2012 NCAA Tournament: Division I Bracketology

[granddaddyscout]

Re: 2012 NCAA Tournament: Division I Bracketology

So at this point it appears that UW, Cornell, UM, SL, MH, and probably BC are in. Either BU or PC will take a spot, so a UMD win appears to knock out UND. Unless you think that the UMD win propels UND past BC. Thoughts?

 

[dave1381]

Re: 2012 NCAA Tournament: Division I Bracketology

I stand corrected -- it doesn't look like Minnesota passes Cornell with a win tonight.

Minnesota has an RPI of .6299, which the average of the RPI for 30 or so games. (probably the St. Cloud & Mankato & Union games get dropped I'm guessing) The RPI for beating UMD is about .7000. So I don't think that's enough to increase the season average to Cornell's RPI which is .6348.

 

[dave1381]

Re: 2012 NCAA Tournament: Division I Bracketology

To be honest, I don't really know how all the PWR seeding stuff works, haha. I just assumed cause NU was a TUC and PC wasn't But hey, in that case, I take back what I said about not playing NU.

That's right, It just happens to be true that one more TUC win doesn't help BU in this case.

 

[dave1381]

Re: 2012 NCAA Tournament: Division I Bracketology

Ok, let's talk about this Northeastern vs. North Dakota for the last at-large. Of course a UMD win over Minnesota means this last at-large no longer exists, but let's talk about it.

Some interesting things about this comparison are
(1) Northeastern wins the head-to-head comparison
(2) North Dakota appears to win comparisons against Mercyhurst & SLU & Harvard that Northeastern does not.
(3) You might think Northeastern wins a comparison against BU that North Dakota does not, but that's an artifact of the way common opponents is calculated in the USCHO PWR -- the NCAA men have moved to a different method which actually gives North Dakota the comparison, and I think the women's committee will follow suit. I assume the comparisons with BU are going to be a nonfactor.

Would the committee have authority to take Northeastern over North Dakota? I think so. The argument would be
(1) As I said before, Northeastern won the head-to-head comparison
(2) They could say that these comparison wins over Mercyhurst/SLU/Harvard that North Dakota are really just a consequence of tiny advantages in RPI and common opponents, while N. Dakota's 4-9-2 record vs. the RPI top 12 is a dismal 11th out of the 12 teams, so really the committee judges North Dakota as the loser of all the aforementioned comparisons.

Now I think the argument just described is horrible because record vs. RPI top 12 is a horrible criterion that's not adjusted for strength-of-schedule. I believe UND's RPI and common opponents are in fact evidence of UND's superior performance relative to Mercyhurst, St. Lawrence, and Harvard, and that the committee should recognize that.

But it's not crazy to think that the committee could really look down at UND because that record vs. the RPI Top 12 is really lousy. The precedent is that last year, when things got really complex for home ice for BC/Mercyhurst/Minnesota, the committee basically ignored other criteria that were close and gave home ice to BC because it had a much better record vs. the RPI top 12 than Mercyhurst and Minnesota.

 

[granddaddyscout]

Re: 2012 NCAA Tournament: Division I Bracketology

Here is the Pairwise with only tomorrow BU vs PC game. Based on Dave’s comments I assume that a BU win strengthens NU’s case and definitly knocks out UND, but they may be out already.


1 Wisconsin 11 31-4-2 .8649 2 .6431* 1 10-3-2 .7333
2 Cornell 10 29-4-0 .8788 1 .6349* 2 9-3-0 .7500
3 Minnesota 9 31-5-2 .8421 3 .6334* 3 11-4-1 .7188
4t Boston College 7 23-9-3 .7000 9 .5717 4 7-6-2 .5333
4t Mercyhurst 7 23-7-3 .7424 5 .5674* 7 4-3-0 .5714
6t North Dakota 5 22-11-3 .6528 12 .5709* 5 4-9-2 .3333
6t St. Lawrence 5 24-9-4 .7027 8 .5686* 6 6-7-1 .4643
8t Northeastern 4 22-7-4 .7273 6 .5669* 9 4-4-2 .5000
8t Boston University 4 22-13-1 .6250 14 .5502 11 7-9-1 .4412


Anyone care to speculate on matchup’s.

 

[Numbers]

Re: 2012 NCAA Tournament: Division I Bracketology

Here is the Pairwise with only tomorrow BU vs PC game. Based on Dave’s comments I assume that a BU win strengthens NU’s case and definitly knocks out UND, but they may be out already.


