2013-2014 Women's D-I PairWise Contentions and Affirmations

[dave1381]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

...and Robert Morris will continue to have the 'better' COP, even if Mercyhurst beats RIT today. Because the Mercyhurst win over RIT would take their record from 3-1 to 4-1, adding only 0.05 to their total.

In fact, Mercyhurst would have to beat RIT *NINE TIMES* in a row for the effect on their COP to overcome the single loss to Ohio State.

Seems 'wrong' to me...

What would be remarkably easy to do is to calculate a KRACH-rating for each common opponent comparisons. What you do is estimate the rating that, conditional on the overall KRACH rating of opponents, best fits the common opponent results.

With such a measure, going 2-0 vs. 1-0 against a common opponent is meaningless if it happens against an 0-32 team, but there's a big difference if it happens against a 32-3 team. And that's as it should be.

 

[Numbers]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

Remember that the committee reserves the right to weight the criteria differently. So they could agree with you and say that one OT loss to Ohio State should not cost the comparison to a Mercyhurst team that did better than RMU in the CHA season and tournament. Humans can do that, but a USCHO program that just goes by raw math is not capable of making those subjective decisions. That's why it says it "attempts to mimic" what the committee will do, not "here's what the committee will decide."

I agree entirely, and this is why in my mind, Mercyhurst is ahead of RMU when the committee goes to work tomorrow. To me, when they look at the details of the CommOpp between Mercy and RMU, they will call it a wash. Mercyhurst is far enough ahead in RPI to negate RMU's lead in the TUCs and CommOpps. The second question is, "What about Quinn v RMU?"

I need some help with this, but I think it shakes out like this:
RPI - Probably about .0075 in favor of Quinn
TUC - In favor of RMU, and by a good margin, more if Quinn loses to Clarkson, but as someone wrote above, this is skewed by which TUCs they played.
CommOpp - Just barely on the side of Quinn, although I don't know if Clarkson is a common opponent.

So, in my world, this comparison falls to Quinnipiac, even if they lose today.

Can someone help me learn more?

And, of course, it's moot if Quinn wins, or if UND wins the WCHA, or if there is upset in HEA.

 

[Eeyore]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

...and Robert Morris will continue to have the 'better' COP, even if Mercyhurst beats RIT today. Because the Mercyhurst win over RIT would take their record from 3-1 to 4-1, adding only 0.05 to their total.

In fact, Mercyhurst would have to beat RIT *NINE TIMES* in a row for the effect on their COP to overcome the single loss to Ohio State.

Seems 'wrong' to me...

Do you have some proposal for how to fix it? Because while that may be wrong, it isn't close to as egregiously wrong as the results you get when you don't normalize the number of games played against each opponent? And within the range of the number of times teams actually play each other, being a half game better against a team that each has played 4-5 times isn't very meaningful.

I do agree that putting that much weight on a single game against OSU is problematic but the problems arise from trying to use record vs. common opponents in this way at all rather than the specific formula used. If you go with a system in which all subjective decision making is dispensed with and the eight team field is selected solely by a mathematical formula (and as ARM has pointed out, this isn't actually true here) then you're going to get oddball cases. The best you can do is to try to come up with a formula that minimizes them. Absent doing something complicated, weighting record vs. common opponents as they are doing now is probably the best compromise available.

 

[robertearle]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

Do you have some proposal for how to fix it?

dave's KRACH suggestion is A-OK with me.

 

[Eeyore]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

dave's KRACH suggestion is A-OK with me.

Me, too. Unfortunately, that falls into the category of "too complicated for the NCAA to get its head around." I think the chances of them deciding to use KRACH for anything in my lifetime is really small.

 

[vicb]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

Quinnipiac loses 6-0 today. The NC$$ Playoff Structure is becoming clearer????

 

[Numbers]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

Quinnipiac loses 6-0 today. The NC$$ Playoff Structure is becoming clearer????

Not sure yet, vicb. It sounds as if BC is winning comfortably, so that really leaves 2 major things undecided - WCHA final and HEA final.

Thus, right now....
Assuming BC and Minny win, so no upsets, the major question mark is which 2 of Mercyhurst, Quinnipiac and RoMo make the field. It is my opinion that the criteria favor Mercyhurst first by a clean break, and then Quinnipiac 2nd. However, looking upthread you will see a lot of discussion about the Quinnpiac/RoMo thing.

