The 2014 Pairwise, Bracketology and History Thread`

[The Exiled One]

Re: The 2014 Pairwise, Bracketology and History Thread`

The St. Cloud Times was told by the SCSU "coaching staff" that they were told CHN is using the correct formula.

"My best friend’s sister’s boyfriend’s brother’s girlfriend heard from this guy who knows this kid who’s going with a girl who saw Ferris pass-out at 31 Flavors last night. I guess it’s pretty serious."

 

[UnionHockeyManiac]

Re: The 2014 Pairwise, Bracketology and History Thread`

The St. Cloud Times was told by the SCSU "coaching staff" that they were told CHN is using the correct formula.

"My best friend’s sister’s boyfriend’s brother’s girlfriend heard from this guy who knows this kid who’s going with a girl who saw Ferris pass-out at 31 Flavors last night. I guess it’s pretty serious."

Comment From Grant, Elk River
I brought this up last week, but once again uscho and college hockey news have different pairwise standings. Uscho has them at 8th and CHN has them at 6. If the pairwise rankings are based upon raw data only, then why is there a discrepancy?

Mick Hatten:
Grant, I asked about this and the SCSU coaching staff has told me that they have been told the College Hockey News PairWise is the correct one. I think USCHO is using last year's formula.

 

[UnionHockeyManiac]

Re: The 2014 Pairwise, Bracketology and History Thread`

College Hockey News:

1 Boston College 58 .6041 1 .5984 .0057 22-4-3 .8103 .8239
2 Minnesota 57 .5949 2 .5880 .0069 19-4-5 .7679 .7614
3 Union 55 .5713 3 .5663 .0051 19-6-3 .7321 .7324
4 Ferris State 55 .5713 4 .5706 .0007 20-6-3 .7414 .7589
5 Quinnipiac 54 .5656 5 .5596 .0060 20-5-5 .7500 .7448
6 St. Cloud State 53 .5594 6 .5553 .0041 15-6-5 .6731 .6642
7 Wisconsin 52 .5582 7 .5521 .0061 16-8-2 .6538 .6339
8 Mass.-Lowell 51 .5579 8 .5553 .0026 19-7-3 .7069 .7014
9 Cornell 49 .5570 9 .5553 .0017 12-5-5 .6591 .6743
10 Michigan 48 .5544 10 .5504 .0041 14-7-3 .6458 .6552
11 Vermont 48 .5454 11 .5426 .0028 15-9-3 .6111 .6119
12 Northeastern 47 .5454 12 .5397 .0057 16-10-3 .6034 .6088
13 Minnesota-Duluth 46 .5437 13 .5398 .0038 13-9-4 .5769 .5741
14 Providence 44 .5411 14 .5366 .0045 15-8-5 .6250 .6196
15 Colgate 44 .5352 16 .5268 .0085 14-11-3 .5536 .5493
16 Maine 43 .5355 15 .5297 .0058 13-10-3 .5577 .5708
17 North Dakota 42 .5342 17 .5323 .0019 14-9-3 .5962 .5827
18 Notre Dame 40 .5326 18 .5281 .0044 16-12-1 .5690 .5373
19 Yale 40 .5321 19 .5299 .0023 12-7-4 .6087 .6126
20 Clarkson 39 .5303 20 .5283 .0020 17-11-2 .6000 .6069
21 Western Michigan 39 .5295 21 .5249 .0046 14-10-4 .5714 .5669
22 Minnesota State 37 .5260 22 .5233 .0026 17-13-0 .5667 .5882
23 Denver 35 .5255 23 .5217 .0038 13-9-6 .5714 .5620
24 New Hampshire 35 .5222 25 .5192 .0030 16-15-1 .5156 .4969
25 Ohio State 34 .5249 24 .5226 .0023 14-9-3 .5962 .5812
26 Alaska-Anchorage 33 .5108 26 .5081 .0026 14-11-3 .5536 .5504
27 Rensselaer 32 .5073 27 .5045 .0027 12-12-4 .5000 .5000
28 St. Lawrence 32 .5072 28 .4999 .0073 11-14-3 .4464 .4621
29 Nebraska-Omaha 30 .5072 29 .5041 .0031 11-13-2 .4615 .4545
30 Brown 29 .5070 30 .5050 .0021 9-11-3 .4565 .4766
31 Bowling Green 28 .5040 31 .5021 .0019 13-12-5 .5167 .5137
32 Lake Superior 27 .5021 32 .4978 .0043 13-14-1 .4821 .4820
33 Air Force 26 .5011 33 .4998 .0013 17-9-4 .6333 .6383
34 Michigan Tech 25 .5011 34 .4994 .0016 12-14-6 .4688 .4801
35 Connecticut 23 .4961 36 .4946 .0015 15-9-4 .6071 .6077
36 Miami 23 .4952 37 .4916 .0036 10-13-3 .4423 .4409
37 Alaska-Fairbanks 23 .4945 38 .4945 .0000 12-12-4 .5000 .4926
38 Bentley 21 .4977 35 .4977 .0000 14-10-4 .5714 .6074
39 Northern Michigan 20 .4911 39 .4904 .0007 11-15-2 .4286 .4470
40 Michigan State 19 .4883 40 .4853 .0030 8-13-6 .4074 .3770
41 Massachusetts 18 .4869 41 .4834 .0035 8-17-4 .3448 .3536
42 Mercyhurst 18 .4859 42 .4859 .0000 15-12-5 .5469 .5577
43 Harvard 17 .4849 43 .4838 .0010 8-12-3 .4130 .3947
44 Bemidji State 16 .4782 44 .4773 .0009 8-15-7 .3833 .3856
45 Merrimack 15 .4716 46 .4694 .0022 7-16-3 .3269 .3235
46 Boston University 13 .4736 45 .4689 .0047 8-16-4 .3571 .3231
47 Robert Morris 12 .4522 47 .4522 .0000 11-13-4 .4643 .4593
48 Canisius 11 .4491 48 .4491 .0000 10-15-3 .4107 .4338
49 Colorado College 10 .4446 49 .4421 .0025 3-18-5 .2115 .1951
50 Dartmouth 9 .4431 50 .4409 .0022 4-16-3 .2391 .2328
51 Penn State 8 .4430 51 .4421 .0009 5-18-1 .2292 .2119
52 Niagara 7 .4424 52 .4424 .0000 9-15-4 .3929 .4104
53 Holy Cross 6 .4357 53 .4336 .0021 8-17-3 .3393 .3542
54 RIT 5 .4321 54 .4319 .0002 8-15-5 .3750 .3650
55 American Int'l 4 .4264 55 .4264 .0000 9-18-1 .3393 .3577
56 Princeton 3 .4256 56 .4235 .0021 4-19-0 .1739 .1667
57 Sacred Heart 2 .4093 57 .4077 .0016 8-20-0 .2857 .2963
58 Alabama-Huntsville 1 .3912 58 .3912 .0000 1-28-1 .0500 .0548
59 Army 0 .3640 59 .3640 .0000 3-22-0 .1200 .1290

USCHO:

1 Boston College 58 22-4-3 .8239 1 .6061* 1
2 Minnesota 57 19-4-5 .7614 2 .5966 2
3t Union 55 19-6-3 .7324 5 .5721* 3
3t Ferris State 55 20-6-3 .7589 3 .5709* 4
5 Quinnipiac 54 20-5-5 .7448 4 .5669* 5
6 Massachusetts-Lowell 53 19-7-3 .7014 6 .5599* 6
7 Wisconsin 52 16-8-2 .6339 11 .5595* 7
8 St. Cloud State 51 15-6-5 .6642 8 .5576* 8
9 Cornell 49 12-5-5 .6743 7 .5567* 9
10t Michigan 48 14-7-3 .6552 9 .5553 10
10t Northeastern 48 16-10-3 .6088 15 .5448* 11
12 Vermont 47 15-9-3 .6119 14 .5447 12
13 Minnesota-Duluth 46 13-9-4 .5741 22 .5427 13
14 Colgate 45 14-11-3 .5493 28 .5360 15
15 Providence 44 15-8-5 .6196 12 .5414* 14
16t Maine 42 13-10-3 .5708 23 .5356 16
16t Yale 42 12-7-4 .6126 13 .5331* 17
18 North Dakota 41 14-9-3 .5827 20 .5326 18
19 Clarkson 40 17-11-2 .6069 18 .5322 19
20t Notre Dame 38 16-12-1 .5373 29 .5306* 20
20t Minnesota State 38 17-13-0 .5882 19 .5287 21
20t Western Michigan 38 14-10-4 .5669 24 .5283 22
23t Denver 35 13-9-6 .5620 25 .5248 23
23t New Hampshire 35 16-15-1 .4969 32 .5231 25
25 Ohio State 34 14-9-3 .5812 21 .5243 24
26 Alaska-Anchorage 33 14-11-3 .5504 27 .5113 26
27 St. Lawrence 32 11-14-3 .4621 37 .5086 28
28t Brown 31 9-11-3 .4766 36 .5088 27
28t Rensselaer 31 12-12-4 .5000 31 .5079 29
30 Bowling Green 29 13-12-5 .5137 30 .5071 30
31 Nebraska-Omaha 28 11-13-2 .4545 39 .5060 31
32 Air Force 27 17-9-4 .6383 10 .5025 32
33 Lake Superior 26 13-14-1 .4820 34 .5012 33
34 Michigan Tech 25 12-14-6 .4801 35 .5007 34
35t Connecticut 23 15-9-4 .6077 16 .4960* 36
35t Miami 23 10-13-3 .4409 41 .4944 37
35t Alaska 23 12-12-4 .4926 33 .4942 38
38 Bentley 21 14-10-4 .6074 17 .4990 35
39t Northern Michigan 20 11-15-2 .4470 40 .4907 39
39t Mercyhurst 20 15-12-5 .5577 26 .4884 40
41 Massachusetts 18 8-17-4 .3536 50 .4860 41
42t Michigan State 17 8-13-6 .3770 46 .4841 42
42t Harvard 17 8-12-3 .3947 44 .4824 43
44 Bemidji State 16 8-15-7 .3856 45 .4792 44
45 Merrimack 15 7-16-3 .3235 51 .4704 46
46 Boston University 13 8-16-4 .3231 52 .4742 45
47 Robert Morris 12 11-13-4 .4593 38 .4531 47
48 Canisius 11 10-15-3 .4338 42 .4498 48
49 Colorado College 10 3-18-5 .1951 56 .4446 49
50 Niagara 9 9-15-4 .4104 43 .4433 50
51 Dartmouth 8 4-16-3 .2328 54 .4408 51
52 Penn State 7 5-18-1 .2119 55 .4383 52
53 Holy Cross 6 8-17-3 .3542 49 .4359 53
54 RIT 5 8-15-5 .3650 47 .4313 54
55 American International 4 9-18-1 .3577 48 .4270 55
56 Princeton 3 4-19-0 .1667 57 .4235 56
57 Sacred Heart 2 8-20-0 .2963 53 .4112 57
58 Alabama-Huntsville 1 1-28-1 .0548 59 .3907 58
59 Army 0 3-22-0 .1290 58 .3636 59

 

[Priceless]

Re: The 2014 Pairwise, Bracketology and History Thread`

Comment From Grant, Elk River
I brought this up last week, but once again uscho and college hockey news have different pairwise standings. Uscho has them at 8th and CHN has them at 6. If the pairwise rankings are based upon raw data only, then why is there a discrepancy?

Mick Hatten:
Grant, I asked about this and the SCSU coaching staff has told me that they have been told the College Hockey News PairWise is the correct one. I think USCHO is using last year's formula.

I'm sure Todd or Ed can clarify, but I am 99.9999% sure USCHO isn't using the formula from last year.

 

[The Exiled One]

Re: The 2014 Pairwise, Bracketology and History Thread`

Comment From Grant, Elk River
I brought this up last week, but once again uscho and college hockey news have different pairwise standings. Uscho has them at 8th and CHN has them at 6. If the pairwise rankings are based upon raw data only, then why is there a discrepancy?

Mick Hatten:
Grant, I asked about this and the SCSU coaching staff has told me that they have been told the College Hockey News PairWise is the correct one. I think USCHO is using last year's formula.