1 Wisconsin 11 31-4-2 .8649 2 .6431* 1 10-3-2 .7333
2 Cornell 10 29-4-0 .8788 1 .6349* 2 9-3-0 .7500
3 Minnesota 9 31-5-2 .8421 3 .6334* 3 11-4-1 .7188
4t Boston College 7 23-9-3 .7000 9 .5717 4 7-6-2 .5333
4t Mercyhurst 7 23-7-3 .7424 5 .5674* 7 4-3-0 .5714
6t North Dakota 5 22-11-3 .6528 12 .5709* 5 4-9-2 .3333
6t St. Lawrence 5 24-9-4 .7027 8 .5686* 6 6-7-1 .4643
8t Northeastern 4 22-7-4 .7273 6 .5669* 9 4-4-2 .5000
8t Boston University 4 22-13-1 .6250 14 .5502 11 7-9-1 .4412


Anyone care to speculate on matchup’s.

My guess would be:
First, the order:
1-UW; 2-Corn; 3-Minn; 4-BC; 5-Mercy; 6-UND or StL; 7-StL or NE; 8-BU or PC

Case A: UND is in. Start with the easy one: 5-Mercy @ 4-BC; Then, I see 2 western home teams, and 1 western road team, so someone will have to fly across country, but I assume only one. Therefore, I hope the committee does: UND @ UW (because UND and Minny just played); StL @ Minny; BU/PC @ Cornell.

Case B: NE is in. Again, start with the easy one: Mercy @ BC. Now, there are only 2 western teams, so 2 teams have to fly. I would hope they go: BU/PC @ Wisc and then NE @ Cornell and StL @ Minny. Reason for the last being, again, to avoid the all-HE matchup between St Lawrence and Cornell.

 

[pakidnyc]

My guess would be:
First, the order:
1-UW; 2-Corn; 3-Minn; 4-BC; 5-Mercy; 6-UND or StL; 7-StL or NE; 8-BU or PC

Case A: UND is in. Start with the easy one: 5-Mercy @ 4-BC; Then, I see 2 western home teams, and 1 western road team, so someone will have to fly across country, but I assume only one. Therefore, I hope the committee does: UND @ UW (because UND and Minny just played); StL @ Minny; BU/PC @ Cornell.

Case B: NE is in. Again, start with the easy one: Mercy @ BC. Now, there are only 2 western teams, so 2 teams have to fly. I would hope they go: BU/PC @ Wisc and then NE @ Cornell and StL @ Minny. Reason for the last being, again, to avoid the all-HE matchup between St Lawrence and Cornell.

I could see if being NE/BU/PC at UW; SLU at Cornell; UND at Minnesota and Mercyhurst at BC. Only one team has to fly again.

Edited to add: I actually don't see any scenario now that NU is in. Think the 8th seed will goto the winner of the HE.

 

[dave1381]

Re: 2012 NCAA Tournament: Division I Bracketology

According to the stuff I can find with them both having a PWR of 7 the RPI is the official tiebreaker. Can anyone confirm that?

This isn't such a simple question. Here are the facts.

-- In all the USCHO PWR, RPI is used as a tiebreaker when teams have the same # of comparisons won.

-- In the men's tournament since 2003, RPI has been the de facto tiebreaker, but not "the official tiebreaker"

-- In the men's tournament prior to 2003, the head-to-head comparisons were the more common tiebreaker

-- The women's tournament has been more like the men's tournament prior to 2003 than since.

-- The women's committee in the past has been more likely to use discretion in breaking ties. For example, last year when BC/Mercyhurst/Minnesota were in essentially a three-way tie for 4th, it was really BC's strong record vs. the RPI top 12 that ended up being the tiebreaker, while other differences between the teams were "smaller."

-- I sensed there was some pushback against the committee for straying too far from the conventional PWR table last year.

So what does this mean for this year and BC-Mercyhurst? As I said, I expect the women's committee is more likely to look at the individual comparison between BC-Mercyhurst as a tiebreak (like the pre-2003 men) rather than use RPI (like the post-2003 men). Mercyhurst wins the individual comparison.

That said, there was some pushback against the committee for using too much discretion. If that pushback was effective, then you could see BC get home ice.

 

[joegrav]

Re: 2012 NCAA Tournament: Division I Bracketology

This isn't such a simple question. Here are the facts.

