How to fix the seeding process

[billhoward]

Maybe the pendulum has swung too far. The NCAA committee seems caught up in applying a formula that's inflexible. No matter how many decimal places of precision in the math-based part of its calculation, there is a margin of real world error, meaning the team calculated as the No. 8 might be anywhere from say the sixth through eleventh best team. So why be so prissy about giving the calculated No. 8 the No. 8 seed?  Consider:

The real bands aren't 1-4, 5-8, 9-12, 13-16. The bands this year seem to be:
1-2 BC and Denver â€" get first round chumps (Mercyhurst, Bemidji State). If there were three weaklings, then the top band would be 1-2-3. (Whether BC really deserves the overall No. 1 is something else.)
3-4 CC and Minnesota â€" must play real hockey teams in first round but at least have top seeds
5-6 Cornell and Michigan â€" just missed top band seeds, legitimate chance to make Frozen Four
7-12 North Dakota, UNH, Harvard, (The) Ohio State, Wisconsin, BU â€" the great middle, no reason to complain about missing the top band of four, could argue they belong higher or lower in the middle and fans play what-if they'd been matched against X not Y in Worcester not Minneapolis.
13-14 Maine, Colgate â€" unlikely to advance but won’t embarrass themselves
15-16 Bemidji State, Mercyhurst â€" No need to book rooms for the whole weekend (KRACH rankings 31 and 51 of 58 D1 schools).
In other words, the seeding committee should really have worried this year about the 2-3 split (who does, doesn’t get a first round pushover) and the 4-5 split (who does, doesn’t get a top seed).

Allow modest reseeding based on the impact of sites and ice surfaces. If you play in your home region / state / town / rink and/or you normally play on an Olympic surface (and your opponent doesn’t), you’re stronger than your seeding indicates (or your opponent will play weaker than its seeding indicates). So while No. 4 (Minnesota this year) is matched up against No. 13 (Maine), the fact that Minnesota has every possible geographic/ice surface advantage this year means it could be paired up against say a No. 12, 11, or 10 seed (eg Ohio State) if need be to avoid some unfairness elsewhere and it would still have the same relative advantage.

Teams that draw well should be placed closer to home. (“For the good of the sport.”) In other words, the NCAA should honor the likes of Cornell and Wisconsin, unless it thinks half-empty rinks help college hockey. This is more important than in college basketball where virtually every game is televised and where students would have a hard time getting seats even if they could get to the regional game site.

Reward recent performance. This would benefit young teams and teams whose star player missed the first half with an injury. Based on its January 1-on performance, Cornell is a top-four seed.

Allow some intra-conference matchups if it avoids tortured juggling elsewhere (especially geographic juggling). Suppose, say, Vermont or RPI or Dartmouth squeaked in to the tournament, would it be so bad if Cornell had RPI as a first round opponent if avoiding RPI forced Cornell from Amherst to Minneapolis?

Make No. 16 a wild card. Meaning if the committee uses a heavily formula-based criteria for who gets in, make it apply to 15 not 16 teams. Then let the committee in its wisdom pick No. 16, a team that belongs for no other reason than "just because." This year that might have been Dartmouth or Vermont (although that would have meant tossing Colgate and Colgate just beat Vermont and Vermont just beat Dartmouth).

Could you trust the NCAA committee to apply this fairly? If the committee was required to report its reasoning, then sit still for a half-hour televised press conference on seeding day and once again the day before the Frozen Four begins, then the fear of looking like dolts might keep them honest. If the committee had the freedom to change a seeding here and there, it would be a better tournament.

OTOH, it's just a game.

[jtwcornell91]

[Q]billhoward Wrote:

 Maybe the pendulum has swung too far. The NCAA committee seems caught up in applying a formula that's inflexible.
[/q]

I feel the opposite.  The CC-DU disgrace shows they're abandoning objective criteria.

