Announcement

Collapse
No announcement yet.

John t whelan ranking simulator

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

  • Patman
    replied
    Originally posted by goblue78 View Post
    It clearly doesn't make sense to lose a game in order to play on the road. I was proposing that the team with the home ice advantage have the option to play on the road. That could clearly be in their advantage. Take a case of 5-12 matchup. That could easily be 65 percent at home and 60 percent on the road. Now the numbers work out.
    Right but its an unavailable choice... Of course this is ultimately the question of if teams should be road warriors.

    Leave a comment:


  • Numbers
    replied
    Re: John t whelan ranking simulator

    Has anyone figured out yet why FlagDUDE's numbers don't match JimDahl or RHamilton?

    Leave a comment:


  • goblue78
    replied
    Re: John t whelan ranking simulator

    It clearly doesn't make sense to lose a game in order to play on the road. I was proposing that the team with the home ice advantage have the option to play on the road. That could clearly be in their advantage. Take a case of 5-12 matchup. That could easily be 65 percent at home and 60 percent on the road. Now the numbers work out.

    Leave a comment:


  • Patman
    replied
    Originally posted by goblue78 View Post
    My point is that a team on the bubble may need to take a risk. It could be that 2-0 in your rink just isn't worth very much and 2-0 in the other guy's rink is worth enough to make the risk worth it. Take the obvious example where a win over someone in your rink would be dropped from RPI while a win in their rink would augment RPI. In that case, playing at home is all downside. (And in the ECAC for example, the first round is 5 vs. 12. A really bad #12 team would be exactly the team you want to play in their rink, and the #5 ECAC team is probably real close to the bubble.)
    I don't think math is going to be your friend.

    Tell you what, take a home win prob of .55 under the two scenarios...

    This is a short sim but I'm headed to a birthday party in dupont. Seeing as the denominator scales with the numerator I'd find it hard to believe there would be an advantage.

    If you disagree w .55, sure, modify. But I'm almost sure of myself that your life is better playing for the first win and the home series than plopping in a loss

    Just surface, N-H-H 3 wins

    +2.6/+2.6

    Vs n a a 2-1

    +2.4/+3.4...

    Don't even get back the same count of adjusted wins. There can be a lot of odd interplay when you add the bonuses... But first principles... Smaller numerator, larger denominator implies a smaller value... And this your base should you add tweaks... And you are likely to lose away... Christ that was a quick trip.
    Last edited by Patman; 11-30-2013, 07:56 PM.

    Leave a comment:


  • goblue78
    replied
    Re: John t whelan ranking simulator

    My point is that a team on the bubble may need to take a risk. It could be that 2-0 in your rink just isn't worth very much and 2-0 in the other guy's rink is worth enough to make the risk worth it. Take the obvious example where a win over someone in your rink would be dropped from RPI while a win in their rink would augment RPI. In that case, playing at home is all downside. (And in the ECAC for example, the first round is 5 vs. 12. A really bad #12 team would be exactly the team you want to play in their rink, and the #5 ECAC team is probably real close to the bubble.)

    Leave a comment:


  • Patman
    replied
    Originally posted by FlagDUDE08 View Post
    The denominator is dependent upon who wins the game. If the home team, it's 0.8 for both. If the away team, it's 1.2 for both. Of course for the non-school-rink situations, because it is always the league that hosts and not a specific school, those are worth 1.0.
    So that definately mutes any strategic advantage. Not that weird things can't happen. Personally, I'd love to lose one and win two all year long as that'll probably get you in

    Leave a comment:


  • FlagDUDE08
    replied
    Re: John t whelan ranking simulator

    Originally posted by Patman View Post
    is the denominator always the same?

    If so, in short, yes, but that perverse incentive will be there for any team. What isn't sure is it worth losing to get into such a situation. That'd be a risky move. I don't think I'd take that risk... could just go 2-2 or worse than 2-1.
    The denominator is dependent upon who wins the game. If the home team, it's 0.8 for both. If the away team, it's 1.2 for both. Of course for the non-school-rink situations, because it is always the league that hosts and not a specific school, those are worth 1.0.

    Leave a comment:


  • FlagDUDE08
    replied
    Re: John t whelan ranking simulator

    November 29th's games are over, and here's how things look:

    Quality Wins rate:

    5.00 Minnesota
    4.75 Ferris State
    4.50 Michigan
    4.25 St. Cloud State
    4.00 Providence
    3.75 Quinnipiac
    3.50 Cornell
    3.25 LSSU
    3.00 Clarkson
    2.75 Notre Dame
    2.50 Minnesota Duluth
    2.25 Boston College
    2.00 Yale
    1.75 Minnesota State Mankato
    1.50 UMASS Lowell
    1.25 New Hampshire
    1.00 Miami
    0.75 Bowling Green
    0.50 Wisconsin
    0.25 Union

    And the tournament field:

    Minnesota
    Michigan
    Ferris State
    Providence

    St. Cloud State
    Quinnipiac
    Cornell
    LSSU

    Notre Dame
    Minnesota Duluth
    UMASS Lowell
    Clarkson

    Yale
    Boston College
    Minnesota State Mankato
    AHA Champ (39 - Mercyhurst)

    A new version of the games list is uploaded today to reflect the tournament, as that deletes a couple of redundant games.

