NCAA Tweaks Pairwise Rankings for Hockey
From College Hockey News
http://www.collegehockeynews.com/news/2011/01/18_committee_tweaks_pairw…
Western College Hockey Blog has a good summary why this is not smart.
http://www.westerncollegehockeyblog.com/2011/1/19/1944371/from-bad-to-w…
This change moved Michigan up from #6 to #5. Notre Dame is tied for 7th. Western is up to 10th (!), Miami at 16th, Alaska at 18th, OSU and Ferris are tied at 24th.
Basically, the ranking's Teams Under Consideration comparision has been expanded from the top 25 RPI teams to all teams with an above .500 RPI. To put that in perspective...Michigan State is now a TUC...wow.
January 19th, 2011 at 5:26 PM ^
Glad to see the boys in blue climbing back up.
January 19th, 2011 at 5:28 PM ^
Jeez, I'm a sophomore and a section 17er, and the college hockey world still baffles me. Between the NHL draft, juniors, and the selection process for the tourney, it's pretty confusing. Maybe by the time I graduate I'll understand everything.
But thanks for posting this. I'm always looking for more knowledge.
January 19th, 2011 at 5:40 PM ^
January 19th, 2011 at 5:45 PM ^
From what I've read, it's pretty much a crapshoot. Right now it helped us, but if State or Ferris or some other middling team falls below .500 RPI, it will hurt us. Likewise, if a team in another conference crosses "the cliff" it will affect the PWR of teams that played that team and not us.
To summarize: Right now, it helped. Down the road, who the eff knows?
January 19th, 2011 at 5:47 PM ^
If every top team takes care of business the rest of the way...the standings could still fluctuate like hell depending which teams get above .500 or drop below .500 in RPI...it has turned into a bigger cluster fuck than it already was...
January 19th, 2011 at 5:57 PM ^
Just win, baby.
January 19th, 2011 at 6:24 PM ^
This could help or hurt us, depending on how the year plays out. It means that more middle-of-the-road CCHA teams will be listed as TUCs*. For instance, this rule change means that FSU, Northern Michigan, and Michigan State were all added as TUCs. This improved our record in this category as we are currently 5-1-1 against these teams, with 3 games remaining against them. So long as we don't screw these games up too badly, this should provide us with an advantage. Furthermore, it pretty much cements OSU as a TUC (we are 1-1-0 against them with 2 games to be played) and leaves LSSU (a team we are 2-0-0 against with 0 games to be played) near the TUC threshold.
On the other hand, there is the possibility that FSU, NMU, and MSU will drop out of the TUC category, and there is the chance that we could lose our remaining games against teams that would not have been considered TUCs under the old system, in which case this would hurt us. However, in all probability, this change will benefit Michigan.
*The Pairwise Ranking system is derived by comparing every team under consideration through a head to head comparison based on 4 categories: RPI, record vs. other teams under consideration, record vs. common opponents, and head to head record. In each comparison, a point is awarded for having a superior RPI, a superior record vs. TUCs, and a superior record against common opponents, and for each head-to-head win. In the event that both teams have the same number of points, RPI rank is used as a tiebreaker. The teams are then ranked in order of number of comparisons won.
EXAMPLE COMPARISON:
Michigan vs. MSU
- Michigan's RPI is .5539. MSU's is .5020. Michigan scores 1 point.
- Michigan's record vs. teams under consideration other than MSU is 8-5-3 (.5938). MSU's is 5-9-2 (.3750). Michigan scores 1 point.
- Michigan's record vs. common opponents to MSU is 10-4-2 (.6875). MSU's record vs. common opponents is 6-7-2 (.4667). Michigan scores 1 point.
- Michigan has beaten MSU twice this season. Michigan scores 2 points.
- MSU has beaten Michigan once this season. MSU scores 1 point.
- Michigan wins the comparison 5-1.
Michigan was compared against every other team under consideration (33 comparisons total) in the same manner. Michigan won 29 of them. This was good for 5th in the country, meaning that if the tornament were held today, Michigan would be the top 2 seed. MSU, incidentally, won 2 comparisons, good for 33rd in the country and would not make the tournament.
