Win out or get screwed for the Big Red

[KeithK]

The point is that comparing conference records against identical (or nearly so) schedules provides a better measure of relative strength.  It's clear cut, doesn't depend on subjective weighting functions like the 25-50-25 in RPI, and doesn't need any strength of schedule compensation because the SoS is the same.  While it's true that non-conference games are needed to compare teams from different conferences with disparate schedules, it's also true that the sample size is quite small. (Not that conference schedules are "large" in a statistical sense, but 22 is significantly better than 7 [Ivy] and 24 is better than 10 [HE, I think]).

Whatever.

[Adam]

But Keith, I think your argument presupposes balanced conference schedules, which not every conference has.  We're lucky to have it in the ECAC.

[jeh25]

Keith K '93 wrote:
 subjective weighting functions like the 25-50-25 in RPI

Arbitrary? Yes. Subjective? No.

Cribbing from my old laxpower.com post on a similar topic, subjective means "taking place within the mind and modified by individual bias" and I take bias to mean "a preference or an inclination, especially one that inhibits impartial judgment."  The RPI weighting may be screwy, but you don't get a different result if you or I calculate it. Thus, I dislike you use of the word subjective.

If you care, check out  http://forums.laxpower.com/read.php?f=1&i=26663&t=26443

Otherwise, you can buy me a beer in Albany.

WordUseMan exits stage left....

[CrazyLarry]

Good point, Keith.  HE does have a balanced schedule.  Of course, no one is arguing BU is a higher seeded team than UNH, it only comes out that way because of the hosting thing.  The real question is, should BU be a #1 over, say, Minnesota?  BU's 5th place finish is irrelevant to that question, except in the indirect (and rather vague) information it reveals about BU's wins and losses during the season.

(Yes, I have ignored the question of whether BU should be rated above BC or Maine, having finished behind them in HE.)  I'll leave that be for now.

[KeithK]

I'm only arguing that BU should lose the comparison with UNH based on conference results and probably Maine and BC too.  I fully agree that you need to use out of conference data, inaccurate as it may be, to judge between BU and Minny.

Strictly speaking Hockey East isn't truly balanced because of the home ice inequity arising from three games against each team.

[KeithK]

WordUseMan (a.k.a. Captain Pedantic, etc.):

Yes, I probably meant something more like arbitrary than subjective.  But since you mention it, I think I'll be argumentative.  In a sense the RPI breakdown is subjective.  It's based on a subjective decision on the importance of strength of schedule as opposed to winning percentage.  The breakdown was changed from 35-50-15 back to 25-50-25 because of a belief that this would result in a ranking that more accurately predicts the relative strengths of teams in the different conferences (in particular dropping the RPI of MAAC teams and raising WCHA and HE teams).  In the absense of data objectively verifying the correctness of this assumption, the decision must be considered at least somewhat subjective.  As far as I know the committee didn't cite any evidence to justify the change (even though something like KRACH might be used to do so).

RPI and PWR are objective methods, once their components and formulas are established.  Setting the qualifications is at least arbitrary and probably somewhat subjecive.

[DeltaOne81]

Wow, here's something from the USCHO boards I never would picked up on...

If QU or Mercy win the MAAC, then it makes them a TUC, giving Maine an extra win or two vs. TUCs (beat Qu once, Mercy twice). BU/Maine are incredibly close in that category, so it would swing the BU/Maine comparison to Maine.- tieing them overall, with the tiebreaker probably going to Maine since they win the head to head Pairwise comparison.

Knocking BU down to a #2 would mean the top 2 eastern schools stay east - us and UNH - instead of just the top 1. So we're safe unless we really crash and drop past Maine. And it'd be almost be 'fair' to send Maine west, with their crappy end of the season.

[Tom P. 98]

I would never have picked up on that either even after realizing that Wayne State is now a TUC.  With all else unchanged, if Mercy or QU wins the MAAC, BU and Maine would tie with 25 comparisons won and Maine would win the tiebreaker with BU.  BU still plays in Worcester as a 2 seed.  Maine gets a 1 seed but would go west for faltering down the stretch.  We get a 1 seed in Worcester and get a possible QF matchup with BU (Worcester is closer to Ithaca than Providence) and get to thank Mercyhurst or Quinnipiac for winning the MAAC by playing them in the first round.  Life would be good.  