-- In all the USCHO PWR, RPI is used as a tiebreaker when teams have the same # of comparisons won.

-- In the men's tournament since 2003, RPI has been the de facto tiebreaker, but not "the official tiebreaker"

-- In the men's tournament prior to 2003, the head-to-head comparisons were the more common tiebreaker

-- The women's tournament has been more like the men's tournament prior to 2003 than since.

-- The women's committee in the past has been more likely to use discretion in breaking ties. For example, last year when BC/Mercyhurst/Minnesota were in essentially a three-way tie for 4th, it was really BC's strong record vs. the RPI top 12 that ended up being the tiebreaker, while other differences between the teams were "smaller."

-- I sensed there was some pushback against the committee for straying too far from the conventional PWR table last year.

So what does this mean for this year and BC-Mercyhurst? As I said, I expect the women's committee is more likely to look at the individual comparison between BC-Mercyhurst as a tiebreak (like the pre-2003 men) rather than use RPI (like the post-2003 men). Mercyhurst wins the individual comparison.

That said, there was some pushback against the committee for using too much discretion. If that pushback was effective, then you could see BC get home ice.

BC and Mercyhurst split their season series 1-1. And if we're going into the discretion portion, iirc BC won one game by multiple goals, whereas Mercyhurst won its game in overtime against the run of play. BC has the higher RPI; I know that's not the "official" tiebreaker, and we don't really know if there is such a thing as an official tiebreaker. But based on what I know about the 'conventional' PWR, it would be PURELY discretion to put the Hurst ahead of BC.

Sticking to the PWR, this would also mean, to me, that NU is definitely out regardless. Which is too bad for them as they had an excellent season.

 

[dave1381]

Re: 2012 NCAA Tournament: Division I Bracketology

BC and Mercyhurst split their season series 1-1.

Oops, when I said "head-to-head comparison" I meant the "individual comparison" between BC & Mercyhurst. Mercyhurst won the individual comparison because it has a better record vs. TUC and common opponents, even though it had the worse RPI.

And if we're going into the discretion portion, iirc BC won one game by multiple goals, whereas Mercyhurst won its game in overtime against the run of play.

That's nonsense. The committee has discretion to do whatever is in the handbook, not discretion to do whatever it wants. It cannot consider margin of victory.

BC has the higher RPI; I know that's not the "official" tiebreaker, and we don't really know if there is such a thing as an official tiebreaker.

Sure we do. We know there is no official tiebreaker because there's a handbook and there's legislation, and there's no tiebreaker in the handbook. The committee can do whatever it wants within the bounds of that legislation. If you read the men's and women's handbook, you'll see the women's committee has a lot more room for discretion than the men's committee.

 

[whoop87]

Re: 2012 NCAA Tournament: Division I Bracketology

Dave1381 - thanks for all the fine work you do keeping everybody in the loop as to how things should pan out given the number of scenarios. Maybe I missed it earlier in the thread here...but this is what I am looking to find out if you have already crunched the numbers:

Case 1 - PC wins HE tournament and auto-bid...who are the Top8 in the Pairwise?
Case 2 - BU wins HE tournament and auto-bid...who are the Top8 in the Pairwise?

I'm most curious to know how these two scenarios impact NU and UND.

 

[dave1381]

Re: 2012 NCAA Tournament: Division I Bracketology

Dave1381 - thanks for all the fine work you do keeping everybody in the loop as to how things should pan out given the number of scenarios. Maybe I missed it earlier in the thread here...but this is what I am looking to find out if you have already crunched the numbers:
Case 1 - PC wins HE tournament and auto-bid...who are the Top8 in the Pairwise?
Case 2 - BU wins HE tournament and auto-bid...who are the Top8 in the Pairwise?
I'm most curious to know how these two scenarios impact NU and UND.

The PC-BU game tomorrow will have no impact on the PWR. What you see is what you'll get.

At this point, it's not so much about the numbers. It's about predicting where the committee might deviate from PWR, which is a matter of gauging how might deviate from the PWR. That involves following the past history of the committee and also understanding how the committee might alter it's practice based on the feedback it received last year.

I discussed the Northeastern-UND selection choice here. I feel the committee should pick UND and will pick UND, but I wouldn't rule out the possibility of Northeastern's selection.

