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.
2019 USCHO Pairwise Predictor
- Thread starter ticapnews
- Start date
Ralph Baer
Let's Go 'Tute!
Re: 2019 USCHO Pairwise Predictor
Their seniors have started to turn pro, so even if somehow the stars align, they will be short.
Get the golf clubs out.
Their seniors have started to turn pro, so even if somehow the stars align, they will be short.
Yep. They seem to have, without notice or comment, modified it to mirror my results and CHN's. So, you can now use the USCHO predictor to verify the Union scenarios, for example, which you previously couldn't:
http://pwp.uscho.com/rankings/pairwise-predictor/?uniq=pwp_5c92f2e092927
The Exiled One
New member
Re: 2019 USCHO Pairwise Predictor
https://twitter.com/TheExiledOne/status/1108477199861063681
They fixed it after I tweeted at them.
https://twitter.com/TheExiledOne/status/1108477199861063681
Re: 2019 USCHO Pairwise Predictor
From what I understand, we changed the Big Ten and WCHA games to home/away instead of neutral in one part of the script on Monday, but a bug down the line in the RPI library (which we've replaced except in this script) caused by an upgrade of PHP and the resulting deprecation of an expression broke that. I don't think we were trying to mirror anyone. You're welcome at any time to reach out to more than one site when you see an issue.
Thanks for taking the effort to reach out, but that wasn't the issue. The common opponents was and is calculated correctly. There's a really good explanation of how that works here: https://www.uscho.com/faq/pairwise-rankings-explanation/
From what I understand, we changed the Big Ten and WCHA games to home/away instead of neutral in one part of the script on Monday, but a bug down the line in the RPI library (which we've replaced except in this script) caused by an upgrade of PHP and the resulting deprecation of an expression broke that. I don't think we were trying to mirror anyone. You're welcome at any time to reach out to more than one site when you see an issue.
Thanks for taking the effort to reach out, but that wasn't the issue. The common opponents was and is calculated correctly. There's a really good explanation of how that works here: https://www.uscho.com/faq/pairwise-rankings-explanation/
TheRevengeance
2016 Hockey East Overlords
Re: 2019 USCHO Pairwise Predictor
If it was working correctly, please explain how there is any situation wherein NU should win the common opponents comparison over SCSU, which they clearly do in the screenshot he provided. Their only common opponents are Union (.000 for both teams) and BC (1.000 for SCSU and something between .500 and .750 for NU). If it works right the SCSU number should never be anything but 1.000 and the NU number should never be able to surpass it.
If it was working correctly, please explain how there is any situation wherein NU should win the common opponents comparison over SCSU, which they clearly do in the screenshot he provided. Their only common opponents are Union (.000 for both teams) and BC (1.000 for SCSU and something between .500 and .750 for NU). If it works right the SCSU number should never be anything but 1.000 and the NU number should never be able to surpass it.
cetihcra
New member
Re: 2019 USCHO Pairwise Predictor
"Starting in the 2011-12 season, the common opponent calculation compares the sum of the winning percentages against each opponent"...per this, CHN isn't calculating COp correctly, but you're correct. USCHO's table HERE shows the correct data. CHN has SCSU's COp w/ NE as .500 to .333.
r
"Starting in the 2011-12 season, the common opponent calculation compares the sum of the winning percentages against each opponent"...per this, CHN isn't calculating COp correctly, but you're correct. USCHO's table HERE shows the correct data. CHN has SCSU's COp w/ NE as .500 to .333.
r
ScoobyFanClub
Registered User
Re: 2019 USCHO Pairwise Predictor
Summing the COP percentages ... and averaging the COP percentages ... give you the exact same thing. It doesn't matter either way. So CHN is not wrong, per se, and neither is USCHO. ... Since everyone is coming up with the same results in the Pairwise, it's doubtful anyone has it wrong.
Summing the COP percentages ... and averaging the COP percentages ... give you the exact same thing. It doesn't matter either way. So CHN is not wrong, per se, and neither is USCHO. ... Since everyone is coming up with the same results in the Pairwise, it's doubtful anyone has it wrong.
cetihcra
New member
Re: 2019 USCHO Pairwise Predictor
There's a generic example in the USCHO explanation page that Ed posted earlier where there is a difference between summing and averaging the COp percentages, it could flip a comparison in extreme cases. I think the example is somebody with a 4-3-1 COp beats a team with a 10-3-1.
r
Edit: gotcha, averaging the W% against each team will yield the same result, my bad.
There's a generic example in the USCHO explanation page that Ed posted earlier where there is a difference between summing and averaging the COp percentages, it could flip a comparison in extreme cases. I think the example is somebody with a 4-3-1 COp beats a team with a 10-3-1.
r
Edit: gotcha, averaging the W% against each team will yield the same result, my bad.
ScoobyFanClub
Registered User
Re: 2019 USCHO Pairwise Predictor
yeah - just to be clear, that's what I meant ... they're both still using the "new" method of going by the W% vs. each team, and not the total W-L ... so at that point, if you add those averages together - or average the averages - it's the same thing, since the divisor is the same for both teams.
yeah - just to be clear, that's what I meant ... they're both still using the "new" method of going by the W% vs. each team, and not the total W-L ... so at that point, if you add those averages together - or average the averages - it's the same thing, since the divisor is the same for both teams.
The Exiled One
New member
Re: 2019 USCHO Pairwise Predictor
SCENARIO FROM YESTERDAY...
<img src=https://i.imgur.com/qrhHbEW.png></img>
SAME SCENARIO TODAY...
