Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodies

[ARM]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

Curious what the math is on this site.

Not the actual math, but at least, the thought behind it:

MYHockey's ratings are computed mathematically, with no subjective weighting. MYHockey rates teams based upon how well they play against other teams and how good those teams are. These two factors are "AGD" or "average goal differential" and "SCHED" or "strength of schedule". AGD is currently calculated by accumulating the goal differential of each game, to a maximum of 7, and then dividing it by the number of games played. Using a max game goal differential of 7 does "penalize" teams that blow out opponents, but this is common in rating systems as it takes away the incentive for teams to run up scores and allows them to have a bad game by capping the mathematical damage. The strength of schedule is computed by averaging the rating of each game opponent. AGD and SCHED are then added together to compute a team's official rating. A teamís rating typically is not reflective of the teamís best or worst games, but a mathematically computed average performance rating.

[D2D]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

Not the actual math, but at least, the thought behind it:

Harvard will show up somewhere after Friday's result vs. BU is in the books (minimum of 5 games played).

[mattj711]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

1. Boston College
2. Minnesota
3. Wisconsin
4. Quinnipiac
5. Clarkson
6. Harvard
7. Bemidji
8. UMD
9. Mercyhurst
10. Cornell

[TonyTheTiger20]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

The poll is back yayyyy

Week IX Posters Poll

Rank Votes  Team (1st place votes)  Range (Mode)  Last week (change)
  1. 187    Boston College (17)     1-3   (1)     1 (0)
  2. 168    Minnesota (1)           1-3   (2)     2 (0)
  3. 155    Wisconsin (1)           1-4   (3)     3 (0)
  4. 132    Quinnipiac              3-8   (4)     4 (0)
  5. 109    Clarkson                4-7   (5)     7 (+2)
  6.  78    Bemidji State           5-10  (6)     8 (+2)
  7.  77    Harvard                 5-10  (7)     5 (-2)
  8.  43    Minnesota-Duluth        6-NR  (9)     ARV (+3)
  9.  41    Boston University       5-NR  (10)    6 (-3)
 10.  33    Mercyhurst              6-NR  (8)     9 (-1)
 
Also Receiving Votes:
      16    Cornell                 9-NR  (10)    ARV 
       3    Vermont                 9-NR  (9.5)   ARV
       1    North Dakota            10-NR (10)    10 (-1) 
       1    Princeton               10-NR (10)    NR
       1    St. Lawrence            10-NR (10)    NR
            
Dropped Out:
            Dartmouth
            Ohio State
            Northeastern

Week IX Chart: http://i.imgur.com/1nWFkPH.jpg

[Eeyore]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

So, by this poll, Minnesota has played 7 of its 12 games against teams in the top ten. Wisconsin has played 6. Boston College and Quinnipiac have combined to play . . . uhm, I may run out of fingers . . . 0.

[FiveHoleFrenzy]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

So, by this poll, Minnesota has played 7 of its 12 games against teams in the top ten. Wisconsin has played 6. Boston College and Quinnipiac have combined to play . . . uhm, I may run out of fingers . . . 0.

Why you gotta harsh the mellow of bc and quinnipea-ack. They have combined to play 3 TUCs, so there is that.

[TonyTheTiger20]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

http://www.nashvillescene.com/binary...hate-eagle.jpg

[PuckRush]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

So, by this poll, Minnesota has played 7 of its 12 games against teams in the top ten. Wisconsin has played 6. Boston College and Quinnipiac have combined to play . . . uhm, I may run out of fingers . . . 0.

Correct. And by the same logic, Minnesota and Wisconsin have LOST to top 10 teams. Guess all 4 should move down and make Clarkson #1. Oh... except Clarkson lost to #9 BU. Let's move them down too. Let's make Bemidji #1. Oh... but they lost to Wisconsin and some other non-Top 10 teams. OK, let's move Wisconsin to #1 then. Oh.. but they lost to Minnesota. OK, let's make them #1. Oh... but they lost to Bemidji. So let's put Bemidji at number one. Wait, that's not going to work. Let's put BC at #1 and Quinnipiac at #4. Perfect.

[ARM]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

And by the same logic...

No, that's totally different logic.

[PuckRush]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

No, that's totally different logic.

