Announcement

Collapse
No announcement yet.

2009-2010 DI Rutter Computer Rankings Thread

Collapse
X
Collapse
 
  • Page of 7
  • Filter
  • Time
  • Show
Clear All
new posts
Previous 1 2 3 4 5 6 7 template Next
  • #1

    2009-2010 DI Rutter Computer Rankings Thread

    Back again for another year of algorithm-induced-bias-only computer based rankings. Latest top 10, for games played through 10/4, are given below.

    For a complete list visit: http://math.bd.psu.edu/faculty/rutte...sRankings.html

    Ratings are based wins, loses, and ties only. Home ice, score, time of year, etc. are not included. Algorithm includes a prior (Bayesian) based on last years results, thus allowing our good friends in the Ivy league to be ranked while they wait for their seasons to start.

    The rating can be used to find the probability of a team winning by finding the area to the left of (your team's rating-opponent's team rating) under a normal curve with a mean of zero and a standard deviation of one (Visit http://www.danielsoper.com/statcalc/calc02.aspx and enter your number). This calculation assumes a tie is not possible. I would be happy to answer any questions, as usual.

    Code:
       	Team  		Rating  
1	Minnesota	1.8423	
2	New Hampshire	1.4638	
3	Mercyhurst	1.3888	
4	Wisconsin	1.1900	
5	St. Lawrence	0.6898	
6	North Dakota	0.6861	
7	Harvard		0.6757	
8	UMD		0.6195	
9	Clarkson	0.5553	
10	Dartmouth	0.5515
    Tags: None
  • #2
    Re: 2009-2010 DI Rutter Computer Rankings Thread

    Wayne State at 20? I'll take that after losing 3 of 4 pts at RPI. It's gonna be an interesting season without the firepower.
    "Here's to the confusion of our enemys!" - Sinatra

    Over 100,000+ posts behind Ralph Baer

    Btw.... Fournier? Where's our rink? FOURNIER! You're an idiot. Bring back our team!!!! LIAR!

    Comment

    • #3
      Re: 2009-2010 DI Rutter Computer Rankings Thread

      For games played through October 12, 2009

      Code:
         	Team  	Rating 
1	Minnesota	2.1687	
2	New Hampshire	1.7208	
3	Mercyhurst	1.5790	
4	Clarkson	1.0389	
5	Harvard		0.6814	
6	UMD		0.6548		
7	Wisconsin	0.6115	
8	Dartmouth	0.5454	
9	Princeton	0.4217	
10	Minnesota State	0.4054
      Clarkson making a move into the top 4 based on their strong (in terms of last year's results and relative to other ranked teams) early season schedule. Wisconsin falls to 7th. Will they crawl back into the top 5, or hover around the 7-10 range, making the NCAA tournament an iffy proposition?

      Comment

      • #4
        Re: 2009-2010 DI Rutter Computer Rankings Thread

        For games played through October 19, 2009

        Code:
           	Team  		Rating  
1	Minnesota	2.3869	
2	Mercyhurst	1.8479
3	New Hampshire	1.1100	
4	Harvard		0.6852	
5	Minnesota State	0.6732	
6	Dartmouth	0.5546	
7	Wisconsin	0.5167	
8	Clarkson	0.4643	
9	Princeton	0.4191	
10	UMD		0.4017
        Minnesota State hits the top 5, but UMD falls to number 10. Is this the year one of the WCHA big three fails to make the NCAA tournament?

        Comment

        • #5
          Re: 2009-2010 DI Rutter Computer Rankings Thread

          Originally posted by LakersFan View Post
          For games played through October 19, 2009

          Code:
             	Team  		Rating  
1	Minnesota	2.3869	
2	Mercyhurst	1.8479
3	New Hampshire	1.1100	
4	Harvard		0.6852	
5	Minnesota State	0.6732	
6	Dartmouth	0.5546	
7	Wisconsin	0.5167	
8	Clarkson	0.4643	
9	Princeton	0.4191	
10	UMD		0.4017
          Minnesota State hits the top 5, but UMD falls to number 10. Is this the year one of the WCHA big three fails to make the NCAA tournament?
          Clarky, 4th to 8th. Ouch.
          (where the heart beats)

          bleep.