This is obviously a very thin reason to believe that CHN has it correct, hence the reason I mocked it. The answer loses even more credibility with the last sentence. USCHO may not be using the right formula, but they sure as hell aren't using last year's formula.

 

[LynahFan]

Re: The 2014 Pairwise, Bracketology and History Thread`

I would argue that Ohio State, if they win their final 8 regular season games, would probably be a #2 seed and would not fall out as an At Large. A few of the teams directly below them, should they run the table and go deep into their tournament, could also qualify as an At Large. Just because it may not have happened in the past 11 years doesn't mean it can't. Remember that PWR metrics have changed (no TUC and addition of bonus) so you can't necessarily compare this season's volatility to what was experienced in the past. So your statement that teams below #25 need to autobid, while probabilistically true, is not 100% accurate.

I thought the consensus was that there should be LESS volatility than in the past, with the elimination of the TUC cliff?

 

[The Exiled One]

Re: The 2014 Pairwise, Bracketology and History Thread`

I thought the consensus was that there should be LESS volatility than in the past, with the elimination of the TUC cliff?

The volatility over time is the same, but if you charted out the PWR volatility of any given team, it now looks more like a slope and less like a set of stairs.

 

[Priceless]

Re: The 2014 Pairwise, Bracketology and History Thread`

I would argue that Ohio State, if they win their final 8 regular season games, would probably be a #2 seed and would not fall out as an At Large. A few of the teams directly below them, should they run the table and go deep into their tournament, could also qualify as an At Large. Just because it may not have happened in the past 11 years doesn't mean it can't. Remember that PWR metrics have changed (no TUC and addition of bonus) so you can't necessarily compare this season's volatility to what was experienced in the past. So your statement that teams below #25 need to autobid, while probabilistically true, is not 100% accurate.

I would point out that my chart is based on history, not math. No team has ever come from below 24th place to do what you suggest. That doesn't mean I'm saying it WON'T happen, just that it never has.

As for volatility, we really don't know what it will look like with the new formula. Also, the RPI is incredibly tight and can fluctuate heavily in a weekend. Is that a result of the new formula, or is it just close this season? If Whelan's site was still up I might be able to tell you.

 

[Patman]

Re: The 2014 Pairwise, Bracketology and History Thread`

This is obviously a very thin reason to believe that CHN has it correct, hence the reason I mocked it. The answer loses even more credibility with the last sentence. USCHO may not be using the right formula, but they sure as hell aren't using last year's formula.

and frankly, its not worth generating a CHN/USCHO ****ing contest. There's strong animosity between the groups.

edit: now, i haven't looked at that article... did they say who was telling them its correct? How does SCSU know? NCAA told them? hockey committee?

 

[JimDahl]

Re: The 2014 Pairwise, Bracketology and History Thread`

I would argue that Ohio State, if they win their final 8 regular season games, would probably be a #2 seed and would not fall out as an At Large. A few of the teams directly below them, should they run the table and go deep into their tournament, could also qualify as an At Large. Just because it may not have happened in the past 11 years doesn't mean it can't. Remember that PWR metrics have changed (no TUC and addition of bonus) so you can't necessarily compare this season's volatility to what was experienced in the past. So your statement that teams below #25 need to autobid, while probabilistically true, is not 100% accurate.

Based on simulations, I would have put the "need an auto bid" line more around #30. That's not to say the mathematical and historical approaches disagree -- it's not very likely for a team in the 25-30 range to go on an 7 or 8 game winning streak to end the season.

As far as Ohio State in particular, they actually seem to have unusual upside. I've been seeing it a lot this season that teams in the 20-25 range face potential jumps of up to 10 spots with a sweep. I'm calculating Ohio State most likely landing between 14-17 if they pull it off this weekend. I'm calculating them needing to win 6 of the final 8 to most likely land at #14 or above.

As to the PWR debate -- I favor the CHN implementation so am using it, but am doing comparisons (including full simulations) to USCHO's. The differences are minor (1 spot here and there), so don't matter much for forecasts, but likely will at the end of the season.

