Announcement

Collapse
No announcement yet.

ECAC Pick the Standings 2017-18

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • lugnut92
    replied
    Re: ECAC Pick the Standings 2017-18

    Here are the statistics through 2 days of entries (not including the Random Number and Mr. Last Year; n=25):
    Code:
    Team       |  Mean±SD   | Median(Range) |
    -----------------------------------------
    Harvard    |  1.80±1.80 |   1   ( 1-10) |
    Quinnipiac |  2.40±1.04 |   2   ( 1- 4) |
    Clarkson   |  3.84±2.44 |   3   ( 1-12) |
    Cornell    |  3.84±1.21 |   4   ( 2- 7) |
    Princeton  |  5.88±2.01 |   5   ( 3-12) |
    Yale       |  6.80±1.38 |   7   ( 4-10) |
    Union      |  6.92±1.98 |   7   ( 3-11) |
    RPI        |  7.48±2.90 |   8   ( 1-12) |
    SLU        |  8.64±2.04 |   9   ( 3-12) |
    Dartmouth  |  9.48±2.22 |  10   ( 2-12) |
    Colgate    | 10.00±1.73 |  10   ( 5-12) |
    Brown      | 11.00±1.35 |  12   ( 7-12) |
    Entries slowed down quite a bit yesterday, so I guess it's time to spam the team threads again.

    Leave a comment:


  • Ralph Baer
    replied
    Re: ECAC Pick the Standings 2017-18

    Originally posted by lugnut92 View Post
    The tiebreakers listed are for ties between contestants, not for the data actually being compared. Luckily the ECAC provides a list of tiebreakers for that situation (though some of those tiebreakers are questionable). I don't think you would find anything horrifically wrong about using an average value (e.g., 6.5 for two teams tied for 6th), it just wouldn't be a true correlation coefficient. Further, you could use the (slightly more complicated) formula listed above the formula I'm using if there were ties.
    Yes, that was my error. Anyway, thanks for the explanation.

    Leave a comment:


  • lugnut92
    replied
    Re: ECAC Pick the Standings 2017-18

    Originally posted by Ralph Baer View Post
    Anyway, the Wikipedia page which you linked mentioned that it should not be used if there are ties and you went to great length to eliminate ties. What terrible things would occur if one just used an average value, or would a weighted average in some position-dependent manner need to be used in order that a number outside of -1 to 1 do not occur?
    The tiebreakers listed are for ties between contestants, not for the data actually being compared. Luckily the ECAC provides a list of tiebreakers for that situation (though some of those tiebreakers are questionable). I don't think you would find anything horrifically wrong about using an average value (e.g., 6.5 for two teams tied for 6th), it just wouldn't be a true correlation coefficient. Further, you could use the (slightly more complicated) formula listed above the formula I'm using if there were ties.

    Leave a comment:


  • Ralph Baer
    replied
    Re: ECAC Pick the Standings 2017-18

    Originally posted by lugnut92 View Post
    I'm changing things ever so slightly with how scores will appear, though the mechanics are the same in the end. Rather than a simple square of the distance between your prediction and the actual standings, I'm going to report the Spearman's rank correlation coefficient between each participant's predictions and the real standings. This change is just because I am difficult (and if people really complain, I can report both metrics). In the end, all scores will range between 1 and -1 (rather than between 0 and 572), and the CLOSEST TO 1 WINS. You have until the puck drops on the first ECAC contests (RPI at Union/Brown at Yale; Friday, 27 October at 7 PM EDT) to get your picks in.

    Tie breakers are:

    Pick correctly the ECACHL Regular Season Champ.
    Have the most teams in the correct order of finish.
    Have the most top 6 teams in the correct order of finish.
    I will flip a coin.
    I had not heard of Spearman's rank correlation coefficient, and I just checked it isn't mentioned either in the book which we used when I took a Prob and Stat course as a senior or in the book which I used when I taught an Intro Stat course one summer when I was a grad TA. It probably wasn't invented yet.

    Anyway, the Wikipedia page which you linked mentioned that it should not be used if there are ties and you went to great length to eliminate ties. What terrible things would occur if one just used an average value, or would a weighted average in some position-dependent manner need to be used in order that a number outside of -1 to 1 do not occur?

