games. Plenty of teams did move a couple spots, though UMD plummeted from #11 to #17.

    You can browse past seasons by changing the year in the URL (e.g. to 2008).

    I know the axis labels get squished together on the full-year version of those, making better ranking history screens has been on the bottom of my to-do list for a couple years now…

  3. Your end of season graph where the y axis=probability is missing the ‘win one game’ (red) plot…

    If I read your end of season probability graph correctly; it suggests that, even if the Sioux win all 8 remaining regular season games, they have almost no chance of finishing the regular season in the top spot?

  4. Actually, the simulation never encountered the “UND wins 0 of the remaining 8” scenario. It kind of blew me away, because that’s never happened before, so I ran some numbers. The KRACH-based probabilities for UND losing vs. each remaining opponent are:
    AA 23.29%
    SCSU 21.88%
    Bemidji St 21.04%
    Mich Tech 6.75%

    So, to lose all 8 would be about a 1 in 2,000,000 chance.

    The seeming impossibility of taking the top spot is a big part of why I added the probability-based graph after the initial post. If UND wins all 8, 1 in 10 scenarios have UND come out #1 in the PWR. However, those scenarios are apparently insanely unlikely (I have an increased appreciation for how that’s possible, given the above explanation of how UND was quite unlikely going to go 0 for 8).

    I personally do think the probability chart is the one to believe. It basically states that if the other good teams out there keep winning, as expected, UND needs to win 6-7 to stay where it is or climb.

Leave a Reply