 

[dave1381]

Re: 2012 NCAA Tournament: Division I Bracketology

I'll summarize my views here:

If I had to bet on one bracket, it's
Providence/BU at (1) Wisconsin
BC at (4) Mercyhurst
UND at (3) Minnesota
SLU at (2) Cornell.

This is the same as the USCHO PWR would predict, except individual comparisons are used to break ties rather than RPI.

As I said before, the committee was under some pressure to go more straight PWR like the men, so there's some chance BC could still host, but that's not what I think is the most likely outcome.

Is there some chance SLU could be sent to Wisconsin instead of Providence/BU? If Providence wins, absolutely no way -- MJ was furious about not getting Dartmouth last year with good reason. Now if BU wins, the committee could swap SLU & BU and avoid an intraconference matchup, though avoiding intraconference play has been very low priority starting with the 07-08 committee. The SLU-BU comparison is 2-2 with BU having the H2H and common opponents edge, and the committee could place the teams as it likes depending at priorities. My best guess though is that the committee faced more pressure to follow "straight pairwise" than necessarily avoid intraconference matchups, so I do think the bracket above is most likely.

I discussed earlier the possibility that Northeastern could be selected over UND. If Northeastern were selected, I'd expect a bracket of
Providence/BU at (1) Wisconsin
BC at (4) Mercyhurst
SLU at (3) Minnesota
Northeastern at (2) Cornell

 

[TonyTheTiger20]

Re: 2012 NCAA Tournament: Division I Bracketology

I have no idea whether or not BC will get home ice, but I'm so excited for 6pm...

 

[joegrav]

Re: 2012 NCAA Tournament: Division I Bracketology

Oops, when I said "head-to-head comparison" I meant the "individual comparison" between BC & Mercyhurst. Mercyhurst won the individual comparison because it has a better record vs. TUC and common opponents, even though it had the worse RPI.


That's nonsense. The committee has discretion to do whatever is in the handbook, not discretion to do whatever it wants. It cannot consider margin of victory.


Sure we do. We know there is no official tiebreaker because there's a handbook and there's legislation, and there's no tiebreaker in the handbook. The committee can do whatever it wants within the bounds of that legislation. If you read the men's and women's handbook, you'll see the women's committee has a lot more room for discretion than the men's committee.

I'm not trying to be sassy or anything. I'm just saying, it's either by the book, or it's the committee's discretion, no? If it's the committee's discretion, why can't they take into account how the two teams appeared to stack up in their head to head series? By the same token but in the other direction, BC has slumped down the stretch. How has Mercyhurst been? A strong finish to the season would be a reasonable discretionary justification to break the tie.

 

[dave1381]

Re: 2012 NCAA Tournament: Division I Bracketology

I'm not trying to be sassy or anything. I'm just saying, it's either by the book, or it's the committee's discretion, no?

No. The point is the men's committee book says almost EXACTLY what to do (the RPI tiebreaker is not legislated but it has been de facto policy for 9 years now), while the women's committee book is much more open for interpretation.

The women's committee book limits the criteria that can be considered, but it DOES NOT tell the women how to combine and compare those criteria, while the men's book does.

For example, the women's committee has discretion to say that one team's record vs. TUC advantage is much larger than the other team's advantage in RPI, and award the comparison to the team with the larger TUC advantage. The men's committee does not have that discretion and would have to award the comparison to team with the advantage in RPI.

Just because the committee has discretion to weight the various criteria as it likes, it does not mean the committee has the discretion to consider criteria like margin of victory and late-season performance that are simply not in the handbook.

And for what it's worth, if the committee had gone by the men's criteria last season, BC wouldn't have hosted. BC was precisely the team that benefited because the committee saw BC had a huge advantage over Mercyhurst and Minnesota in record vs. the RPI top 12, while the committee ignored smaller differences between the schools in RPI. You win some, you lose some.

 

[joegrav]

Re: 2012 NCAA Tournament: Division I Bracketology

No. The point is the men's committee book says almost EXACTLY what to do (the RPI tiebreaker is not legislated but it has been de facto policy for 9 0years now), while the women's committee book is much more open for interpretation.

The women's committee book limits the criteria that can be considered, but it DOES NOT tell the women how to combine and compare those criteria, while the men's book does.

For example, the women's committee has discretion to say that one team's record vs. TUC advantage is much larger than the other team's advantage in RPI, and award the comparison to the team with the larger TUC advantage. The men's committee does not have that discretion and would have to award the comparison to team with the advantage in RPI.

