March 10, 2008
NCAA tournament Bracket ABCs
Analysis, Breakdown and Comparison for March 10
by Adam Wodon/Managing Editor
Less than two weeks now before we know the field for the NCAA tournament. And you'll be able to figure it out right along with us. In particular, soon you'll be able to utilize our "You Are The Committee" application to put in your predicted results for upcoming conference tournament games, and see how that would affect the Pairwise (i.e. the selection and seeding for the NCAAs).
Before we get into the rundown of the current situation, I wanted to point out something that I see as pervasive among the online chatter about the NCAA tournament selections. Things have come quite a long way in regards to education on how the process works, even with all the changes over the years. The education process that those of us currently at CHN first started when with USCHO in 1997, first educated coaches themselves and media members. The public was ahead of the curve. Later, the "insiders" got it straight mostly, and the fans are way up to date — the occasional conspiracy theorists notwithstanding. When anyone doesn't get it, the online community is usually quick to make a correction.
But I still see a common error.
Last year, we were the only publication to accurately describe how the Pairwise tie would be broken. But that's not the error I refer to.
(See our primer on what the Pairwise is, and how it works to select the NCAA field.)
What I see is many people still trying to arrange their predicted brackets by placing too much emphasis on attendance concerns. They'll flip this team and that team, and assert that the committee will do this because of attendance. Then, invariably, they are surprised that the committee didn't. Sometimes, this even leads to cries of "conspiracy."
This was most notable in 2003, the first year of the 16-team field. At first, no one could figure out what the committee did, because it went against many past practices. Soon we discovered that the committee simply laid the teams out in a strict 1-16 seeding (according to Pairwise), and placed them in a 1-16, 8-9, 2-15, 7-10, etc... pattern — adjusting for first-round intra-conference matchups and other sacrosanct issues.
Since then, the committee has steadfastly stayed true to this principle. It wavered only slightly last year when it kept Air Force in the West for a 1-15 matchup against Minnesota, instead of having it be 1-16.
It is certainly possible the committee will waver from this philosophy this season. It can do anything it wants. But we have seen no evidence in five years that the committee will care at all about attendance concerns, as it used to do in the 1990s. Until we see that happen, we'll assume the committee will just go with the truest brackets possible.
Remember too, though — we have argued this isn't necessarily wise. I'm just saying what the committee has done. There is still merit in tweaking the seedings to get better matchups, avoid second-round intra-conference games, and so on. We support such tweaking when it makes sense (subjectively, of course), because the Pairwise is too imperfect a system to be so beholden to it when it comes to seeding.
But this is what the committee has done every year for five years now.
On to the current list ...
1. Colorado College
The only change over the top four from a week ago is CC flip-flopping places with North Dakota at No. 1 and No. 4. Colorado College's two-game sweep of Denver pushed its RPI up enough to surpass that of idle Michigan and North Dakota (two ties vs. St. Cloud State). That alone wouldn't have flipped CC's comparison with North Dakota — but CC also gained on the Sioux in Record vs. Common Opponents.
This flip-flop points out the precarious nature of these seeds right now, as they certainly can all easily switch around again. None of the teams in the top four will play a TUC (Team Under Consideration) this week, but if any of them stumble, the RPI will be affected. Because No. 5 Miami also doesn't play a TUC this week, the top four should stay stable, though that will change down the road if Miami beat a TUC at The Joe (more below). But placement within the top four is in flux.
2-3-4. Michigan, New Hampshire, North Dakota
There's very little New Hampshire can do to lose comparisons to CC or North Dakota, or to win comparisons it's currently losing to Michigan and Miami. The only possibility is if North Dakota plays and beats CC, while UNH plays and beats Northeastern or BC. Neither will happen this weekend, but could happen the week after.
Assuming this is the top four, then we deal with placement. Colorado College must be in Colorado Springs for the West Regional, by rule. By logic, New Hampshire will be in Worcester at the Northeast Regional. That leaves the other two. At this point, I believe this will come down to whether Wisconsin makes the tournament, and which team, Michigan or North Dakota, ends up higher seeded overall. That's because Wisconsin must be in the Midwest Regional in Madison; and Wisconsin is a likely No. 4 seed. Then we run into what's become a relatively common scenario since the 16-team tournament: Do you "protect" the No. 1 seed? The answer has generally been "yes," and we agree.