I find it very ironic that BU, who played the toughest schedule, thus allowing them enough comparison wins to put them into position for a #1 seeding, might lose that seeding at the very end because Maine put four games against the weakest conference on their schedule.

[ugarte]

A butterfly flaps its wings . . .

[Al DeFlorio]

Tom P. 98 wrote:
I find it very ironic that BU, who played the toughest schedule, thus allowing them enough comparison wins to put them into position for a #1 seeding, might lose that seeding at the very end because Maine put four games against the weakest conference on their schedule.

Once again illustrating the silliness of TUC in particular and PWR in general.  That having been said, I sure as hell hope it keeps Cornell east.

[CrazyLarry]

There's another possibility.  If Minnesota beats CC, they'll flip their comparison with Maine based on COP.  If Maine also flips the comparison with BU via the Mercyhurst/QU route, then Minn,Maine,BU will all be tied for the #1 seed (Minnestoa's loss to Dartmouth should become irrelevant by the time we get to this point, so I'm returning to them that point in the PWR).  Minnesota will almost surely have the RPI lead among the 3 teams, and get the #1 seed.  We'd definitely stay east then with 2 western and 2 eastern #1 seeds.

(I saw INCH has Minnesota as a 1 seed in their bracketology, which they don't confine to ITSET.  I tried to figure out if it was even possible.  I think it is.  Somebody correct me if I'm wrong)

[KeithK]

[Q]Once again illustrating the silliness of TUC in particular and PWR in general. [/Q]

Al, you've made this many times and while I don't entirely disagree, I did want to comment.  In principle it makes complete sense to judge a team based on how well it plays against good competition.  "Good competition" is defined by drawing a line in the sand, RPI or Win% or whatever.  No matter where you draw the line, there will always be teams on the bubble and thus other teams whose PWR will benefit or hurt from small perturbations.  The thing is, the system is designed to make sense this coming Sunday, on selection day when all the games have been played.  At that point we evaluate who is a TUC and then judge teams accordingly.  The silliness arises when we try to rank teams before the end of the season and when we watch how PWR progresses through time.

I'm not saying that we shouldn't try to do predictions, etc. before all is said and done.  It's entertaining.  But we should realize that the ranking system isn't designed to be smooth over time - it's just designed to "work" (in some fashion) at the end of the season.

[Al DeFlorio]

Sorry, Keith, but I disagree.  

The fact that one weak team of necessity has to win the MAAC--and also the CHA--doesn't automatically make that team "good competition," to use your phrase from above.  If QU should beat Mercyhurst--or vice versa--in the MAAC championship game, it won't make either one a stronger team than they were before.  A win over QU--or Mercyhurst-- should not be equated to a win over a legitimate TUC--and yet it will in the NCAAs eyes come Sunday.

[Tom's last paragraph above states it very nicely, by the way.]

Maybe the phrase "teams-under-consideration" has become meaningless, and the criterion should be changed to read "record-against-teams-with-RPI-equal-to-or-greater-than-.500." (To which I'd like to see added "and-with-winning-percentage-greater-than-.500."  But that's topic for another discussion.)

Lastly, Keith, for the umpteenth time, I really do understand that TUC is intended to be applied only at the end of the season.  Really I do.  My problem is the very arbitrary cutoff between TUCness and non-TUCness, and the fact that a win (or loss) against a .4999999999 RPI team doesn't count for squat whereas a win (or loss) against a .5000000001 team counts fully.  And, I think it's real silliness when a win over a .4262 RPI team (Bentley, should it win the MAAC), would be counted, if what you're trying to measure is success against "good competition."

Capisce?;-)



Post Edited (03-19-03 14:15)

[KeithK]

Al, it seems in my brain deadness I've missed some of the point of your umpteem earlier posts :-).  I totally agree, calling Wayne State a TUC for these purposes is ridiculous.

[Al DeFlorio]

Tada!!

Another piece of silliness is the possibility that by beating a team--say in the semifinals of your tournament--and thereby knocking it out of TUCness--you could in fact knock yourself down in the rankings (and even perhaps out of the tournament).  There was speculation on a USCHO thread that Cornell might be facing that situation with Brown on Friday, but it turned out to be not the case.  IMHO, any ranking system where you can hurt your own standing by winning a game is simply wrongheaded.