Just because the committee has discretion to weight the various criteria as it likes, it does not mean the committee has the discretion to consider criteria like margin of victory and late-season performance that are simply not in the handbook.

And for what it's worth, if the committee had gone by the men's criteria last season, BC wouldn't have hosted. BC was precisely the team that benefited because the committee saw BC had a huge advantage over Mercyhurst and Minnesota in record vs. the RPI top 12, while the committee ignored smaller differences between the schools in RPI. You win some, you lose some.

BC also benefited from the committee's odd choices in 2007 when we got @Dartmouth instead of @Wisconsin. I'm not complaining about it, I'm suggesting an argument that could be used in favor of one team or the other. I didn't realize that it was explicitly stated that they can weight the PWR criteria however they'd like.

I, for one, really like the men's committee's system wherein the only fudging they really do with the brackets is to avoid in-conference matchups. The math is pretty simple and easy for everyone to understand, well in advance, and it leaves a lot of this nonsense out.

 

[dave1381]

Re: 2012 NCAA Tournament: Division I Bracketology

BC also benefited from the committee's odd choices in 2007 when we got @Dartmouth instead of @Wisconsin.

Right, but that was more about avoiding intraconference matchups.

I'm not complaining about it, I'm suggesting an argument that could be used in favor of one team or the other. I didn't realize that it was explicitly stated that they can weight the PWR criteria however they'd like.

Just to be clear, it's not explicitly stated, it just says "these are the criteria the committee can use" and it's clear in practice the women's committee weight things as they like. Meanwhile, the men's handbook says, "these are the criteria, and each one is worth one point within the individual comparison, etc, etc"

I, for one, really like the men's committee's system wherein the only fudging they really do with the brackets is to avoid in-conference matchups. The math is pretty simple and easy for everyone to understand, well in advance, and it leaves a lot of this nonsense out.

My opinion is that I would ultimately like the committee to have less discretion as well, but I like that only if the selection criteria were improved.

For example, I would definitely want the common opponents to look like the new men's calculation, because it's clearly ludicrous the BU is winning that comparison over UND in the USCHO PWR now, even though UND has a better win pct. against each common opponent. My guess is the women's committee is already using the new men's committee common opponent calculation -- as ARM pointed out earlier this thread, Mercyhurst would've hosted last year if the women were still using the old common opponent calculation. The USCHO women's PWR should probably change that next year.

I'd also like criteria that could be aggregated in a clean way. For example, if we were to use KRACH as a criterion, instead of giving 1 point to the winner and 0 to the loser, you could give .75 to one team and .25 to the other if the KRACH expected win percentage were 75%. So every criterion is on some 0-to-1 scale, and you average those.

 

[dave1381]

Re: 2012 NCAA Tournament: Division I Bracketology

The final USCHO PWR is up. No real changes to speak off, as I said.

Not sure what anyone was looking for from this game in terms of RPI changes, but again, the case for Northeastern relies on whether the committee think Northeastern's advantage in record vs. RPI top 12 (4-4-2 vs. 4-9) and common opponents (3-1 vs. BU while UND was 1-1) outweigh North Dakota's superior RPI and advantage in common opponents vs. Mercyhurst/SLU that Northeastern doesn't have (thus giving N. Dakota more wins in total comparisons, which would be decisive on the men's side).

 

[Bumbles]

Re: 2012 NCAA Tournament: Division I Bracketology

The final USCHO PWR is up. No real changes to speak off, as I said.

Not sure what anyone was looking for from this game in terms of RPI changes, but again, the case for Northeastern relies on whether the committee think Northeastern's advantage in record vs. RPI top 12 (4-4-2 vs. 4-9) and common opponents (3-1 vs. BU while UND was 1-1) outweigh North Dakota's superior RPI and advantage in common opponents vs. Mercyhurst/SLU that Northeastern doesn't have (thus giving N. Dakota more wins in total comparisons, which would be decisive on the men's side).

I see them taking UND if not for the sole purpose of being able to reduce travel.

My predicted bracket:
BU @ Wisconsin
Mercyhurst @ BC
UND @ Minnesota
SLU @ Cornell

I'd like to see them switch BU and SLU so they can as much as possible avoid an intraconference first round but I think some ppl have already pointed out that the Committee really doesn't seem to care about that.