 

[Numbers]

Re: The 2014 Pairwise, Bracketology and History Thread`

Based on simulations, I would have put the "need an auto bid" line more around #30. That's not to say the mathematical and historical approaches disagree -- it's not very likely for a team in the 25-30 range to go on an 7 or 8 game winning streak to end the season.

As far as Ohio State in particular, they actually seem to have unusual upside. I've been seeing it a lot this season that teams in the 20-25 range face potential jumps of up to 10 spots with a sweep. I'm calculating Ohio State most likely landing between 14-17 if they pull it off this weekend. I'm calculating them needing to win 6 of the final 8 to most likely land at #14 or above.

As to the PWR debate -- I favor the CHN implementation so am using it, but am doing comparisons (including full simulations) to USCHO's. The differences are minor (1 spot here and there), so don't matter much for forecasts, but likely will at the end of the season.

Jim,
CHN implementation means: Apply the .8/1.2 to all three parts of the RPI for that game? So, each game has its own separate RPI, which you then average? Making it easy to see which games to drop?

 

[JimDahl]

Re: The 2014 Pairwise, Bracketology and History Thread`

Jim,
CHN implementation means: Apply the .8/1.2 to all three parts of the RPI for that game? So, each game has its own separate RPI, which you then average? Making it easy to see which games to drop?

Yep, which I think is more in line with what hockey has done in the past.

 

[Numbers]

Re: The 2014 Pairwise, Bracketology and History Thread`

So, not much changes tonight. However, by any measure teams 12 - 23 in the PWR are very close. With that in mind, I have a question:

What is the playoff procedure for HE this year? I ask because right now 6 of those teams would find themselves in the field. However, if it is as it has been - 8 teams only qualify, then 2 of those 6 have to lose their first round matchups, and that means it would be hard to hold position.

Does anyone know?

Thanks.

 

[Patronick]

Re: The 2014 Pairwise, Bracketology and History Thread`

So, not much changes tonight. However, by any measure teams 12 - 23 in the PWR are very close. With that in mind, I have a question:

What is the playoff procedure for HE this year? I ask because right now 6 of those teams would find themselves in the field. However, if it is as it has been - 8 teams only qualify, then 2 of those 6 have to lose their first round matchups, and that means it would be hard to hold position.

Does anyone know?

Thanks.

1-5 get a bye.


Seeds 6-8 host seeds 9-11 respectively in a one game playoff.

Then the regular best of three series with 1-4 getting home ice.

 

[Numbers]

Re: The 2014 Pairwise, Bracketology and History Thread`

Very strange thing happened tonight. Somewhere between last night and tonight (it is 9 PM CST as I write this), CHN and USCHO PWR rankings have coalesced. Unfortunately, I was not paying sufficient enough attention to notice whose changed. However, as I type, I also notice it is the identical list to JimDahl on SiouxSports.

Addition after all games posted on CHN except the Alaska game:

#1s - BC, Minny, Union, SCSU
#2s - Ferris, QU, Wisco, Lowell
#3s - N'eastern, Mich, Maine, NoDak
#4s - Cornell, NoDame, Vermont, AHA (UMD just off the bubble)

Worcester: BC v AHA, Lowell V NoDak
St Paul: Minny v Cornell, Wisco v No"Eastern
Bridgeport: Union v Vermont, QU v Maine
Cincinnati: SCSU v NoDame, Ferris v Michigan

Explanation:
For the #4 seeds, first put AHA v BC. Then, NoDame to Cincinnati. Then, Vermont has to play Union since Cornell can't.
The #2 seeds come out perfect for attendance as bracketed. For the #3s, first put Michigan in Cincinnati for attendance. Then, Lowell has to face NoDak (not Maine or Northeastern). Then, QU is higher seeded than Wisco, so they should play Maine. Northeastern to St Paul to face Wisconsin.

 

[JimDahl]

Re: The 2014 Pairwise, Bracketology and History Thread`

Very strange thing happened tonight. Somewhere between last night and tonight (it is 9 PM CST as I write this), CHN and USCHO PWR rankings have coalesced. Unfortunately, I was not paying sufficient enough attention to notice whose changed. However, as I type, I also notice it is the identical list to JimDahl on SiouxSports.

