interesting stuff. thanks for posting. my two cents: this post nicely illustrates the absurdity of the pairwise rankings (as i understand them). any ranking system in which a team could be better off losing a game than winning it is beyond repair.
Hockey pairwise update: still a #1 seed.
[Ed: bump. I won't reinvent such a clear, informative wheel.]
Even though Michigan did not play, there were some interesting developments in the Pairwise rankings this weekend. For a more detailed breakdown on how the Pairwise works, see here. Michigan currently sits 4th in the Pairwise, just ahead of Denver, just like they were last week.
Major events that occurred this weekend:
- Denver lost to St. Cloud St. on Friday, dropping their RPI enough that we are still ahead of them. More importantly, we also own the TUC comparison with them. If they win out, they might pass us in RPI, but they can’t pass us in TUC by winning out (as I explain below). So if we win out, we own this comparison.
- Nebraska-Omaha is currently winning their comparison with us, by virtue of a dramatically improved TUC record (a result of Bemidji State dropping out of the TUC field). They sit tied for sixth in the Pairwise
- BC swept UNH, solidifying their comparison win against us and taking away virtually any possibility of a comparison win against a team currently above us. (The path to beating BC in our comparison is so byzantine as to be irrelevant. It is also meaningless, because it wouldn’t change which region the teams are sent to.)
- Michigan State and Ohio State were both swept. Yay! The OSU sweep cost Michigan a little bit, but it was probably worth it.
- EDIT: upsets in the ECAC tournament mean that Union plays Colgate in the quarterfinals of the ECAC playoffs. Had Union drawn 8th seed Quinnipiac, they would have had a chance at getting as many as four wins over TUCs, but now they can only get two. Since the difference in TUC record between the two teams is half a game (same ties and losses, Michigan has one more win), this means that if Michigan wins out, they can't be passed, since any team they play after this weekend would be a TUC (except, maybe, for LSSU, if they somehow get past Notre Dame). Yes, it's weird that this matters, but the important thing is that it was good for Michigan.
This weekend’s important games:
- Michigan hosts either BG or LSSU (EDIT: BG beat Northern, so we play BG). Neither are a TUC, so it would be nice to draw BG, since they are really bad, but either way, a sweep is a must.
- Denver hosts Minnesota State. Other than the Michigan series, this is the series of the weekend for Michigan fans, and your rooting interest is a little strange. Weirdly enough, you are rooting for a Denver sweep. The reason is that Minnesota State is currently just above the TUC cliff, and Denver has a 3-0-1 record against them. Also, Neb.-Omaha is 2-0 against Minnesota State. So dropping them out of the TUC field would actually help us in both comparisons. A sweep would do that; I don’t think Denver winning 2 out of 3 is enough. The improvement in RPI that Denver gets from a sweep is more than offset by the resulting precipitous drop in their TUC record, and virtually eliminates the chance that they could pass us if we win out. If Denver does lose a game, root for them to lose a second one as well.
- Neb.-Omaha hosts Bemidji State. This one is a little awkward as well. A Bemidji win makes them a TUC and really hurts UNO, but helps Denver a bit. Still, as long as we beat Denver in RPI we’re ok there, so go ahead and root for Bemidji State to get at least one win.
- Union hosts an as-yet unknown (and probably bad) ECAC team. EDIT: due to upsets in the ECAC tournament, Union plays Colgate, who is the Bowling Green of the ECAC. It also means that Union only has, at most, two more TUC games left, so if Michigan wins out, it is virtually impossible for Union to pass us. In any event root against Union in case Michigan doesn't win out. You are hoping that someone (probably Yale) eventually takes them out in the ECAC tournament.
- Merrimack hosts Maine, who just crushed them last weekend. Any wins by Maine this weekend will prevent Merrimack from passing us until at least next weekend, so root hard for Maine. This comparison is very close, and will swing Merrimack’s way if they get a sweep.
- Finally, Miami hosts either Northern or Alaska. Any Miami loss will help keep us ahead of them; those two head-to-head losses mean we have to win every comparison, so root against Miami.