          Comment

          • #6
            Re: 2009-2010 DI Rutter Computer Rankings Thread

            Originally posted by LakersFan View Post
            Code:
               	Team  		Rating  	
3	New Hampshire	1.1100
            Originally posted by LakersFan View Post
            Code:
               	Team  	Rating 	
2	New Hampshire	1.7208
            Wow, that 4-3 tie really hurt UNHs rating compared to last week.
            113-162-27 .419

            252-113-40 .672

            Comment

            • #7
              Re: 2009-2010 DI Rutter Computer Rankings Thread

              Originally posted by RStarr View Post
              Clarky, 4th to 8th. Ouch.
              Suspect that early season surprises tend to have a larger impact on the rankings, as there are less games being used to calculate the final raw numbers.

              Comment

              • #8
                Re: 2009-2010 DI Rutter Computer Rankings Thread

                I try to minimize the impact of early season games by having each team "win" a virtual game against the 2008-09 version of themselves (i.e. the Bayesian Prior). If a top 10 team loses to another top 10 team early, the effect is small. If a top 10 team loses or ties a team with a negative rating (UNH/Niagara or Clarkson/Vermont), the effect is a more dramatic. Right now these results make up 1/5 of a teams rating. By the end of the year, it will be 1/30 or less. As long as these results are not trends, the long term impact is small.

                And I have not yet added tie/win/loss due to ref incompetency into my model . Maybe next summer.

                Comment

                • #9
                  Re: 2009-2010 DI Rutter Computer Rankings Thread

                  Originally posted by LakersFan View Post
                  I try to minimize the impact of early season games by having each team "win" a virtual game against the 2008-09 version of themselves (i.e. the Bayesian Prior). If a top 10 team loses to another top 10 team early, the effect is small. If a top 10 team loses or ties a team with a negative rating (UNH/Niagara or Clarkson/Vermont), the effect is a more dramatic. Right now these results make up 1/5 of a teams rating. By the end of the year, it will be 1/30 or less. As long as these results are not trends, the long term impact is small.

                  And I have not yet added tie/win/loss due to ref incompetency into my model . Maybe next summer.
                  Just what I suspected. Thanks for the technical explanation bringing me back to math class form yesteryear.

                  This means the variations/movements from week to week will be greater early in the year.

                  Comment

                  • #10
                    Re: 2009-2010 DI Rutter Computer Rankings Thread

                    Originally posted by LakersFan View Post
                    I try to minimize the impact of early season games by having each team "win" a virtual game against the 2008-09 version of themselves (i.e. the Bayesian Prior). If a top 10 team loses to another top 10 team early, the effect is small. If a top 10 team loses or ties a team with a negative rating (UNH/Niagara or Clarkson/Vermont), the effect is a more dramatic. Right now these results make up 1/5 of a teams rating. By the end of the year, it will be 1/30 or less. As long as these results are not trends, the long term impact is small.

                    And I have not yet added tie/win/loss due to ref incompetency into my model . Maybe next summer.
                    I usually like the prior and think it's informative, but not so much this year. How much would the rankings change under a totally neutral prior, if that's possible?

                    Comment

                    • #11
                      Re: 2009-2010 DI Rutter Computer Rankings Thread

                      Originally posted by dave1381 View Post
                      I usually like the prior and think it's informative, but not so much this year. How much would the rankings change under a totally neutral prior, if that's possible?
                      These results are without the prior, so they are a reflection of the rankings if the season finished today (after 0-6 games, depending on the team). Note, the Ivy League schools all have 0 ratings because they have not played a game.

                      WARNING: If you are fan of a western teams, you may not like this result.