[ugarte]

[Q]jtwcornell91 Wrote:

 [Q2]billhoward Wrote:

 Maybe the pendulum has swung too far. The NCAA committee seems caught up in applying a formula that's inflexible.
[/Q]
I feel the opposite.  The CC-DU disgrace shows they're abandoning objective criteria.[/q]They are somewhere in between. They are inflexible about 4-team bands but willing to play favorites within the bands.

It is as if they are just pissed about the codebreakers running bracketology analyses.

[Josh 03]

What about a play in game to make it a field of 17 instead of 16.  Maybe even 2 play in games.  That way, the top 2 seeds don't get an auto-win against the AH and CHA winner.  Instead, those teams have to take on the #15 or #16 team to get to the 1st round.  That could equalize the bands a bit more.

It's a step backwards for AH and CHA, but I'm just spewing ideas here.

[Stephen Turner]

Interesting idea, but as a percentage, a lot of teams make the tournament now, especially compared to basketball.  A few good teams need to be left out in order to have some validity.

As for subjectivity/objectivity there needs to be some subjectivity, otherwise why do you need a committee.  Wouldn't it be better for Cornell (and for college hockey in general) if we were in Amherst, even if we were 6 or 7 instead of 5?  There needs to be some flexibility, common sense always has a roll.

[billhoward]

[Q]ugarte Wrote: ...  They are inflexible about 4-team bands but willing to play favorites within the bands.  It is as if they are just pissed about the codebreakers running bracketology analyses.[/q]

Right: "NCAA selection process" and "transparency" are not words often heard together, at least not in the past.

(Flipping the two and placing Denver ahead of Colorado College can be argued both ways. Denver did beat CC for the WCHA title the night before and won their head to head series by 3 games to 2. CC has a beatable team it must play while Denver gets a pushover. Sort of like what happened to Cornell in 2003 except then we were the No. 1 seed.)

[billhoward]

Interesting concept. And the following two years the NCAA first round has sites in Albany and then in Rochester. That's when it would really be worth - assuming we were not a high enough #1 seed - offering to drop down to a the second band of seeds (4-6-7-8) if it allowed us to stay in the region, if such an option were to exist. That would be interesting, especially if the whole thing were televised, sort of like NFL draft day with a countdown timer: The #1 seed overall gets to pick its region if it doesn't like its placement (so long as it doesn't force a host school out), then the #2 overall gets to accept or reject its region, etcetera, and when you get to #4 overall, #4 has the option of dropping to a #5 seed and picking its region. And you only have five minutes to decide. But you don't know what's behind Door Number Three yet - you mind wind up with a less desirable opponent. This would create all sorts of havoc with intra-conference matchups. It would be more popular in the East / Northeast where you actually drive to the home region site; if you're CC and the regional is in Minneapolis, having home ice doesn't exactly let your fans follow en masse.

[KeithK]

[q] What about a play in game to make it a field of 17 instead of 16. Maybe even 2 play in games.[/q]Play-in games are horrible.  Part of the appeal of a 16 team tournament is that you don't have byes.  The point of the auto-bids is to give every team in D1 something to play for in terms of the national scene.  The road is already hard enough for the AH/CHA teams - why make it harder?  Just so you can get Dartmouth or Northern Michigan into the tournament as #15?  If you can't get into the top 14 you don't deserve to play for the national championship.

Besides, hockey is over the NCAA guideline of 25% anyway so there's no way they would increase the size of the field.  Well, unless they thought a CHA/AH matchup would bring in Superbowl ratings.

[Steve M]

see below

[Steve M]

The way to fix it, within the existing rules, is to put a higher priority on keeping teams close to home (especially amongst the upper seeds) and put a lot more flexibility in the competitive equity department.  I don't mind them drawing a solid line between the #1 and #2 seeds (not flipping Cornell with Minny) while swapping teams within the seed bands (as they did with DU and CC).  But if the committee is going to give themselves the flexibility to do such swaps, it would be much better for them to stop worrying about whether seed 1d plays seed 2a in the 2nd round and give even greater priority to attendance than they did.  