In other words, the difference between a 4 seed like Wisconsin and a 4 seed like Army is, no offense, pretty big. So if a team like, say, Michigan, is the No. 1 overall seed, it should get "protected" and not have to play Wisconsin in Madison. If that means moving Michigan to the Albany Regional (slightly farther away but still a flight), then so be it.
But here's another wrinkle ... North Dakota and Wisconsin are in the same conference. This complicates matters. Now it's not just about protecting Michigan, but also about avoiding first-round intra-conference matchups. In 2003, the first year of the 16-team tournament, this exact scenario happened, and the committee pitted No. 1 overall Cornell against Minnesota State instead of switching Minnesota State to play a WCHA first-round team, or changing its seed. Cornell went ballistic — though it made the Frozen Four anyway.
If Michigan were the No. 1 overall seed in this scenario, the committee could pull New Hampshire out of Worcester and solve the problem. But that's another ball of wax. Should be interesting to keep an eye on this.
Any way you slice it, with six or seven WCHA teams poised to make the NCAAs, there will be all sorts of issues to deal with.
5. Miami
Now, take everything we wrote above, and scrap it, because Miami is poised to be a "savior" to the semi-nightmare scenario explained above. First, Miami must get past Bowling Green this weekend. Then, it must split at The Joe, as long as the win comes against a TUC.
Why does this matter? Miami's record vs. TUC is currently 6-2-1. Since it doesn't meet the 10-game minimum threshold, then that portion of the comparison doesn't count. If it counted right now, then Miami would flip the comparison with both Colorado College and North Dakota, and would be the No. 2 overall seed.
All Miami needs to do to become a No. 1 seed is split at The Joe. Why? Because once it does, it will pass either CC or North Dakota, or both, depending. Why? Because CC and North Dakota can't both win out. One of the two will lose a game vs. TUC, whether that be in the semifinals, or to each other in the WCHA final — and that will keep the record vs. TUC of one of those teams below Miami's.
However, here's another rub. North Dakota plays Michigan Tech this weekend. If North Dakota sweeps that series, then splits in St. Paul, then that's a 3-1 record to the record vs. TUC, and keeps them above Miami no matter what. So both teams could then stay ahead of Miami, unless Miami wins the CCHA tournament.
However, if North Dakota sweeps Michigan Tech this weekend, that is likely to knock Michigan Tech out of TUC status (A TUC is a team in the Top 25 of RPI). Tech is currently No. 23 on the list, with other teams on its heels. If that happens, that wipes out what would be 5 wins vs. TUC for the Sioux, and, again, they would lose out to Miami.
Of course, there's another scenario — Miami passes CC individually, but still loses out to it in the overall Pairwise.
But, if Miami cracks the top four, it solves some issues expressed above.
6-10. Denver, Clarkson, Boston College, St. Cloud State, Michigan State
Let's take a breather here. These guys look safe for the tournament, except, necessarily St. Cloud. The thing that insulates the non-WCHA teams from trouble, is that four of the five teams just below this range are WCHA teams. They can't all go on a run in the playoffs. These teams are all going to lose games, in some cases to each other. Unless any of these teams — BC, Clarkson, Michigan State — get swept in their best-of-3 series this weekend, you can pencil them in.
St. Cloud State, on the other hand, could lose an awful lot of comparisons in a hurry if it gets swept by Wisconsin this weekend, and things break right elsewhere. It would take a big drop and some upsets to fall out of it completely, though.
11-13. and 15. Minnesota State, Minnesota, Wisconsin, Minnesota-Duluth
There's four WCHA teams here, with three on the positive side of the bubble. It's virtually impossible to sort all of the combinations out, because there are so many, depending on what happens in the WCHA playoffs. It's not rocket science to suggest that those teams which win this weekend will be in good shape, and those that lose may be in trouble.
Without a bunch of upsets in other conferences, it does seem almost a lock that six teams will get in, and probably seven. That's because there's very little downside pressure on teams like Minnesota coming from those just outside the bubble — i.e. Vermont, Providence, Harvard, Notre Dame.