    Leave a comment:


  • Patman
    replied
    Re: John t whelan ranking simulator

    Originally posted by goblue78 View Post
    Just thought of another issue with the new RPI calculation. In the conference playoffs, couldn't there be a clear advantage for a higher ranked team to play on the road? The difference between a .8 multiplier and a 1.2 multiplier could easily be the difference for a team on the bubble. Shouldn't a higher ranked team have the option to opt to play a two-out-of-three playoff series on the road? If not, why not? Note that the home advantage (as traditionally estimated) is nowhere near the RPI differential, so, other than the fact that you lose the gate receipts, shouldn't it be the normal call if you're on the bubble?
    is the denominator always the same?

    If so, in short, yes, but that perverse incentive will be there for any team. What isn't sure is it worth losing to get into such a situation. That'd be a risky move. I don't think I'd take that risk... could just go 2-2 or worse than 2-1.

    Leave a comment:


  • goblue78
    replied
    Re: John t whelan ranking simulator

    Just thought of another issue with the new RPI calculation. In the conference playoffs, couldn't there be a clear advantage for a higher ranked team to play on the road? The difference between a .8 multiplier and a 1.2 multiplier could easily be the difference for a team on the bubble. Shouldn't a higher ranked team have the option to opt to play a two-out-of-three playoff series on the road? If not, why not? Note that the home advantage (as traditionally estimated) is nowhere near the RPI differential, so, other than the fact that you lose the gate receipts, shouldn't it be the normal call if you're on the bubble?

    Leave a comment:


  • FlagDUDE08
    replied
    Re: John t whelan ranking simulator

    Originally posted by LTsatch
    see above
    That's how the ratings work.

    Leave a comment:


  • FlagDUDE08
    replied
    Re: John t whelan ranking simulator

    As of games ended 27 November 2013:

    5.00 Minnesota
    4.75 Ferris State
    4.50 Michigan
    4.25 St. Cloud State
    4.00 Providence
    3.75 Quinnipiac
    3.50 Cornell
    3.25 Boston College
    3.00 LSSU
    2.75 Clarkson
    2.50 Minnesota Duluth
    2.25 Minnesota State Mankato
    2.00 Yale
    1.75 Notre Dame
    1.50 UMASS Lowell
    1.25 Wisconsin
    1.00 Bowling Green
    0.75 Miami
    0.50 Union
    0.25 New Hampshire

    And the tournament field:

    Minnesota
    Michigan
    Ferris State
    Providence

    St. Cloud State
    Quinnipiac
    Cornell
    Boston College

    LSSU
    Minnesota Duluth
    Clarkson
    Yale

    UMASS Lowell
    Notre Dame
    Wisconsin
    AHA Champ (38 - Air Force)

    Leave a comment:


  • FlagDUDE08
    replied
    Re: John t whelan ranking simulator

    As of games ended 26 November 2013:

    5.00 Minnesota
    4.75 Ferris State
    4.50 Michigan
    4.25 St. Cloud State
    4.00 Providence
    3.75 Quinnipiac
    3.50 Cornell
    3.25 Boston College
    3.00 LSSU
    2.75 Clarkson
    2.50 Minnesota Duluth
    2.25 Notre Dame
    2.00 Yale
    1.75 Minnesota State Mankato
    1.50 Wisconsin
    1.25 Bowling Green
    1.00 UMASS Lowell
    0.75 Miami
    0.50 Union
    0.25 Northern Michigan

    And the tournament field:

    Minnesota
    Michigan
    Ferris State
    St. Cloud State

    Providence
    Quinnipiac
    Cornell
    Boston College

    LSSU
    Minnesota Duluth
    Clarkson
    Notre Dame

    Yale
    UMASS Lowell
    Wisconsin
    AHA Champ (38 - Air Force)

    Leave a comment:


  • Steve_MN
    replied
    Re: John t whelan ranking simulator

    Really good discussion for the numbers geeks among us. Keep up the good work, guys.

    Leave a comment:


  • goblue78
    replied
    Re: John t whelan ranking simulator

    I can guarantee the variance of the ratings will increase, since counting games at either .8 or 1.2 (with roughly equal probabilities) is guaranteed to increase variance. On the other hand, I can also guarantee that, jsut as in every other year, variance will decrease as the year winds down, because it will still be the case that the 20th game has, on average, only about 1/20th the weight of the previous games. Add to that the removal of the TUC cliff and variance at the end of the season will almost surely be lower. That doesn't mean, of course, that it might not still be the case that in a particular circumstance a team might not drop six places with a loss if they are particularly narrowly perched above the five teams below them, even at the end of the season. But there is nothing in the new system that won't stabilize the results overall at the end of the season.

    Leave a comment:

Working...
X