obviously the system is assuming that maryland will get 14 wins. if they do get it, then they should be a lock, with the #1 SOS and 10 wins against top 50 rpi. 

Posted by isack24 on 12/12/2011 2:33:00 PM (view original):

Posted by coach_billyg on 12/12/2011 2:19:00 PM (view original):

Posted by cburton23 on 12/11/2011 9:20:00 PM (view original):
Then I have a problem with the real life committee too. If you beat somebody twice then by definition you are better than they are

trying to describe the world in these absolute black and whites will never get you anywhere

Exactly, and that's been my problem with cburton's position.  I think he probably has an argument in this particular circumstance, even if the majority disagrees.  But there's no way that you can say 2-0 makes you the better team every time, without looking to anything else. 

I think my Bears/Packers example pretty clearly proves that, and it's frustrating that someone can look at that and say, "I guess the Bears must be the better team." 

I do think its arguable which should be first in this case. . however the principle he is espousing is not workable on its face. . and its not as if, even if correct, his argument is a slam dunk.  Hard cases make bad law.

You have to remember this game is based on numbers, if you get so much credit for beating a top 50 rpi team an half that amount of credit for beating 51-100 rpi teams then 2 more wins vs top 50 makes a heck of a difference. At the same time head to head only proves you match up better against that team than they match up with you. It doesn't prove you match up better against everyone to determine that you must look at the entire body of work.

Posted by tianyi7886 on 12/12/2011 4:14:00 PM (view original):
obviously the system is assuming that maryland will get 14 wins. if they do get it, then they should be a lock, with the #1 SOS and 10 wins against top 50 rpi. 

Obviously...hence the flaw in calling someone a lock when they may or may not even be eligible. locks should be teams that really are in fact locks, no? you dont call a team national champs just because they make the final four

Posted by bow2dacowz on 12/12/2011 7:02:00 PM (view original):

Posted by tianyi7886 on 12/12/2011 4:14:00 PM (view original):
obviously the system is assuming that maryland will get 14 wins. if they do get it, then they should be a lock, with the #1 SOS and 10 wins against top 50 rpi. 

Obviously...hence the flaw in calling someone a lock when they may or may not even be eligible. locks should be teams that really are in fact locks, no? you dont call a team national champs just because they make the final four

obviously the coding logic isn't taking into the account of the 14 win requirement, but does it really matter? unless this teams makes the NT without winning the 14th game, who cares? 

Nevermind dude. i already stated that it was semantics and not a big deal...i just think its dumb. i am still allowed an opinion right?

Sure, but why bring up these minor issues w/o any impact on the actual game or seeding determination when there are much larger issues at hand. D3 tark has shown that losing to top 50 teams significantly boosts NT seeding, which is simply wrong. 

Or that absolutely mauling a bad team in conference that you have no choice but to play can significantly HURT your seed and tournament chances?   One of the two biggest problems in real life I have with RPI:  A 20 point loss to a good team could boost your prospects significantly while a fifty point win over a bad team hurts them significantly.
 

 

Posted by a_in_the_b on 12/13/2011 10:32:00 AM (view original):

Or that absolutely mauling a bad team in conference that you have no choice but to play can significantly HURT your seed and tournament chances?   One of the two biggest problems in real life I have with RPI:  A 20 point loss to a good team could boost your prospects significantly while a fifty point win over a bad team hurts them significantly.
 

 

But that wasn't the case in tark. Compare these 4 teams:

Drew: 28-1, 5 rpi, 57 SOS, 4-0 against rpi 1-50, 9-0 against rpi 50-100, 15-1 against rpi 100+ CT champion
Greensboro: is 23-6, 11 rpi, 7 SOS. 5-6 against rpi 1-50, 7-0 against rpi 50-100, lost in CT championship game
Lynchberg: 22-6, 8 rpi, 6 SOS, 6-5 against rpi 1-50, 7-0 against rpi 50-100, lost in CT semis
Hamline: 28-1, (forgot the exact rpi, but I think it was 7), 7-1 against rpi 1-50, 5-0 against rpi 50-100, 16-0 against rpi 100+

coolman broke it down further, with rpi 1-25, 25-50, with Greensboro being 5-6 against rpi 1-25, while hamline is 4-1 against rpi 1-25. Drew is 2-0 against rpi 1-25.