Certainly, one of these will not make it. With the right mix of wins and losses, another will get bumped out. But it's hard to see three of these slipping.
14. Boston University
The "TUC Cliff" (see the primer) comes heavily into play here. Remember, again, that a TUC is any team in the Top 25 of RPI. And "Record vs. TUC" makes up one of the four Pairwise components. The last two teams in the Top 25 of RPI are Northeastern and Providence, and BU is a combined 6-0-1 against those teams. If those teams stumble this weekend, they could drop out of the Top 25, and BU's 6-0-1 goes out the window.
Two things could help BU, one in its own control.
First off, the No. 26 team in RPI is currently Massachusetts-Lowell, which is the team BU plays this weekend. By beating Lowell, BU can help keep Providence and Northeastern in.
The other thing is, at No. 27 is Union, which plays Cornell (No. 21). BU defeated Cornell this year, so it wants to keep Cornell in the Top 25, and it wants to keep Union down below Northeastern and Providence. So BU is rooting for Cornell this weekend.
To the upside, BU has very little chance of catching many teams, other than Minnesota.
16-18. Notre Dame, Vermont, Harvard
The best chance is for one of these teams to catch BU. Basically, it just means, these teams need to play better than BU over the next couple of weeks. These teams don't have a real shot of catching any of those WCHA teams.
When you look at things like Common Opponents records, it's easy to see why the WCHA has that Pairwise superiority. It strikes me as completely legit. There is merit in the committee deciding that each conference should have a five-team cap, or something like that. But as it stands now, there's no reason to suggest that putting seven WCHA teams in the tournament is a sign that the Pairwise is flawed.
One thing the above scenarios seem to prove is that the WCHA is by far the toughest conference from top to bottom. Evidence supported by the WCHA's dominance in winning the NCAA tourney. The conspiracy theorists would have a field day if the committee bypassed a WCHA team in favor of an Eastern team based on the perceived east coast bias. I'm sure the NCAA sweats every year when the WCHA places so many teams highly in the pairwise as well as KRACH.
Mar 10 2008, 6:13 pm by Gregg P. Hartford, WI
Another noteworthy stat is the fact that EACH of the WCHA teams has played 2-3 TIMES the number of TUC games as most others (e.g., Miami and, to a lesser degree, Michigan and Michigan State). The more QUALITY teams you play, the more likely your TUC winning percentage will drop -- life in the WCHA is cannibalistic by definition when it comes to Pairwise. Another key Pairwise comparison is UND v. UNH -- all decided by the fog-shortened 2-period tie in Boston early in the season. If UND had won that game, UND tips the UNH comparison in its favor. One final note -- the two wins UND has against potential CHA champion Bemidji State will be very positive for UND's TUC numbers, even if MTU drops out of the TUC class, should BSU win out. A loss to MTU this weekend, coupled with two wins by the Sioux might be a GREAT stroke of "luck" for the Sioux in the end. Plus, the conspiracy theorists will have more to whine about while their teams stay home come the last weekend of March
Mar 11 2008, 12:07 am by Sioux_Me, Grand Forks
Thanks for the good insights. However, the Bemidji wins won't help North Dakota. Bemidji will not become an automatic TUC by winning the CHA. Teams don't become automatic TUCs anymore.
Mar 11 2008, 12:15 am by CHN Editor
Sioux_Me, you are correct "IF" they won the game, but "IF" they lost that would tip the scale just that much larger in UNH's favor. If we lived by "WHAT_IF's" then lots of things would change. The main adavantage for UNH is due to the split with UND and the sweep of CC and the clear superiority in the HockeyEast. I welcome any match-up against a WCHA team.
Mar 11 2008, 9:46 am by JJ - Lowell, MA
Adam:
I ran some PWR scenarios with MTU losing 2-0 to UND and, unless there are upsets out east, MTU stays 23 in RPI.
One question - is it better for UND to have MTU drop as a TUC but have NE stay a TUC? The Sioux are 1-0 v NE and could be 5-1 v MTU.