    Leave a comment:


  • Saints#1
    replied
    Re: ECAC Pick the Standings 2017-18

    1. Harvard
    2. Quinnipiac
    3. Clarkson
    4. Cornell
    5. Princeton
    6. St. Lawrence
    7. Yale
    8. Union
    9. RPI
    10. Colgate
    11. Dartmouth
    12. Brown

    Leave a comment:


  • Rainman
    replied
    Re: ECAC Pick the Standings 2017-18

    Thanks lugnut for keeping score this year! I'm running the ECAC Pick'em contest if any of you want to enter

    1. Harvard
    2. Quinnipiac
    3. Clarkson
    4. Cornell
    5. Princeton
    6. Yale
    7. Union
    8. Colgate
    9. St. Lawrence
    10. Dartmouth
    11. RPI
    12. Brown

    Leave a comment:


  • cudmud
    replied
    Re: ECAC Pick the Standings 2017-18

    1. Quinnipiac
    2. Harvard
    3. Cornell
    4. Yale
    5. Princeton
    6. RPI
    7. Union
    8. Clarkson
    9. Colgate
    10. SLU
    11. Dartmouth
    12. Brown

    Leave a comment:


  • Turk
    replied
    Re: ECAC Pick the Standings 2017-18

    1. Quinnipiac
    2. Harvard
    3. Clarkson
    4. Cornell
    5. Princeton
    6. RPI
    7. Yale
    8. Union
    9. Dartmouth
    10. Colgate
    11. St. Lawrence
    12. Brown

    Leave a comment:


  • Section5forLife
    replied
    Re: ECAC Pick the Standings 2017-18

    1. Harvard
    2. Qunnipiac
    3. Union
    4. Clarkson
    5. RPI
    6. Princeton
    7. Cornell
    8. Colgate
    9. Yale
    10. Brown
    11. Dartmouth
    12. St. Lawrence

    Leave a comment:


  • lugnut92
    replied
    Re: ECAC Pick the Standings 2017-18

    Here are the statistics through ~1 day of entries (not including the Random Number and Mr. Last Year; n=20):
    Code:
    Team       |  Mean±SD   | Median(Range) |
    -----------------------------------------
    Harvard    |  1.90±2.00 |   1   ( 1-10) |
    Quinnipiac |  2.60±1.05 |   2.5 ( 1- 4) |
    Cornell    |  3.70±1.13 |   4   ( 2- 6) |
    Clarkson   |  3.75±2.55 |   3   ( 1-12) |
    Princeton  |  6.05±2.21 |   5   ( 3-12) |
    Yale       |  6.85±1.31 |   7   ( 5-10) |
    Union      |  7.00±2.00 |   6.5 ( 4-11) |
    RPI        |  7.50±3.05 |   8.5 ( 1-12) |
    SLU        |  8.40±1.96 |   8.5 ( 3-11) |
    Dartmouth  |  9.25±2.40 |   9   ( 2-12) |
    Colgate    | 10.25±1.80 |  11   ( 5-12) |
    Brown      | 10.85±1.42 |  11   ( 7-12) |

    Leave a comment:


  • UFan
    replied
    Re: ECAC Pick the Standings 2017-18

    1. Quinnipiac
    2. Harvard
    3. Cornell
    4. Union
    5. Clarkson
    6. St. Lawrence
    7. Princeton
    8. Colgate
    9. RPI
    10. Yale
    11. Dartmouth
    12. Brown

    Leave a comment:


  • FlagDUDE08
    replied
    Re: ECAC Pick the Standings 2017-18

    No surprise, RPI's the first to receive both a first and last place vote. Pretty much an every year occurrence.

    CCT in second this time.

    Leave a comment:


  • FlagDUDE08
    replied
    Re: ECAC Pick the Standings 2017-18

    I guess I should make some picks, then.

    1: Quinnipiac. Unless proved otherwise.
    2: Hahvahd, so long as Madsen remains healthy.
    3: Cornell. They're Cornell.
    4: Princeton. My longshot for the year.
    5: RPI. Compared to the old guy, the new guy is night and day.
    6: CCT. I don't know why people keep getting fooled by October. They'll prove in time they're not back.
    7: Ylae*. Pretty much had to fit them in somewhere.
    8: SLU. I really just don't see them struggling all season long...
    9: Dartmouth. Always a perplexing choice...
    10: Brown. Getting better, but still a ways to go.
    11: #26 Union. They like to copy Troy, but let's just say it's not for the good this time around.
    12: Colgate. New barn, but not that new of a product on the ice.

    Leave a comment:


  • joecct
    replied
    Re: ECAC Pick the Standings 2017-18

    RPI Standings Board
    1. RPI
    2. Dartmouth
    3. SLU
    4. QU
    5. Colgate
    6. Cornell
    7. Yale
    8. Brown
    9. Princeton
    10. Harvard
    11. Union
    12. Clarkson

    Leave a comment:


  • Ralph Baer
    replied
    Re: ECAC Pick the Standings 2017-18

    Originally posted by joecct View Post
    Ralph

    You posted a picture on one of the threads of the standings board at Houston. There was talk at the time of making that the RPI entry in this contest. I can't find it. If you can, would you post the RPI entry??

    All I can recall is that you were in 1st and we were in last.
    I posted the original tweet of the men's board http://board.uscho.com/showthread.ph...=1#post6528326

    Lugnut92 himself posted this image http://board.uscho.com/showthread.ph...=1#post6531130 of the women's board.
    Last edited by Ralph Baer; 10-24-2017, 03:15 AM.

    Leave a comment:

Working...
X