 

[Eeyore]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

Not sure yet, vicb. It sounds as if BC is winning comfortably, so that really leaves 2 major things undecided - WCHA final and HEA final.

Thus, right now....
Assuming BC and Minny win, so no upsets, the major question mark is which 2 of Mercyhurst, Quinnipiac and RoMo make the field. It is my opinion that the criteria favor Mercyhurst first by a clean break, and then Quinnipiac 2nd. However, looking upthread you will see a lot of discussion about the Quinnpiac/RoMo thing.

Unless, of course, RIT wins the CHA title, in which case the battle for the 7th and 8th seeds gets even messier.

 

[Numbers]

Unless, of course, RIT wins the CHA title, in which case the battle for the 7th and 8th seeds gets even messier.

And... Some battle. 3 loses and a win when trying to audition for the committee...

 

[dave1381]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

I agree entirely, and this is why in my mind, Mercyhurst is ahead of RMU when the committee goes to work tomorrow. To me, when they look at the details of the CommOpp between Mercy and RMU, they will call it a wash. Mercyhurst is far enough ahead in RPI to negate RMU's lead in the TUCs and CommOpps. The second question is, "What about Quinn v RMU?"

I need some help with this, but I think it shakes out like this:
RPI - Probably about .0075 in favor of Quinn
TUC - In favor of RMU, and by a good margin, more if Quinn loses to Clarkson, but as someone wrote above, this is skewed by which TUCs they played.
CommOpp - Just barely on the side of Quinn, although I don't know if Clarkson is a common opponent.

So, in my world, this comparison falls to Quinnipiac, even if they lose today.

Can someone help me learn more?

And, of course, it's moot if Quinn wins, or if UND wins the WCHA, or if there is upset in HEA.

Not sure yet, vicb. It sounds as if BC is winning comfortably, so that really leaves 2 major things undecided - WCHA final and HEA final.

Thus, right now....
Assuming BC and Minny win, so no upsets, the major question mark is which 2 of Mercyhurst, Quinnipiac and RoMo make the field. It is my opinion that the criteria favor Mercyhurst first by a clean break, and then Quinnipiac 2nd. However, looking upthread you will see a lot of discussion about the Quinnpiac/RoMo thing.

Both you and previously Eeyore have been leaving out the head-to-head criterion, which is going to be THE difference in Robert Morris getting in over Quinnipiac for the last slot if Minnesota & BC win their autobids.

 

[Numbers]

Both you and previously Eeyore have been leaving out the head-to-head criterion, which is going to be THE difference in Robert Morris getting in over Quinnipiac for the last slot if Minnesota & BC win their autobids.

And if you are right,I have no complaint. To me, when I look at the criteria, it seems almost a toss up. I personally would choose Quinnipiac by a slight margin, but either way... Fine with me. Whoever is going to fly to Minneapolis anyway. Good luck there.

 

[FireKnight]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

Boy those 7-9 spots are really tight. USCHO now has M'Hurst at 7, QU at 8 and Bob Morris at 9.

 

[dave1381]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

The RPI has been updated. The Quinnipiac-RMU edge is .0050 (not .0075, if it were .0075 it would be much more up in the air like you say).

To put the RPI difference into perspective, if RMU had 1 more win & 1 less loss on its existing schedule, its RPI would be .0086 higher. If RMU had one more tie & 1 less loss, its RPI would be .0043 higher.

So that's why I say the 1-0-1 head-to-head for RMU beats the .0050 RPI difference, at least going by the committee's precedent in picking Dartmouth over Clarkson in 2008.

 

[dave1381]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

Boy those 7-9 spots are really tight. USCHO now has M'Hurst at 7, QU at 8 and Bob Morris at 9.

USCHO PWR uses RPI as the tiebreaker, so don't expect that to be of much predictive value when Quinnipiac and RMU are so close in three of the criteria and RMU has the head-to-head edge. In a very similar situation, the 2008 committee took Dartmouth (who had the 1-0-1 head-to-head edge) over Clarkson (who had Dartmouth beat by tiny margins in the other three criteria). Amazingly, Seeley was coach of that Clarkson team and this Quinnipiac team and I don't think he has ever made the NCAA tournament.