Mar 11 2008, 10:21 am by GChristian, St. Paul
GChristian - those are great points ... definitely, if you swapped 5-1 for 1-0 - the 1-0 is better, which is part of the problem with the system. But if you only lose the 1-0 and keep the 5-1, then just losing the 1-0 may not hurt too bad - though it would hurt vis-a-vis Miami.
Mar 11 2008, 11:24 am by CHN Editor
Ah, yes -- the exceptions to the general rules and the changed rules rear their ugly heads again. Now I know why I like the KRACH rankings better! Thanks for picking up on my error. In the end, if a team wins out, there is not much that can happen to it in terms of Pairwise negatives -- unless they were a bubble team to start! Keep up the great work!
Mar 11 2008, 12:49 pm by Sioux_Me, Grand Forks
Could it potentially help UND to drop one game to Tech this weekend?
Mar 11 2008, 12:57 pm by Nate, Grand Forks
"the clear superiority of Hockey East"?!? How precious.
Mar 11 2008, 5:25 pm by JM, Minneapolis
'The clear superiority of Hockey East'?!? How precious ... Look again. What JJ in Lowell was (I think) trying to say was that UNH's record of 3-1-0 vs. CC and NoDak, coupled with their clear superiority *within* Hockey Easy (having clinched the RS title with 4 games to play), suggests they're on a roll and WCHA battle-tested, certainly more so than any other HEA team. Time will tell if those results translate to post-season success ...
Mar 11 2008, 8:40 pm by JL, upstate NY
I reviewed the Pairwise definitions - and Bemidji State's schedule (12 games v. current TUC teams) - and think they ARE a TUC team if they win their conference tournament this Sunday. Am I still missing something?
Mar 12 2008, 8:25 am by Sioux_Me, Grand Forks
That doesn't make you a TUC. Check out our Pairwise Primer (linked above). A TUC is a team in the Top 25 of the RPI.
Mar 12 2008, 8:45 am by CHN Editor
Directly from your "Primer" -- something you might want to consider reading sometime (again): "The NCAA mandates that a conference receives an automatic bid to the NCAA tournament if it exists for at least two years and has at least six teams (Note: CHA, with five teams, is currently receiving an exemption)." This portion of the "Primer" is followed by the TUC definition: "Team Under Consideration: A team under consideration is one which has an RPI of .500 or higher. (Modification, 2006: TUC is a team in the top 25 of RPI. Also, in order for this component to matter in the comparison of one team to another, each team must have played at least 10 games against other TUCs.) Further, if a team wins its conference tournament, and is thus an automatic qualifier to the tournament, it is added to the list and automatically becomes a TUC for purposes of this criterion." Am I missing another "double-secret, we didn't mean to publish that criterion" adjustment to what appears to be clear?
Mar 12 2008, 1:03 pm by Sioux_Me, Grand Forks
I did qualify my comments above by stating "IF Bemidji State wins its conference championship" -- scheduled for Sunday, March 12th (catch it live on CSTV!) -- that would then make it an automatic TUC team. The same holds true for Army (or Air Force, RIT, or Mercyhurst/Sacred Heart (should that winner slip past Army) if it wins the AHA, eh?
Mar 12 2008, 1:08 pm by Sioux_Me, Grand Forks
Yeah, the primer needs to be fixed. Apologies for the error. But the information I previously stated is accurate. They are no longer automatically a TUC by virtue of winning the conference title.
Mar 12 2008, 1:12 pm by CHN Editor
Thanks JL, that is what I meant by it and you're definitely right about time will tell. If history has anything to say about it, I will not see them in Denver, but I can say they've looked a lot better this yr than years past.
Mar 12 2008, 5:02 pm by JJ - Lowell, MA
All of our collective "what-if's" will quickly become "coulda, woulda, shoulda's" -- or the ultimate, "told ya!" -- over the next couple of weeks. Needless to say, if you can't find quality D-I hockey to watch now, you're never going to find it. Any way you slice and dice the brackets, there will be "winners" or "losers." In the end, that's why they play the games anyway! Even the most die-hard WCHA fans have got to admit that simple numbers means intra-conference match-ups must occur at NCAA Regional sites -- one or two as first-round match-ups if we get six, seven or (it almost seems downright greedy) eight into the NCAA Tournament field.