It appears that USCHO changed -- all 3 now seem to match the "CHN formula", weighting all three components of RPI (instead of just win%, as USCHO had been until tonight). Assuming I'm interpreting their table correctly, it would have been nice if USCHO had announced that they changed, but it wouldn't be surprising they did it silently.

Given that most people probably don't compare multiple sites or follow this stuff closely enough to realize there was any controversy, it's good for everyone that they converged.

 

[Dutchman]

Re: The 2014 Pairwise, Bracketology and History Thread`

It appears that USCHO changed -- all 3 now seem to match the "CHN formula", weighting all three components of RPI (instead of just win%, as USCHO had been until tonight). Assuming I'm interpreting their table correctly, it would have been nice if USCHO had announced that they changed, but it wouldn't be surprising they did it silently.

Given that most people probably don't compare multiple sites or follow this stuff closely enough to realize there was any controversy, it's good for everyone that they converged.

Thanks for this note. This is great news. Given how close things are this year, it will be great we are all reading from the same page.

 

[Priceless]

Re: The 2014 Pairwise, Bracketology and History Thread`

1-5 get a bye.


Seeds 6-8 host seeds 9-11 respectively in a one game playoff.

Then the regular best of three series with 1-4 getting home ice.

Are teams reseeded after that new first round?

Very strange thing happened tonight. Somewhere between last night and tonight (it is 9 PM CST as I write this), CHN and USCHO PWR rankings have coalesced. Unfortunately, I was not paying sufficient enough attention to notice whose changed. However, as I type, I also notice it is the identical list to JimDahl on SiouxSports.

Addition after all games posted on CHN except the Alaska game:

#1s - BC, Minny, Union, SCSU
#2s - Ferris, QU, Wisco, Lowell
#3s - N'eastern, Mich, Maine, NoDak
#4s - Cornell, NoDame, Vermont, AHA (UMD just off the bubble)

Worcester: BC v AHA, Lowell V NoDak
St Paul: Minny v Cornell, Wisco v No"Eastern
Bridgeport: Union v Vermont, QU v Maine
Cincinnati: SCSU v NoDame, Ferris v Michigan

Explanation:
For the #4 seeds, first put AHA v BC. Then, NoDame to Cincinnati. Then, Vermont has to play Union since Cornell can't.
The #2 seeds come out perfect for attendance as bracketed. For the #3s, first put Michigan in Cincinnati for attendance. Then, Lowell has to face NoDak (not Maine or Northeastern). Then, QU is higher seeded than Wisco, so they should play Maine. Northeastern to St Paul to face Wisconsin.

That looks like what I worked out last night. Attendance in Cincinnati is pretty much a lost cause, but that's about as good as it will get. Maybe if Ohio State moves into the field.

It appears that USCHO changed -- all 3 now seem to match the "CHN formula", weighting all three components of RPI (instead of just win%, as USCHO had been until tonight). Assuming I'm interpreting their table correctly, it would have been nice if USCHO had announced that they changed, but it wouldn't be surprising they did it silently.

Given that most people probably don't compare multiple sites or follow this stuff closely enough to realize there was any controversy, it's good for everyone that they converged.

People are paying attention. I've been fielding more questions than usual on twitter about the pairwise and the differences between sites. I just wish more TV media understood pairwise if they're going to try to talk about them on a broadcast (for starters, it's pairwise, not power ranking).

 

[DeepRed72]

Re: The 2014 Pairwise, Bracketology and History Thread`

It appears that USCHO changed -- all 3 now seem to match the "CHN formula", weighting all three components of RPI (instead of just win%, as USCHO had been until tonight). Assuming I'm interpreting their table correctly, it would have been nice if USCHO had announced that they changed, but it wouldn't be surprising they did it silently.

Given that most people probably don't compare multiple sites or follow this stuff closely enough to realize there was any controversy, it's good for everyone that they converged.

Hopefully it is the correct formula now that all are using it. Perhaps the change was made because someone from the committee, who chose to remain anonymous, told the USCHO that the "CHN formula" is what will be applied.

 

[Biddco]

Re: The 2014 Pairwise, Bracketology and History Thread`

For a moment last night UMD was 11th in the USCHO PWR and 16th in the CHN PWR. Sounds like the discrepency has been fixed.