 

[robertearle]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

USCHO PWR uses RPI as the tiebreaker, so don't expect that to be of much predictive value when Quinnipiac and RMU are so close in three of the criteria and RMU has the head-to-head edge.

One more thing to factor in, if you haven't yet... an RIT win today would likely raise their RPI above .500, making them a 'team under consideration', and changing the TUC percents for anyone (Quinnipiac, RMU. Mercyhurst) that played them.

 

[FireKnight]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

USCHO PWR uses RPI as the tiebreaker, so don't expect that to be of much predictive value when Quinnipiac and RMU are so close in three of the criteria and RMU has the head-to-head edge. In a very similar situation, the 2008 committee took Dartmouth (who had the 1-0-1 head-to-head edge) over Clarkson (who had Dartmouth beat by tiny margins in the other three criteria). Amazingly, Seeley was coach of that Clarkson team and this Quinnipiac team and I don't think he has ever made the NCAA tournament.

Oh, I remember the choosing of Dartmouth quite well and the situation is strikingly similar. However, this is a new committee, so who knows what they will do. I would imagine that QU's RPI could change quite a bit too, depending on the outcome of the 2 yet decided ECAC games. If Cornell wins out, their 1-0-1 record v. the Big Red should help their RPI.

Also, if RIT manages to beat M'Hurst, you're going to add a 3-2-0 TUC to RMU and a 0-1-0 to QU.

 

[Numbers]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

With Clarkson and Cornell winning, it appears the top part of the bracket shakes out like this:
If BC wins and Cornell wins, 1-UM, 2-Corn, 3-Clark, 4-UW, 5-BC, 6-Harv
If BC wins and Clarkson wins, 1-UM, 2-Clark, 3-UW, 4- BC, 5-Corn, 6-Harv
If BC loses and Cornell wins, 1-UM, 2-Corn, 3-Clark, 4-UW, 5-Harv, 6-BC with Harvard only slightly ahead of the HU/BC compare
If BC loses and Clarkson wins, 1-UM, 2-Clark, 3-UW, 4-Cornell, 5-Harv, 6-BC again with HU only slightly winning the HU/BC compare.

 

[dave1381]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

Really, it's a joke that UND isn't solidly into this tournament. KRACH says they're on par with BC and Harvard, not slightly lagging these no-shows in Mercyhurst, RMU, and Quinnipiac.

UND actually has the common opponents edge against Mercyhurst and Quinnipiac, though not RMU, and the RPIs are all quite close. What dooms UND is the nonsense TUC criterion. But if I were on the committee, I'd try to find a way to interpret the mandate to get UND into this tournament. The common opponent measures are better than anything else they have available in comparing these teams.

 

[dave1381]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

I don't think it's crazy to go RMU #7 and UND #8 even with a WCHA final loss. You just have to recognize that the TUC criterion is worthless. You're saying the one team that's beaten the #1 team has a disadvantage against the teams under consideration. It's just all unintentional irony at this point.

 

[Numbers]

Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations

Really, it's a joke that UND isn't solidly into this tournament. KRACH says they're on par with BC and Harvard, not slightly lagging these no-shows in Mercyhurst, RMU, and Quinnipiac.

UND actually has the common opponents edge against Mercyhurst and Quinnipiac, though not RMU, and the RPIs are all quite close. What dooms UND is the nonsense TUC criterion. But if I were on the committee, I'd try to find a way to interpret the mandate to get UND into this tournament. The common opponent measures are better than anything else they have available in comparing these teams.

Yes, but remember that NoDak gave Minnesota its only loss, so if you take out Minnesota from the TUC records, you subtract 1-3 from North Dakota, and then it looks really close too.

I understand KRACH ranks UND high. I wonder if that might be because the 4 games against Minny end up with a high level of value because Minnesota has the 1 loss this year, so they are not OFF THE CHART like they were last year.

Maybe the real answer is: Mercyhurst, Quinn, RMU and UND are all too close to call.

As for "no-shows", where was UND when they played OSU and could have won a couple of games and we wouldn't be wondering about this today.