What's new
USCHO Fan Forum

This is a sample guest message. Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

  • The USCHO Fan Forum has migrated to a new plaform, xenForo. Most of the function of the forum should work in familiar ways. Please note that you can switch between light and dark modes by clicking on the gear icon in the upper right of the main menu bar. We are hoping that this new platform will prove to be faster and more reliable. Please feel free to explore its features.

Pairwise and Bracketology 2013 Edition

  • Thread starter Thread starter Priceless
  • Start date Start date
Status
Not open for further replies.
Re: Pairwise and Bracketology 2013 Edition

If memory serves, Jim Dahl puts every permutation possible into a computer and it spits out the results, but I think that is after the QF round as the number of permutations is still too great.

Not sure he even does that Priceless. He mostly uses Monte Carlos, because even after quarterfinal weekend, the number of permutations is: 2^17 * 3^2, which is obviously a very huge number, and for every one, the computer has to run the PWR calculation. This is about 2.5 million permutations. That is the number of seconds in 1 whole month. So, even if he could code that operation, if his computer could calculated and log 1 permutation every second, it would take a month to do the job, and that's too long.:):):):) :):):)
 
Re: Pairwise and Bracketology 2013 Edition

Not sure he even does that Priceless. He mostly uses Monte Carlos, because even after quarterfinal weekend, the number of permutations is: 2^17 * 3^2, which is obviously a very huge number, and for every one, the computer has to run the PWR calculation. This is about 2.5 million permutations. That is the number of seconds in 1 whole month. So, even if he could code that operation, if his computer could calculated and log 1 permutation every second, it would take a month to do the job, and that's too long.:):):):) :):):)


I wouldn't say a month... a few hours depending on how well he wrote the code. Less if he can multi-core the job.

edit: you know what, i don't really know, the last time I computed the pairwise it was sitting around four seconds per, I think... but I'm not the best programmer and I've learned some more tweaks since the last time I did it. I have a skeleton program together right now but no data to test and I haven't set up the RPI calculation which is a pain in the ***. The game removal bit is the hardest part... IIRC, there's also a bit of removal in the SOS w/ out whole game removal... SOS doesn't include the games you played vs. opponent... I think?

----

Monte Carlo is really the only way otherwise.... the rest is a matter of shifting the probabilities with respect to a model.
 
Last edited:
Re: Pairwise and Bracketology 2013 Edition

Priceless,

I took a quick gander. If BU beats MC, MC drops off the cliff no matter what. So, give BU a sweep. Then, make Union lose their 1/4 final series in 2 games. Denver wins today, so Wisco plays Omaha. Give Omaha the win. I think that will keep BU ahead of Union, and even give them a little leeway.

Edit: In that last scenario, Omaha has to sweep.
And, should UAF win tonight (a loss does BU no good), then it is impossible to flip the BU/UAF compare without also flipping the NoDame/BU compare to NoDame. So, the above, with BU re-winning the Union and Wisconsin compares, is the best case. Union is necessary, because it's a +1, -1 situation.

Wisco plays UMD, not Omaha
 
Re: Pairwise and Bracketology 2013 Edition

Wisco plays UMD, not Omaha

Thanks, that got straightened out in the other thread, too. I was mistaken about the tiebreakers. In the case of BU, the idea is that BU has to win the compare with Union - so BU and Dartmouth have to sweep. Then, BU can still win the Wisco compare without damaging the other compares, so have UMD sweep. It actually would work better, I think. In that case, BU wins the compare with Wisco, too.

And, the last paragraph still holds.
 
Re: Pairwise and Bracketology 2013 Edition

Pairwise with Quarterfinal Conference Tourney coming next:
1 – QU, 2 – Minny, 3 – Miami, 4 – Lowell
5 – NoDak, 6 – Denver, 7 – BC, 8 – UNH (All with equal PWR wins)
9 – Mankato, 10 – WMU, 11 – Yale, 12 – SCSU
13 – Niagara, 14 – RPI, 15 – NoDame, 16 – Union

So:
Providence: QU (1), BC (7), SCSU (12-lower of WCHA teams), NoDame (15)
GR: Minny (2), Denver (6), WMU (10), Union (16 – avoid ECAC matchup)
Toledo: Miami (3), NoDak (5-highest left), Yale (11), Niagara (13 – save 2 flights)
Manchester: Lowell (4), UNH (8-host), Mankato (9-higher of WCHA teams), RPI (14)

Really only 2 possible quibbles with this bracket: Mankato and SCSU are interchangeable, because the entire 2-band is really tied, and Nia/RPI switch may be debateable.