Even a Michigan sweep could see them dropping down to 5th or 6th in the Pairwise. Remember that the Pairwise is very fluid, and there is still another very consequential weekend to go after this one. Go Blue!
When I watch a hockey game I root for my team to win. It seems absurd to me to want my team to lose under certain circumstances as some of these fanbases will be. I wouldn't mind if pairwise rankings were used as a criterion for tiebreaks but the way it is now is ludicrous. Maybe they could reform the way that TUC are judged? I have to admit that I'm no expert on hockey but this kind of system is nearly as bad as the BCS.
of the PWR is that is objective and transparent. You know exactly who's in right at the end of the season and you (can, if you put in enough work) figure out exactly which scenarios will get your team in/out. It's all based on the numbers, there's no committee, smoke-filled rooms and all that, no human polls or anything, strictly, objectively based on the criteria.
The downside, of course, is that the criteria can be quite subjective, especially the TUC cliff, which has varied a couple times, but always has the same problems with who is on the edge and you want counted/not counted.
But better this way than, I believe the 1989-1990 season, when Michigan had a team that was expected to make the tournament, barely, for the first time in quite a while and the first time in the Red Berenson era, was passed over, by a committee, for a Wisconsin team that pretty clearly had a worse resume than Michigan. Of course, the Badgers went on to win the title that year, so it's hard to criticize too much, but I believe Red's still a little hot over that and if you listen to various of his radio shows, occasioanlly the alumni guest will touch on it.
The PWR has ups and downs and certainly isn't perfect, but it could be a lot worse. And you almost always want your own team to win, losing to make a team a TUC is going to pretty significantly lower your RPI since it probably isn't a fantastics team and it worsens your TUC record. If someone else has a really bad record against them and your numbers can take the hit, you can find a scenario where a loss might be slightly better. Mostly it's teams you won't play that make up the TUC cliff anyways and you're rooting for other teams that have no realistic at-large potential to win/lose.
For example, Michigan would never want to win in 3 against say, LSSU, even if it would make LSSU a TUC, because Michigan's RPI would plummet, our TUC benefit would only be 4-1 and several other teams we're competing with would get a 2-0 benefit, so the RPI drop isn't worth it.
Now, if Michigan was like 4-0 against LSSU and say Denver was 2-2 against them and Michigan had a significantly higher RPI than Denver, a win in 3 might help Michigan because Michigan would be 6-1 and still have a higher RPI. But it still only really helps if Michigan still beats Denver in RPI, Denver holds some head-to-head advantage over Michigan or the common opponents point and the TUCs are close and Michigan only wins the comparison if Michigan wins the TUC point over Denver.
Same applies if you lose RPI to the team you want to flip with regardless, but don't drop below the team behind you and the benefit to the TUC gives you the comparison, then you might want to lose as well.
But both of those are super rare. And if Michigan plays Bowling Green instead, Michigan has no reason to lose based on LSSU being on the bubble.
Do you get it now? :-p The PWR can be super complex to really get a handle on all its ins and outs, but as with all sports, it still comes down to, "just win". Beat everybody and it doesn't matter who is a TUC.
I definitely see the benefit of taking it out of fallible voters' hands. I think the TUC could use some kind of tweaking though because I find it hard to believe that teams barely on the cusp of tourney consideration should have such a heavy influence on those teams who are at the top of hockey.
For instance, should a .500 RPI team's success/failure down the stretch really determine who the top seeds should be? Teams can get hot at the end of the year and be considered the most dangerous even if they lost a game to a bad team during the first couple of weeks. In this example, I think a system that takes voter opinion into account can be useful because it "marginalizes" the effect of the mediocre teams on these higher seeds. I'm struggling to articulate my point here, but the gist is that I think a voter system would be a positive influence because it would stabilize the effects of, for example, a LSSU sweep of FSU on Michigan's resume if both LSSU and FSU were just barely TUC.
I do not advocate complete abolition of the PWR. Rather, I think the system could be improved by adding a voter element to it in these instances. Maybe some kind of adjusted scale could be used where the .500 cutoff wasn't so strict. This change would recognize the fact that a .497 team isn't all that different from a .504 team.