Actually, my post has no logic at all. Which, is kind of my point

[ARM]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

Actually, my post has no logic at all. Which, is kind of my point

But you can't use it as an excuse to post an image of cupcakes, which is really the only point of Eeyore's post, so that then TTT can post that same image he always does, like he's RStarr or something. Message board debates weren't nearly so graphical back in my day (although now that I think of it, they'd have had to involve push pins.)

[pokechecker]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

Actually, my post has no logic at all. Which, is kind of my point

it looks to me like your point is you cannot predict the outcome of two teams playing one another on the basis of how they played against common opponents. I agree, however the problem is that only 8 teams will get to play in the NCAA tournament, and those teams are selected by the process you say has no logic (and I agree, in fact nearly everyone agrees). So right now BU & Mercyhurst would be out, but wait, the CHA champ has an auto bid. So Mercyhurst or some other CHA team will bump UMD if the tournament was played this weekend.
But is Harvard really better than BU? Is Bemidji better than UMD or any team below? The only way to know is to have them play.
In the end a couple of teams get screwed out of playing in the tournament.

[pokechecker]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

Interesting ranking of Women's D1 on myhockeyrankings, I can't say I would disagree too much. Curious what the math is on this site.
http://myhockeyrankings.com/rank.php?y=2014&a=f&v=903

While I don't disagree with the ranking, using goal differential is just wrong, it rewards teams for playing cupcakes instead of playing tough opponents.
I don't know about the rest of you, but opening weekend I was much more interesting in watching MN take on BU rather than Penn State.

[PuckRush]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

it looks to me like your point is you cannot predict the outcome of two teams playing one another on the basis of how they played against common opponents. I agree, however the problem is that only 8 teams will get to play in the NCAA tournament, and those teams are selected by the process you say has no logic (and I agree, in fact nearly everyone agrees). So right now BU & Mercyhurst would be out, but wait, the CHA champ has an auto bid. So Mercyhurst or some other CHA team will bump UMD if the tournament was played this weekend.
But is Harvard really better than BU? Is Bemidji better than UMD or any team below? The only way to know is to have them play.
In the end a couple of teams get screwed out of playing in the tournament.

You read my mind! So now I get to make my next point, which is what I think about when ranking teams. I rank based on my guess of which team I think will, right now, win a 5 game series on neutral ice against any of the teams below it. The #1 team will win that series against all teams, #2 will win that series against all but the #1 team, and so on. So the CHA conundrum is irrelevant to my ranking. I won't put a CHA team in the top 8 because it satisfies the selection committee's paradigm. I also admit that I have no faith in my predictive abilities when I get to the back half of the top 10!

Is that the consensus on how everyone ranks the Top 10? Share your view... I think it would be interesting and instructive to understand our different approaches.

[Timothy A]

Re: Fun With Numbers '14-'15: USCHO Posters Poll, Pairwise What-Ifs, and Other Goodie

The poll is back yayyyy

Week IX Posters Poll

Rank Votes  Team (1st place votes)  Range (Mode)  Last week (change)
  1. 187    Boston College (17)     1-3   (1)     1 (0)
  2. 168    Minnesota (1)           1-3   (2)     2 (0)
  3. 155    Wisconsin (1)           1-4   (3)     3 (0)
  4. 132    Quinnipiac              3-8   (4)     4 (0)
  5. 109    Clarkson                4-7   (5)     7 (+2)
  6.  78    Bemidji State           5-10  (6)     8 (+2)
  7.  77    Harvard                 5-10  (7)     5 (-2)
  8.  43    Minnesota-Duluth        6-NR  (9)     ARV (+3)
  9.  41    Boston University       5-NR  (10)    6 (-3)
 10.  33    Mercyhurst              6-NR  (8)     9 (-1)
 
Also Receiving Votes:
      16    Cornell                 9-NR  (10)    ARV 
       3    Vermont                 9-NR  (9.5)   ARV
       1    North Dakota            10-NR (10)    10 (-1) 
       1    Princeton               10-NR (10)    NR
       1    St. Lawrence            10-NR (10)    NR
            
Dropped Out:
            Dartmouth
            Ohio State
            Northeastern

Week IX Chart: http://i.imgur.com/1nWFkPH.jpg

Nice chart, but UW should be red, not yellow.