I had not noticed earlier the possibility of Union being 'in' at this point.

Edit: Difference between CHN, USCHO and Whelan's DIY is the method of breaking ties. For anyone using that site at elynah, remember that the code there is written to break ties among teams with the same number of comparison wins by looking at the compares between tied teams. CHN and USCHO, as well as the committee (apparently??) use RPI as a tiebreaker in that situation. Hence, Union being #16 tonight. They win the tie with BU because of a higher RPI.
 
Last edited:
Re: Pairwise and Bracketology 2013 Edition

Interesting. ECAC partisan that I am, it is still not possible for 4 ECAC teams to make the tournament. Even if Yale, RPI and Union all survive their QF rounds, the fact that those three teams will have two games against each other in AC has to eliminate somebody, right?

Actually, the best case is: all get to AC along with QPac. Union wins twice, then the loser of RPI-Yale beats QPac in the consolation game, coming out of the weekend net up in RPI. Though I'm still not sure whichever of Yale-RPI wins the first game and loses to Union can survive.
 
Last edited:
Re: Pairwise and Bracketology 2013 Edition

Interesting. ECAC partisan that I am, it is still not possible for 4 ECAC teams to make the tournament. Even if Yale, RPI and Union all survive their QF rounds, the fact that those three teams will have two games against each other in AC has to eliminate somebody, right?

Actually, the best case is: all get to AC along with QPac. Union wins twice, then the loser of RPI-Yale beats QPac in the consolation game, coming out of the weekend net up in RPI. Though I'm still not sure whichever of Yale-RPI wins the first game and loses to Union can survive.

Are you using the DIY sim on elynah? You could always try to put in results from other leagues, too. Like, what if NoDame loses this weekend? That helps your 4th ECAC team alot. I am sure there are ways for it to happen, although it might be tough, in reality, since QU is not likely to lose twice at AC.
 
Re: Pairwise and Bracketology 2013 Edition

Is it possible the CCHA is a one bid league this year?

I don't foresee WMU AND ND losing this weekend, but it wouldn't shock me if they did.

If WMU and ND lose this weekend, and Miami ends up winning the tournament, is it possible the CCHA is a one bid leauge?
 
Re: Pairwise and Bracketology 2013 Edition

Is it possible the CCHA is a one bid league this year?

I don't foresee WMU AND ND losing this weekend, but it wouldn't shock me if they did.

If WMU and ND lose this weekend, and Miami ends up winning the tournament, is it possible the CCHA is a one bid leauge?

I will play with this a little. My first answer is that I doubt Western falls out. But, we shall see...

OK. Here's what I found with just a little playing with CCHA results only. Notre Dame apparently did well against Michigan this year. You notice NoDame has a high rpi compared to their PWR rank. PWR is punishing them because of TUCs. So, if Michigan becomes a TUC, that helps NoDame ALOT.
So, to get NoDame out of the field, and try to get WMU out as well, Michigan's results are imprtant. If Michigan sweeps WMU, for NoDame to fall out of the field, Michigan needs to lose/tie at the Joe, or lose/lose. If Michigan beats WMU in 3 games, then Michigan needs to lose/lose. I am assuming there is a 3rd place game.
As for WMU, taking at least a game from Michigan is strong insulation. If Michigan sweeps, then, with no other results accounted for, WMU comes out 13th. Then, you would need RPI and BU to have good tourneys. Specifically, if RPI has a good tourney, but Union comes out standing tall in Atlantic City, then WMU could fall out.

So, to answer: It's possible. Most things are still possible. But, it's highly unlikely.