All things considered, these are the musings of a relative newcomer to hockey fandom. No system can be perfect, clearly. I'm just throwing my thoughts out there. UM has a shot at a 1 seed, they're awesome, and I hope they bring home a title this year!
The change this year in who qualified as a TUC was from the RPI top 25 to teams above 0.500 in RPI. This added anywhere from 3 to 7 teams, depending on which side of the "cliff" certain teams ended up. The problem with adding these teams is that you are adding mediocre teams against whom a contending team might go 3-0-1 or 4-0 in four games. If contending teams are more like 2-1-1 against a team at the cliff, whether they are above or below it is largely irrelevant. It also becomes conference-dependent; if the CCHA has a number of mediocre teams below the cliff, and the WCHA gets them just above the cliff (sort of like it is now), that can hurt CCHA teams both in seeding and, in Western's case, in qualifying for the tournament. The more teams you add, the more important the TUC cliff is likely to be. Also, having the TUC cliff so low means that a team's record against a Ferris State or Alaska is equivalent to its record against Boston College or North Dakota, so the TUC comparison is also highly schedule-dependent.
I would rather they made it the RPI top 20, since those are really the teams under consideration. No one outside that group has a legitimate chance to make it on Selection Sunday. Also, there is a lot more parity among those teams than there is if you include the top 30 or so, so it is a more equitable comparison. Yes, the cliff is still an issue, but it keeps the metric a simple one and the oddities like this weekend's Denver series become far less common.
thanks for posting these. I can never keep up with the Pairwise and RPI and all that in college hockey. you make it much easier. I'll just root for M to sweep through the CCHA playoffs and secure that 1 seed.
Agreed, but at least it's black and white, unlike basketball.
And thanks, mfaninohio, for posting. I had zero understanding of hockey seeding and pairwise until these updates came along.
Quick update on the NMU-BG game:
The teams are tied at 1-1 heading into the 1st overtime period.
Let's hope for an exhausting quadruple overtime win for BG.
Then the last place falcons can come into Yost on 4 days rest and get demolished.
Not that it will matter because Michigan will take care of anyone who comes to town next weekend.
Whoa there, nothing is as horrible as the BCS as this is a system which sets up a playoff bracket, instead of divying up tens of millions of dollars between 5 conferences.
Glad to see another M-fan in Ohio representing so well. Keep it up man.
Root against Miami.... duh
Root against Union... DONE (Rt 7 Rivalry FTW!)
Root hard for Maine... not gonna happen
you can root against Merrimack instead of rooting for Maine. Also remember, "the enemy of my enemy is my friend."
that set up some pretty rare loopholes. I don't know all of the particulars, but IIRC, Notre Dame wouldn't have qualified for the NCAA playoffs in 2008 had they not LOST in OT to Miami in the CCHA semi finals that year. Of course, they would have played Michigan in the final had they won and anything could have happened to earn the autobid, but just the same, the Irish got in as a 4th seed in the NCAA's (like 14th overall) buzzed their way through the tourney and eventually did beat Michigan in the national semi-finals in OT.
I think there was also a case in the past where Wisconsin was better off losing a series in the WCHA than winning in three games to also qualify for the NCAA playoffs due to the make up of the PWR/TUC etc.
These are the loopholes the NCAA is trying to close, although the recent tweak to the TUC cliff is intended to level the playing field for teams in conferences that don't have the benefit of playing as robust a schedule (with as many TUC win opportunities) as say the WCHA or HEA.
The WCHA fans are howling about how unfair the PWR is, especially when one of their teams gets screwed out of the NCAA's in favor of say a mediocre ECACHL team because of the anomalies the PWR criteria has. I think Brian isn't a proponent of this system for many of the same reasons... you want the best 15 + the Atlantic Hockey League Rep in the tourney. But, in a small measure of justice for their arrogance, anytime I see a WCHA team get screwed, it makes me smile.
I can't wait for the day we get a football playoff so we can play RPI / TUC / PWR / QWIGIBO games and cheer for the team ahead of us to win.