FanPost

The Pairwise Rankings: And you thought the BCS was weird.

[Bumped.  Handy, pocket-sized guide to MSU hockey rooting interests at bottom. -KJ]

As the ice-rink version of March Madness approaches, hockey fans have a much better idea of where their team sits and what their chances of making the tournament are than basketball fans do. The reason? A ranking table called the Pairwise Rankings. Unlike the basketball selection committee, which uses votes by ten members to determine the at-large bids and seeds and allows committee members to weigh various factors differently, the hockey selection procedure is quite rigid. The Pairwise Rankings began as an attempt to mimic the committee's decision-making process and does so quite well - to date, it has predicted the field exactly every year.

After the jump, an explanation of the Pairwise Rankings, their oddities, and what it all means for MSU hockey this year:

The Pairwise Rankings are limited to the top 25 teams in the RPI, known as "Teams Under Consideration". Each TUC is compared directly against the others as follows:

  • RPI: Better RPI gets one point.
  • Record vs. TUC: Better record gets one point, as long as both teams have played at least 10 games against other TUCs. (Head-to-head matchups are not included here.)
  • Common opponents: Better record gets one point. Here there is no minimum number of games, and every game against a common opponent counts even if one team played more games against a particular common opponent than the other.
  • Head-to-head: One point for each win over the other team.

Teams are then ranked by the number of comparisons against other teams won; comparisons that are tied go to the team with the better RPI. Teams who have won the same number of comparisons are ranked by RPI as well, although on the bubble head-to-head comparisons may be used to break ties instead. This ranking is used to choose the at-large teams and seed the field 1-16 (swaps within a seed line are common to accommodate home teams or avoid first-round conference matchups, but teams are not moved up or down a seed line unless there is no alternative).

The Pairwise Rankings are notorious for their volatility; Brian at mgoblog wrote a detailed explanation of the flaws three years ago. To summarize:

  • Assuming no head-to-head play (or a split of the series), a huge edge in one category can be totally wiped out by losing the other two by a tiny bit. An example: Vermont owns the comparison with St. Cloud State at the moment despite an RPI difference of over .03, thanks to one extra TUC win (8-8-3 instead of 7-8-3) and the lack of a tie in common opponents (2-1-0 to 2-1-1).
  • TUC record can fluctuate severely if teams near the cutoff line jump in and out of the RPI top 25. Nebraska-Omaha's sweep of Michigan last weekend dropped the Wolverines out and moved UNO in; as a result, our TUC record changed from 7-6-2 to 5-5-3 without us playing a single game. This can lead to some perverse incentives: in the second round of the CCHA playoffs, which is best of 3, it can sometimes be better to win the series 2-1 than 2-0, particularly if you had a good record against that team in the regular season. If a 2-0 sweep drops them out of the top 25, that record will no longer help you in TUC comparisons, so it may well be worth the RPI hit from a loss to keep those wins in the TUC count. (As it happens, in 2007 there was a scenario on the final day which could have seen us left out of the tournament altogether - but it required us to tie the third-place game. A win or loss meant we were safe, the latter because it would have brought our excellent record against Lake Superior State into play for TUC points.)
  • Common opponents puts a massive emphasis on a tiny number of non-conference games. Losses to Minnesota and Wisconsin in the College Hockey Showcase make it much, much harder to win any comparisons against WCHA teams, as the common opponent point is likely lost. (A win over Michigan Tech offsets this partially, but Tech is so bad that we're probably better off not letting WCHA teams count their regular demolitions toward common opponents.) Meanwhile, wins over Rensselaer and Clarkson give us a big edge over ECAC teams in the Pairwise. A split with Maine probably helps against Hockey East, as Maine is near the top of the standings (so most teams will be below .500 against them).

Just this past weekend, with no games of our own to affect the results, we moved from 13th (barely on the good side of the bubble, as there are usually two auto-bids below 16th) to 9th to 11th. At the moment, aside from "win to get our RPI up and pass people that way", some things that would help MSU's chances of reaching the tournament and getting a good seed:

  • Sweep Ferris State. With them having the RPI edge and our common opponent edge hanging on by a thread (12-7-5 instead of 12-8-4), we could use the security of two extra points against them. Even 1-0-1 isn't good enough if they flip the common opponents, as RPI would break the tie in their favor.
  • Get Michigan back into the top 25. Painful as this may be, we could use the boost of our 3-1 record against them in the TUC category. Even without an RPI move, this could flip our comparisons with New Hampshire and possibly Yale (if they pick up a couple of TUC losses to allow the comparison to count). Similarly, we want to keep Minnesota and our 0-1 record against them out, keep Lake Superior State and Nebraska-Omaha in (1-0-1 against each), and drop Northern Michigan (1-1 split) and Alaska (0-1-1).
  • Ideally, we'd like to draw an opponent in the CCHA second round who is safely into the top 25 so we can sweep them without knocking them out. If it's someone farther down, prefer Alaska or NMU so that if we do knock them out, we're knocking out a loss as well as the wins. We do not want Michigan here, as it's possible we'd push them out if we sweep. Best case is to get Michigan to do just enough to stay in.

As the regular season ends and conference tournament play begins, I'll revisit this and look at MSU's rooting interests around the NCAA.

This is a FanPost, written by a member of the TOC community. It does not represent the official positions of The Only Colors, Inc.--largely because we have no official positions.