                      Code:
                         	Team  	Rating  
1	Minnesota	1.45
2	Mercyhurst	1.21
3	New Hampshire	0.89
4	Clarkson	0.67
5	Vermont		0.61
6	Minnesota State	0.55
7	Northeastern	0.22
8	Boston University	0.18
9	Quinnipiac	0.15
10	Rensselaer	0.03
11	Ohio State	0.02
12	Brown		0
13	Princeton	0
14	Cornell		0
15	Yale		0
16	Dartmouth	0
17	Harvard		0
18	St. Lawrence	-0.02
19	Providence	-0.03
20	Connecticut	-0.09
21	North Dakota	-0.09
22	Minnesota Duluth	-0.14
23	Boston College	-0.16
24	Wisconsin	-0.18
25	Niagara		-0.27
26	Syracuse	-0.3
27	Wayne State	-0.38
28	Bemidji State	-0.42
29	Colgate		-0.5
30	St. Cloud State	-0.71
31	Maine		-0.87
32	Robert Morris	-0.87
33	Union		-0.93

                      Comment

                      • #12
                        Re: 2009-2010 DI Rutter Computer Rankings Thread

                        I like it because my Mavericks and Gophers are both in the top 8.

                        Originally posted by LakersFan View Post
                        These results are without the prior, so they are a reflection of the rankings if the season finished today (after 0-6 games, depending on the team). Note, the Ivy League schools all have 0 ratings because they have not played a game.

                        WARNING: If you are fan of a western teams, you may not like this result.


                        Code:
                           	Team  	Rating  
1	Minnesota	1.45
2	Mercyhurst	1.21
3	New Hampshire	0.89
4	Clarkson	0.67
5	Vermont		0.61
6	Minnesota State	0.55
7	Northeastern	0.22
8	Boston University	0.18
9	Quinnipiac	0.15
10	Rensselaer	0.03
11	Ohio State	0.02
12	Brown		0
13	Princeton	0
14	Cornell		0
15	Yale		0
16	Dartmouth	0
17	Harvard		0
18	St. Lawrence	-0.02
19	Providence	-0.03
20	Connecticut	-0.09
21	North Dakota	-0.09
22	Minnesota Duluth	-0.14
23	Boston College	-0.16
24	Wisconsin	-0.18
25	Niagara		-0.27
26	Syracuse	-0.3
27	Wayne State	-0.38
28	Bemidji State	-0.42
29	Colgate		-0.5
30	St. Cloud State	-0.71
31	Maine		-0.87
32	Robert Morris	-0.87
33	Union		-0.93

                        Comment

                        • #13
                          Re: 2009-2010 DI Rutter Computer Rankings Thread

                          Graph of rating over time this season. Very confusing in the middle of the pack. Next week I will try ranking, which should be cleaner. And for those of you who are color blind, my apologizes. Still playing around with the visualization. Note, teams are ordered by the latest ranking.

                          *****http://math.bd.psu.edu/faculty/rutter/graph.png******

                          Comment

                          • #14
                            Re: 2009-2010 DI Rutter Computer Rankings Thread

                            If you turn it 90 degrees, I think I see the Mercury space rocket!




                            http://en.wikipedia.org/wiki/File:La...000-000686.jpg
                            Toe Blake On goalies: "You get four goals off them, or five, but the goal you've got to have to win, somehow the great ones don't let you get it.”

                            Comment

                            • #15
                              Re: 2009-2010 DI Rutter Computer Rankings Thread

                              Originally posted by LakersFan View Post
                              These results are without the prior, so they are a reflection of the rankings if the season finished today (after 0-6 games, depending on the team). Note, the Ivy League schools all have 0 ratings because they have not played a game.
                              Thanks. I appreciate it. At this point I really have no clue which of these will be more accurate in the end (among the teams who've already played).

                              Comment

                              Previous 1 2 3 4 5 6 7 template Next
                              Working...
                              X