January 19th, 2011 at 10:58 PM ^
seriously. helped a lot.
January 19th, 2011 at 5:52 PM ^
it's a little more than a "tweak..." I feel like this will help Michigan, since I want to say that traditionally a lot of the CCHA teams that are middle of the pack teams that Michigan generally beats up on pretty good, are right around 25 in the PWR, so expanding the 25 will generally include a team or two that Michigan did pretty well against (for example Ferris, this year who I believe wouldn be just out of the top 25 of the PWR, but our 3-0-1 against them really helps our TUC. Obviously our 2-1 (soon to be 3-1) against State helps too, but as the OP said, including State this year is just... yeah... at least Ferris is decent)
I'm not sure I really like the new change though, even though it does and probably will continue to help Michigan. It's almost too much I think. No matter what you do, there is going to be something super objective about where the TUC cliff is and depending on where it is, it can make a pretty big difference on the whole PWR rankings. I think I would be a lot more OK with this if they at least cut teams with a losing record. I believe that teams below .500 can't get in anyways, without an auto-bid, even if the committee were to, for whatever reason, decide to not go with the PWR. So, in this case we would drop Bemidji, Brown, Alaska-Anchorage, Northern, Michigan State and St. Cloud State, leaving us with 28 teams, instead of 34. At least then we have slightly under 50% of total programs instead of slightly over.
No matter what happens, deciding who counts as a TUC is one of the most subjective elements of the PWR, so nobody is really going to be happy, but if your RPI (which is about 54% average win percentage of your opponents opponenets and only 25% your own win percentrage ) is only above .500 because you've played a hard schedule, you should definitely not be under consideration.
January 19th, 2011 at 5:55 PM ^
Just keep winning and it won't matter.
January 19th, 2011 at 5:56 PM ^
Michigan has 11 games left. They win 8 of them (which is very doable), they'll get a first round bye in the CCHA tournament and probably have very little to worry about PWR-wise. They could get off to a good start on that by sweeping Alaska this weekend.
January 19th, 2011 at 7:08 PM ^
I know that I'll be in the minority here, but I think that this is a good rule change. It doesn't even come close to fixing the humongous mass of stupid that is the Pairwise ranking system, but it does alleviate one of my pet peeves, namely that the hard cap of 25 teams typically compounded the advantages that the WCHA usually enjoys in the Pairwise rankings.
While the top teams in the WCHA are typically no better than the top teams in the Hockey East or CCHA, the upper-middle tier of that conference is usually the best in the country, often by a wide margin. This gives every team in the WCHA a boost to their RPI, as 75% of the RPI is determined by schedule strength. This wasn't an issue until the hard cap of 25 teams in the Pairwise was installed because it meant that lower tier WCHA teams would be considered TUCs at the expense of upper-middle tier ECAC teams and middle-tier CCHA and Hockey East when a) they weren't that good and b) they weren't really better than the teams they passed in from other conferences.
The effects of this were to give top WCHA teams an undeserved boost in the TUC category. This allowed for all kinds of weird and unfair crap to happen. For instance, Wisconsin made the tournament as an at-large with a losing record a few years ago. And had this not been changed, it would have contiuned to unfairly help the WCHA by allowing every top WCHA team a boost for beating up on a thoroughly mediocre Bemidji State team.
By switching to a cap of .5000, the NCAA removes this unfair advantage that the WCHA has enjoyed for most of the decade. WCHA teams will still get a deserved advantage in the RPI, but this RPI advantage will not have as big an effect on TUC records.
Of course, this doesn't solve the bigger issue of how going 9-4-0 earns a point over 11-5-0, but going 11-5-0 earns a point over 8-4-0, but that's why the NCAA needs to employ more advanced statistical methods if they're going to use a strict statistical system to determine who goes to the tournament. It also slightly exacerbates the issue of how this comparison considers a win over a top team is equal to a win over a mediocre team. But on the whole, I think it'll be slightly better than what we have now.