Which, as a Bronco fan, should make you happy.:):) Here at Pairwise and Bracketology, we try to please.:):)
 
Last edited:
Re: Pairwise and Bracketology 2013 Edition

The script for all remaining games:

20130314 Me 0 ML 1 nc
20130315 Me 0 ML 1 nc
20130315 NH 0 Pv 1 nc
20130315 Vt 0 BC 1 nc
20130315 Mr 0 BU 1 nc
20130316 NH 0 Pv 1 nc
20130316 Vt 0 BC 1 nc
20130316 Mr 0 BU 1 nc
20130322 BU 0 BC 1 nc
20130322 Pv 0 ML 1 nc
20130323 BC 0 ML 1 nc

20130315 CC 0 DU 1 nc
20130315 NO 0 Mk 1 nc
20130315 AA 0 SC 1 nc
20130315 BS 0 Mn 1 nc
20130315 MT 0 ND 1 nc
20130315 MD 0 Wi 1 nc
20130316 CC 0 DU 1 nc
20130316 NO 0 Mk 1 nc
20130316 AA 0 SC 1 nc
20130316 BS 0 Mn 1 nc
20130316 MT 0 ND 1 nc
20130316 MD 0 Wi 1 nc
20130321 Mk 0 ND 1 nc
20130321 DU 0 Wi 1 nc
20130322 Wi 0 SC 1 nc
20130322 ND 0 Mn 1 nc
20130323 Mn 0 SC 1 nc

20130315 Cr 0 Qn 1 nc
20130315 Bn 0 RP 1 nc
20130315 SL 0 Ya 1 nc
20130315 Da 0 Un 1 nc
20130316 Cr 0 Qn 1 nc
20130316 Bn 0 RP 1 nc
20130316 SL 0 Ya 1 nc
20130316 Da 0 Un 1 nc
20130322 Un 0 Qn 1 nc
20130322 Ya 0 RP 1 nc
20130323 Un 0 Ya 1 nc
20130323 RP 0 Qn 1 nc

20130315 FS 0 OS 1 nc
20130315 MS 0 Mm 1 nc
20130315 BG 0 Nt 1 nc
20130315 Mi 0 WM 1 nc
20130316 FS 0 OS 1 nc
20130316 MS 0 Mm 1 nc
20130316 BG 0 Nt 1 nc
20130316 Mi 0 WM 1 nc
20130322 OS 0 Mm 1 nc
20130322 WM 0 Nt 1 nc
20130323 Nt 0 Mm 1 nc

20130315 Mh 0 HC 1 nc
20130315 Ca 0 AF 1 nc
20130315 RM 0 Ct 1 nc
20130315 RT 0 Ni 1 nc
20130316 Mh 0 HC 1 nc
20130316 Ca 0 AF 1 nc
20130316 RM 0 Ct 1 nc
20130316 RT 0 Ni 1 nc
20130322 HC 0 AF 1 nc
20130322 Ct 0 Ni 1 nc
20130323 AF 0 Ni 1 nc

Again, the higher seed wins every game. To have a lower seed win, change the 0 to a 2. Remember to add a 3rd game if necessary and/or to change future rounds accordingly.
If these results hold (which of course they won't) the final pairwise would be

1 Quinnipiac (EC) 29
2 Minnesota (WC) 28
3 Miami (CC) 27
4 Mass-Lowell (HE) 25
5 St Cloud (WC) 23
6 Boston Coll (HE) 23
7 North Dakota (WC) 22
8 MSU-Mankato (WC) 21
9 Niagara (AH) 21
10 Yale (EC) 20
11 New Hampshire (HE) 20
12 Denver U (WC) 19
13 Western Mich (CC) 16
14 RPI (EC) 16
15 Notre Dame (CC) 14
16 Providence (HE) 13
---
17 Union (EC) 13
18 Wisconsin (WC) 12
19 Boston Univ (HE) 11
20 Air Force (AH) 11

which gives us a nice bracket - if this were next year, when Miami and Notre Dame weren't in the same conference.

Code:
[B]Manchester (UNH)	Providence (Brown)	Toledo (BGSU)		Grand Rapids (Michigan)[/B]
Lowell			Quinnipiac		Miami			Minnesota
St Cloud		Mankato			Boston C		N Dakota
New Hamp		Niagara			Denver			Yale
RPI			Providence		Notre Dame		W Michigan

Too bad it can't happen. Notre Dame goes to Grand Rapids, W Michigan goes to Manchester and RPI goes to Toledo.
 