<img src=https://i.imgur.com/ty8nhnF.png></img>
https://pwp.uscho.com/rankings/pairwise-predictor/?uniq=pwp_5c93db55c42f0
Your predictor is still broken, as Northeastern cannot get to a 4-3 common opponent record with SCSU.
Then how come it looks different than the screenshot I made yesterday?
SCENARIO FROM YESTERDAY...
<img src=https://i.imgur.com/qrhHbEW.png></img>
SAME SCENARIO TODAY...
<img src=https://i.imgur.com/ty8nhnF.png></img>
https://pwp.uscho.com/rankings/pairwise-predictor/?uniq=pwp_5c93db55c42f0
Your predictor is still broken, as Northeastern cannot get to a 4-3 common opponent record with SCSU.
Last edited:
TheRevengeance
2016 Hockey East Overlords
Re: 2019 USCHO Pairwise Predictor
Are you sure you're not breaking it by leaving the NCHC consolation blank? It looks like when that game is filled SCSU goes back up to 1st.
Are you sure you're not breaking it by leaving the NCHC consolation blank? It looks like when that game is filled SCSU goes back up to 1st.
The Exiled One
New member
Re: 2019 USCHO Pairwise Predictor
Not for me it doesn't. Look at the individual comparisons and see what it shows for St. Cloud and Northeastern in the common opponents column.
Re: 2019 USCHO Pairwise Predictor
I don't know what's going on in your browser but when I follow your link, I get no consolation winner for NCHC. When I pick CC, Northeastern is 5th. When I pick UMD, Northeastern is 6th. And in both cases it's 1-1-0 St. Cloud (from 1-0 vs. BC and 0-1 vs. Union) and 3-3-0 Northeastern (3-1 vs. BC and 0-2 vs. Union), resulting 1.000 SCSU and .750 NU. We can't reproduce what you are seeing.
The browser stores your choices in a session that is associated with a cookie. It's possible that you've got double information stored somehow. Please try the Start Over button or if that doesn't work, clearing cookies for the domain pwp.uscho.com.
I don't know what's going on in your browser but when I follow your link, I get no consolation winner for NCHC. When I pick CC, Northeastern is 5th. When I pick UMD, Northeastern is 6th. And in both cases it's 1-1-0 St. Cloud (from 1-0 vs. BC and 0-1 vs. Union) and 3-3-0 Northeastern (3-1 vs. BC and 0-2 vs. Union), resulting 1.000 SCSU and .750 NU. We can't reproduce what you are seeing.
The browser stores your choices in a session that is associated with a cookie. It's possible that you've got double information stored somehow. Please try the Start Over button or if that doesn't work, clearing cookies for the domain pwp.uscho.com.
Re: 2019 USCHO Pairwise Predictor
Is there any way a team other than St. Cloud is #1, therefore putting them in Fargo?
Is there any way a team other than St. Cloud is #1, therefore putting them in Fargo?
Re: 2019 USCHO Pairwise Predictor
Fargo is the closest regional to St. Cloud and since host North Dakota is out, that seems the most likely scenario. You’d expect that the NCAA would fly the AHA winner there, but I suppose there’s a scenario in which the number 15 seed could swap there for travel reasons.
Sent from my iPhone using Tapatalk
Fargo is the closest regional to St. Cloud and since host North Dakota is out, that seems the most likely scenario. You’d expect that the NCAA would fly the AHA winner there, but I suppose there’s a scenario in which the number 15 seed could swap there for travel reasons.
Sent from my iPhone using Tapatalk
Re: 2019 USCHO Pairwise Predictor
FINALS:
Northastern 2, BU 1
AIC 3, Robt Morris 2
Cornell 6, Brown 0
St Cloud 5, CC 2
Is Union out now? I thought their path forward was predicated on CC winning their tournament.
FINALS:
Northastern 2, BU 1
AIC 3, Robt Morris 2
Cornell 6, Brown 0
St Cloud 5, CC 2
Is Union out now? I thought their path forward was predicated on CC winning their tournament.
ticapnews
Never Forget!
Re: 2019 USCHO Pairwise Predictor
If Boston College doesn't win this game (0-0 1st) tomorrow could be very boring. It looks like Harvard, Cornell and BGSU are in, the winner of the B1G and AHA are in so it comes down to either BC or Providence. Tomorrow will be about teams in the 4 seed spot (ND/PSU, Cornell, Harvard, BGSU) trying to snag a 3 seed and avoid playing one of the big boys.
We now know that St Cloud will be #1 in Fargo and play the AHA champion. We can guess that if BC loses tonight the Friars will be placed in Providence as the 4 seed. That means the UMass bracket (likely 2-7-10-15) will be in Manchester.
As I type this BC takes a 1-0 lead with just :13 left in the first period. Killer.
If Boston College doesn't win this game (0-0 1st) tomorrow could be very boring. It looks like Harvard, Cornell and BGSU are in, the winner of the B1G and AHA are in so it comes down to either BC or Providence. Tomorrow will be about teams in the 4 seed spot (ND/PSU, Cornell, Harvard, BGSU) trying to snag a 3 seed and avoid playing one of the big boys.
We now know that St Cloud will be #1 in Fargo and play the AHA champion. We can guess that if BC loses tonight the Friars will be placed in Providence as the 4 seed. That means the UMass bracket (likely 2-7-10-15) will be in Manchester.
As I type this BC takes a 1-0 lead with just :13 left in the first period. Killer.