The committee could have put together the following bracket that would have allowed many more fans to see their teams play live, while staying well within the published NCAA selection rules:

West-
4 Minnesota
7 North Dakota
10 Ohio State
14 Colgate

Midwest-
2 Colorado College
6 Michigan
11 Wisconsin
15 Bemidji State

East-
3 Denver
5 Cornell
12 Boston U.
13 Maine

Northeast-
1 Boston College
8 New Hampshire
9 Harvard
16 Mercyhurst

I know Adam and Jayson wouldn't like it, because it would be tougher for USCHO to predict the brackets, but as long as they're allowing swaps such as 1b with 1c, they won't be able to predict the brackets perfectly anymore anyway.

And for those who say the above brackets would be unfair from a competitive equity standpoint, my response is that PWR is not a good enough seeding method to produce accurate forced rankings of teams such that a few swaps within bands (or ignoring 1-8, 2-7, 3-6, 4-5 matchup criteria completely) is truly going to upset the competitive balance of the tourney.  College Hockey is a great sport, and the opportunity should be given to as many fans as possible to see their teams play live in the tourney IMHO.

[CrazyLarry]

I know we'd all love to be in an east regional, but the system is really pretty good.  As everyone says, no one predictor (RPI, TUC record, H2H, COP, or even KRACH) has enough precision to define seeding.  PWR takes a number of criteria (not the best criteria, no, but not worthless ones, either, at least they are reasonably well motivated) and uses them all together.  I think that has some value.

The #1 seeds get chosen and ranked, and everything follows from there, and that is as it should be.  The best #1 seeds get the easiest draw, and the lowest #1 seed gets the toughest draw.  And the best #2 seed gets the lowest #1 seed to contend with.  I can't see why you would do it any other way.

Would people really rather play CC or DU than Minnesota?  Having seen them all play on TV, not in a million years.

[Trotsky]

1. Publish a fully-deterministic procedure prior to the season which will be used to rank all teams at the end of the season.
2. Play the season.  No selection show is necessary -- selections are completely transparent.
3. Call team #1.  They have a complete choice of all regionals and all seeds.  There are *no* restrictions at all -- there are no bands, no host requirements, no intra-conference prohibitions.
4. Call the next team.  They have a choice of all regionals and all seeds not yet taken.
5-17. Repeat  Step 4.
18. The worst team gets whatever's left.

[Steve M]

I'd rather play BC in Worcester on NHL ice than Minnesota on their home Olympic sheet, but that is besides my point, as I'm not whining about the seeding.  My point is that PWR isn't accurate enough to perfectly seriatim rank the teams 1-16, so some tinkering within the seed bands can be done without destroying the competitive balance of the tournament.  Is everyone convinced beyond a reasonable doubt that BC is better than DU and CC, that Harvard and OSU are better than Wisconsin, and Maine is better than Colgate?  I'm not and I'm sure most people would agree with me.  Given this, why not do some swapping within seed bands, within reason using some judgement, to better boost attendance and give more people a chance to see their schools' teams?  The NCAA rules allow for this and I firmly believe the committee should use this flexibility to a greater extent.  I think they will in the future if they must to avoid attendance disasters.

[Steve M]

I like your idea providing the fully deterministic procedure involves a better ranking system than PWR.

[CrazyLarry]

Since when do teams get to pick their opponent, and why do you think Schafer will be any better at doing it than the Committee?  Also, do you really think Schafer will pick going to Worcester in BC's region rather than playing a struggling, injured Minnesota.  And why would he want the headache?  I'd rather he concentrate on winning two games, rather than scouting 16 teams.

Also, this year's tourney has lots of East-West intrigue.  Interesting Matchups.  I like it.  Of course, I'm in LA, the travel isn't making me upset.

Also, if you want to get rid of the hosting thing, which everybody hates - go to the Eastern regionals when your team isn't there.  Filling those places up is the only way to start convincing the NCAA to lose the host thing.