Re: Pairwise and Bracketology 2013 Edition

The script for all remaining games:

20130314 Me 0 ML 1 nc
20130315 Me 0 ML 1 nc
20130315 NH 0 Pv 1 nc
20130315 Vt 0 BC 1 nc
20130315 Mr 0 BU 1 nc
20130316 NH 0 Pv 1 nc
20130316 Vt 0 BC 1 nc
20130316 Mr 0 BU 1 nc
20130322 BU 0 BC 1 nc
20130322 Pv 0 ML 1 nc
20130323 BC 0 ML 1 nc

20130315 CC 0 DU 1 nc
20130315 NO 0 Mk 1 nc
20130315 AA 0 SC 1 nc
20130315 BS 0 Mn 1 nc
20130315 MT 0 ND 1 nc
20130315 MD 0 Wi 1 nc
20130316 CC 0 DU 1 nc
20130316 NO 0 Mk 1 nc
20130316 AA 0 SC 1 nc
20130316 BS 0 Mn 1 nc
20130316 MT 0 ND 1 nc
20130316 MD 0 Wi 1 nc
20130321 Mk 0 ND 1 nc
20130321 DU 0 Wi 1 nc
20130322 Wi 0 SC 1 nc
20130322 ND 0 Mn 1 nc
20130323 Mn 0 SC 1 nc

20130315 Cr 0 Qn 1 nc
20130315 Bn 0 RP 1 nc
20130315 SL 0 Ya 1 nc
20130315 Da 0 Un 1 nc
20130316 Cr 0 Qn 1 nc
20130316 Bn 0 RP 1 nc
20130316 SL 0 Ya 1 nc
20130316 Da 0 Un 1 nc
20130322 Un 0 Qn 1 nc
20130322 Ya 0 RP 1 nc
20130323 Un 0 Ya 1 nc
20130323 RP 0 Qn 1 nc

20130315 FS 0 OS 1 nc
20130315 MS 0 Mm 1 nc
20130315 BG 0 Nt 1 nc
20130315 Mi 0 WM 1 nc
20130316 FS 0 OS 1 nc
20130316 MS 0 Mm 1 nc
20130316 BG 0 Nt 1 nc
20130316 Mi 0 WM 1 nc
20130322 OS 0 Mm 1 nc
20130322 WM 0 Nt 1 nc
20130323 Nt 0 Mm 1 nc

20130315 Mh 0 HC 1 nc
20130315 Ca 0 AF 1 nc
20130315 RM 0 Ct 1 nc
20130315 RT 0 Ni 1 nc
20130316 Mh 0 HC 1 nc
20130316 Ca 0 AF 1 nc
20130316 RM 0 Ct 1 nc
20130316 RT 0 Ni 1 nc
20130322 HC 0 AF 1 nc
20130322 Ct 0 Ni 1 nc
20130323 AF 0 Ni 1 nc

Again, the higher seed wins every game. To have a lower seed win, change the 0 to a 2. Remember to add a 3rd game if necessary and/or to change future rounds accordingly.
If these results hold (which of course they won't) the final pairwise would be

1 Quinnipiac (EC) 29
2 Minnesota (WC) 28
3 Miami (CC) 27
4 Mass-Lowell (HE) 25
5 St Cloud (WC) 23
6 Boston Coll (HE) 23
7 North Dakota (WC) 22
8 MSU-Mankato (WC) 21
9 Niagara (AH) 21
10 Yale (EC) 20
11 New Hampshire (HE) 20
12 Denver U (WC) 19
13 Western Mich (CC) 16
14 RPI (EC) 16
15 Notre Dame (CC) 14
16 Providence (HE) 13
---
17 Union (EC) 13
18 Wisconsin (WC) 12
19 Boston Univ (HE) 11
20 Air Force (AH) 11

which gives us a nice bracket - if this were next year, when Miami and Notre Dame weren't in the same conference.

