Announcement

Collapse
No announcement yet.

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

Collapse
This topic is closed.
X
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

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

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

    Comment


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

      Originally posted by TonyTheTiger20 View Post
      I found it --



      Not following your math, though... BC is 1st in both polls and behind only Dartmouth in the PWR/RPI?

      Even if you're including Rutter, that still wouldn't be enough to get Minnesota ahead of BC (1-2-3 for BC, 2-3-1 for Minnesota).
      since two of the polls are essentially the same, I average them and end up with 1 poll
      I use KRACH and Rutter as well, so there are 4 polls
      but USA Today comes out late, so sometimes I don't use it
      so:
      KRACH
      Rutter
      coaches poll (average the 2)
      RPI type polls(average the 2)
      so I am using 4 polling types to come up with a number that simply is the sum of the places in the 4 polls

      and it's always possible I miscalculated, I 'll recalculate later today.

      Comment


      • Originally posted by pokechecker View Post
        since two of the polls are essentially the same, I average them and end up with 1 poll
        I use KRACH and Rutter as well, so there are 4 polls
        but USA Today comes out late, so sometimes I don't use it
        so:
        KRACH
        Rutter
        coaches poll (average the 2)
        RPI type polls(average the 2)
        so I am using 4 polling types to come up with a number that simply is the sum of the places in the 4 polls

        and it's always possible I miscalculated, I 'll recalculate later today.
        I see what happened -- KRACH is still up from last year; it doesn't have this year's rankings up yet.
        Grant Salzano, Boston College '10
        Writer Emeritus, BC Interruption
        Twitter: @Salzano14


        Click here for the BC Interruption Pairwise, KRACH, and GRaNT Calculators

        Comment


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

          1. UW
          2. UM
          3. BC
          4. Hawvawd
          5. Q
          6. Clarkson
          7. Bemiji
          8. Mercy
          9. UMD
          10. UND
          Wisconsin Hockey: 1, 3, 5, 7, 9, 11 WE WANT MORE!
          ------------------------------------------------------------------------------
          Come to the Wisconsin Evangelical Lutheran Synod
          ------------------------------------------------------------------------------
          Originally Posted by Wisko McBadgerton:
          "Baggot says Hughes and Rockwood are centering the top two lines...
          Timothy A --> Great hockey mind... Or Greatest hockey mind?!?"

          Comment


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

            Originally posted by Trillium View Post
            Keeping it the same after Sun-Tues scores
            I hate to say I might be wroo... wroooooo.... wroooooon.... wroooooooong. OK, I said it.

            Mercyhurst showed me something vs OSU. They probably deserve a top 10 for now given how confusing all 4 conferences are making me.

            Comment


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

              Originally posted by TonyTheTiger20 View Post
              I see what happened -- KRACH is still up from last year; it doesn't have this year's rankings up yet.
              DOH!

              I was wondering how BC could be below Clarkson, Cornell, UND ...
              so corrected:

              1 BC 2-1-3=6
              1 MN 3-2-1=6

              3 WI 5-3-2=10

              4 Harvard 6-4-4=14
              4 Quinnipiac 4-5-5=14

              6 Dartmouth 1-10-10=21
              7 Clarkson 7-7-8=22

              8 BU 10-6-9=25
              9 Bemidji 11-10-6=27

              10 Murcyhurst 9-8-14=31
              11 UMD 13-9-11=33

              really, Dartmouth & Clarkson, BU & BSU, Murcyhurst & UMD are tied for their positions as well

              Comment


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

                So does anyone know how the RPI 'adjustment to remove negative effect from defeating weak opponent' works?
                Go Gophers!

                Comment


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

                  Originally posted by 5mn_Major View Post
                  So does anyone know how the RPI 'adjustment to remove negative effect from defeating weak opponent' works?
                  If a win would lower a team's RPI it is deleted from their record and not included in the calculation.

                  Comment


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

                    Originally posted by Eeyore View Post
                    If a win would lower a team's RPI it is deleted from their record and not included in the calculation.
                    AFAICT the way it would affect the RPI is through the opponents' win %, yes? So if the win % of defeated opponent A is less than the average win % of all your opponents, it gets dropped. But the win still counts towards your own win %. Is that right?

                    Comment


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

                      Originally posted by wwhyte View Post
                      AFAICT the way it would affect the RPI is through the opponents' win %, yes? So if the win % of defeated opponent A is less than the average win % of all your opponents, it gets dropped. But the win still counts towards your own win %. Is that right?
                      No. If the game would lower the RPI, then it is dropped entirely, like it never existed. During Minnesota's perfect season, the RPI at selection time was based on five wins over North Dakota, and everything else was dropped because everything else would lower the RPI because all of the other teams were weaker opponents. Yet another thing that RPI is not capable of deciphering.
                      "... And lose, and start again at your beginnings
                      And never breathe a word about your loss;" -- Rudyard Kipling

                      Comment


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

                        Originally posted by wwhyte View Post
                        AFAICT the way it would affect the RPI is through the opponents' win %, yes? So if the win % of defeated opponent A is less than the average win % of all your opponents, it gets dropped. But the win still counts towards your own win %. Is that right?
                        No, it's dropped completely. To expand on what Arlan said, if including the game would drop your opponents' winning percentage times that factor's coefficient by more than it increases your winning own percentage times its coefficient, then the game is dropped. (Technically we also have to look at its effect on opponents' opponents' winning percentage but that probably isn't meaningful in most cases.

                        Comment


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

                          Originally posted by Eeyore View Post
                          No, it's dropped completely. To expand on what Arlan said, if including the game would drop your opponents' winning percentage times that factor's coefficient by more than it increases your winning own percentage times its coefficient, then the game is dropped. (Technically we also have to look at its effect on opponents' opponents' winning percentage but that probably isn't meaningful in most cases.
                          It's all very science.

                          Comment


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

                            Science...

                            https://www.youtube.com/watch?v=-FIM...=RD-FIMvSp01C8
                            At the outset, we could hang with the dude...

                            Comment


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

                              I can give you guys an example.

                              Picture each game as having a "Game RPI" -- 1.000 times .3 for a win, .000 times .3 for a loss, then opponents winning percentage time .24 and oppoppwin% times .46.

                              Say this team has played 10 games. If that "Game RPI" is lower than their season RPI (and it was a win), you subtract that game RPI from the sum of all of the game RPIs, and divide by 9 (total number of games minus that game).

                              Maybe that explanation made things worse haha but that's another way of looking at it. It's how I thought through making my spreadsheet's calculation.
                              Grant Salzano, Boston College '10
                              Writer Emeritus, BC Interruption
                              Twitter: @Salzano14


                              Click here for the BC Interruption Pairwise, KRACH, and GRaNT Calculators

                              Comment


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

                                Originally posted by TonyTheTiger20 View Post
                                I can give you guys an example.

                                Picture each game as having a "Game RPI" -- 1.000 times .3 for a win, .000 times .3 for a loss, then opponents winning percentage time .24 and oppoppwin% times .46.

                                Say this team has played 10 games. If that "Game RPI" is lower than their season RPI (and it was a win), you subtract that game RPI from the sum of all of the game RPIs, and divide by 9 (total number of games minus that game).

                                Maybe that explanation made things worse haha but that's another way of looking at it. It's how I thought through making my spreadsheet's calculation.


                                you guys are nuts

                                Comment

                                Working...
                                X