Mar 12 2008, 9:25 pm by Sioux_Me, Grand Forks
Oops! I forgot about the "Holy Cross Rule" -- the rule enacted to prevent clearly undeserving teams from theoretically sneaking in the "back door" of the NCAA Tournament (via TUC comparisons). It was a moot point because Holy Cross won out in their conference and came in the front door. Ultimately, the question was addressed on the ice -- where it should have been. It is ALMOST ancient history, but I do believe the Crusaders crucified the Gophers at that year's Regional (no, not too strong a phrase under the circumstances!). It is all coming back to me now...
Mar 12 2008, 9:38 pm by Sioux_Me, Grand Forks
Adam,
Not sure how you figure that "BU has very little chance of catching many teams, other than Minnesota".
Assumptions:
If BU loses two this weekend, they're not only not catching anyone, they're done.
If PC and NU drop off the TUC list, that will damage BU's comparisons to the point where they will need the HE auto-bid.
Either of those happening makes any other discussion moot, so let's assume they don't happen - BU gets past Lowell and NU and PC stay TUC.
UML leaving TUC cost BU two pairs *below* them, Harvard and Quinnipiac, by slim TUC matches. With those two playing in the ECACHL QF, one of them will have to lose and BU will gain one pair.
If BU were to sweep UML, quick math (assuming slight drop to SOS/SOOS) says that would add over 50 points (.0050) to RPI. The loser of Wisco/SCSU and the MNs should similarly lose >50. Combined with the related TUC drop, that would give BU any of the UW, MN, MSM pairs.
By Sunday, BU could be 3rd tier or could be out.
Mar 13 2008, 2:48 pm by Todd
Todd, you are right about Harvard and QU below them - I only mentioned about catching teams above them. You are right, though, that flipping those comparisons will help BU move up. ... Regarding the teams above them ... you are right if what you say about RPI is true. It seems like a big RPI discrepancy. I'm not sure the needle move that much for all those teams mentioned. But maybe you're right.
Mar 13 2008, 3:02 pm by CHN Editor
Adam, here's how I made the estimation on RPI, which is usually pretty accurate. BU right now is 17-15-4, giving them a .5278 win %. If they sweep, they would become 19-15-4 or .5526 a gain of .0248 (actually .0249 w/o rounding). Anyway, since that accounts for 25% of the RPI, divide the 248-ish points by 4 (or * by .25) and it will add 62 points. Since UML's record and SOS are < BU's current Sched Str (the other 75%), there will be a slight drop on that end. The reason that I am assuming less of an impact from *playing* UML than by beating them is that 2 games out of BU's then-38 is more impactful than 2 (more) games with UML's record (36 games) relative to the other slates of the other opponents (O) (34-38 games ea) and the opponents' opponents (OO) (even more diffuse). In sum, I don't think their SOS will drop as much as 12 just from playing UML twice, but there are so many other factors to be added from other O & OO, that I'm just thumbnailing "a little <62" or "over 50".
Mar 13 2008, 7:52 pm by Todd
Similar estimations lead me to believe that if PC and NU match results this weekend, they will change places as BC is 200 RPI higher than UVM and PC's SOS will make up more than .0002 on NU. If either of them matches MTU's results then MTU should increase it's hold on 23 because NoDak is 500 higher than BC. However, if PC or NU should win one more game than MTU, that difference should let them pass Tech. Note that even though PC and NU are the lower seeds, PC was 2-0-1 vs BC and NU and UVM split (1-1-1), although UVM was 1-0-1 at home. Not unreasonable to think that either "underdog" might take a game, or even the series. Things are so knotted from 20-30 in RPI that there are also scenaria where UML could join PC and NU as TUCs - including a BU 2-1 series win which could add a 4-2 to BU's TUC record. BU's RPI wouldn't raise by as much as with a sweep, but it may not need to to catch Minn or even Wisco, should they lose out. Anyway... don't want to derail the thread over one team.
Mar 13 2008, 8:12 pm by Todd
Minnesota State played like hell against Minnesota in the 1st round of wcha playoffs. I hope they still get a spot with the 3 OT games.
Mar 19 2008, 10:31 pm by Dave, Mankato