Code:
[B]Manchester (UNH)	Providence (Brown)	Toledo (BGSU)		Grand Rapids (Michigan)[/B]
Lowell			Quinnipiac		Miami			Minnesota
St Cloud		Mankato			Boston C		N Dakota
New Hamp		Niagara			Denver			Yale
RPI			Providence		Notre Dame		W Michigan

Too bad it can't happen. Notre Dame goes to Grand Rapids, W Michigan goes to Manchester and RPI goes to Toledo.

when does the pairwise predictor roll out ,we can play around with .......
 
Re: Pairwise and Bracketology 2013 Edition

when does the pairwise predictor roll out ,we can play around with .......

Those usually roll out Sunday night after all the QF's are done. Still way too many permutations before the Quarters.
 
Re: Pairwise and Bracketology 2013 Edition

Those usually roll out Sunday night after all the QF's are done. Still way too many permutations before the Quarters.

for some reason thought it was last night .... Think I asked you the same question last year... promise ill get it straight next year.
 
Re: Pairwise and Bracketology 2013 Edition

for some reason thought it was last night .... Think I asked you the same question last year... promise ill get it straight next year.
No worries. I'm sure you aren't the only one wondering - just the only one who asked :)
 
Re: Pairwise and Bracketology 2013 Edition

when does the pairwise predictor roll out ,we can play around with .......

Predictor comes after this weekend. This script that Priceless talks about is for use with John Whelan's DIY calculator, now hosted on elynah. The site is mentioned about 2 pages back. Well, here: http://www.elynah.com/tbrw/2013/rankings.diy.shtml

Priceless explains how to use it. His script there is assuming all games end 1-0. The only thing that matters, of course, is who wins. You can simulate the rest of the season if you want to, and that is what Pr has done in the above post.

Play with it. IF you get stuck, or can't figure something out, come back here. Some of us are here on and off through the day.
 
Re: Pairwise and Bracketology 2013 Edition

Predictor comes after this weekend. This script that Priceless talks about is for use with John Whelan's DIY calculator, now hosted on elynah. The site is mentioned about 2 pages back. Well, here: http://www.elynah.com/tbrw/2013/rankings.diy.shtml

Priceless explains how to use it. His script there is assuming all games end 1-0. The only thing that matters, of course, is who wins. You can simulate the rest of the season if you want to, and that is what Pr has done in the above post.

Play with it. IF you get stuck, or can't figure something out, come back here. Some of us are here on and off through the day.
Thanks a bunch .
 
Re: Pairwise and Bracketology 2013 Edition

Predictor comes after this weekend. This script that Priceless talks about is for use with John Whelan's DIY calculator, now hosted on elynah. The site is mentioned about 2 pages back. Well, here: http://www.elynah.com/tbrw/2013/rankings.diy.shtml

Priceless explains how to use it. His script there is assuming all games end 1-0. The only thing that matters, of course, is who wins. You can simulate the rest of the season if you want to, and that is what Pr has done in the above post.

Play with it. IF you get stuck, or can't figure something out, come back here. Some of us are here on and off through the day.
numbers Is this the same one that can be found in college hockey news ????
 
Re: Pairwise and Bracketology 2013 Edition

numbers Is this the same one that can be found in college hockey news ????

I don't believe there is a link to it there. The one on CHN only allows to you predict currently scheduled games, so you can't input your own outcomes for the Conference Finals weekend, because CHN's little predictor there doesn't allow that.

The one on elynah requires you to put in your own code. Each team has a 2-letter code, like Priceless has listed above. So, you can literally put in whatever games you want.

To be more precise, if you go to the DIY on elynah, there is a block and at the very top check where it says which things to calculate, check PWR. Then, skip all the middle part, that is about the criteria, because you just want the same as it really is. Then, almost at the bottom of the page, there is a long scroll with all the game results in it. Right above that, there are 2 checks : one says, "Use actual game results from CHN" The other says "Specify results." Click "specify results". Then, click inside the results box, scroll to the end, and add whatever results you want. When you are done, click "Calculate Ratings" at the very bottom. Have fun.
 
Re: Pairwise and Bracketology 2013 Edition

Could someone tell me which conference tourneys have a 3rd place game:

I believe I know this:
WCHA - No
HE - No
ECAC - Yes

Help?
 
Status
Not open for further replies.
Back
Top