Hamline has the strongest resume among the four, and it's not even close, but Greensboro and Lynchberg got the #2 seed because they lost more games against highly ranked opponents?!?

Please note that Hamline has higher rpi than both team.

Clearly the new seeding is valuing losses against good teams by quite a bit. 

Posted by tianyi7886 on 12/13/2011 11:31:00 AM (view original):

Posted by a_in_the_b on 12/13/2011 10:32:00 AM (view original):

Or that absolutely mauling a bad team in conference that you have no choice but to play can significantly HURT your seed and tournament chances?   One of the two biggest problems in real life I have with RPI:  A 20 point loss to a good team could boost your prospects significantly while a fifty point win over a bad team hurts them significantly.
 

 

But that wasn't the case in tark. Compare these 4 teams:

Drew: 28-1, 5 rpi, 57 SOS, 4-0 against rpi 1-50, 9-0 against rpi 50-100, 15-1 against rpi 100+ CT champion
Greensboro: is 23-6, 11 rpi, 7 SOS. 5-6 against rpi 1-50, 7-0 against rpi 50-100, lost in CT championship game
Lynchberg: 22-6, 8 rpi, 6 SOS, 6-5 against rpi 1-50, 7-0 against rpi 50-100, lost in CT semis
Hamline: 28-1, (forgot the exact rpi, but I think it was 7), 7-1 against rpi 1-50, 5-0 against rpi 50-100, 16-0 against rpi 100+

coolman broke it down further, with rpi 1-25, 25-50, with Greensboro being 5-6 against rpi 1-25, while hamline is 4-1 against rpi 1-25. Drew is 2-0 against rpi 1-25.

Hamline has the strongest resume among the four, and it's not even close, but Greensboro and Lynchberg got the #2 seed because they lost more games against highly ranked opponents?!?

Please note that Hamline has higher rpi than both team.

Clearly the new seeding is valuing losses against good teams by quite a bit. 

From what I can garner from what you just said. . I don't think I am actually arguing with you.  My main additional point is how much you get hurt by just PLAYING, win or lose, a 300 RPI team. . even if you can't avoid it.  Heaven help you if your division in your conference acquires a ghost ship and you have to play them twice.

I thought your point was that playing the 300 rpi team hurts your rpi, which gives you a lower seed, while losing to a top rpi team boosts your SOS, and thus, your rpi, helping you with a better seed.

The example I provided above shows that the 300 rpi team hurting your rpi isn't the problem right now, the problem is that the value of playing a top 50 rpi team (even losing to one) is too great. Losing to an rpi 1-50 team now boosts your seeding.

In the above example, before the CT championship game was simmed, I was sitting at #5 in the projection report, with Greensboro a few spots below. After I won the CT game and Greensboro losing it, I get dropped to #11 while Greensboro overtakes me. This makes it very clear that losing to a top rpi team now carries huge value. 

Posted by tianyi7886 on 12/13/2011 2:53:00 PM (view original):
I thought your point was that playing the 300 rpi team hurts your rpi, which gives you a lower seed, while losing to a top rpi team boosts your SOS, and thus, your rpi, helping you with a better seed.

The example I provided above shows that the 300 rpi team hurting your rpi isn't the problem right now, the problem is that the value of playing a top 50 rpi team (even losing to one) is too great. Losing to an rpi 1-50 team now boosts your seeding.

In the above example, before the CT championship game was simmed, I was sitting at #5 in the projection report, with Greensboro a few spots below. After I won the CT game and Greensboro losing it, I get dropped to #11 while Greensboro overtakes me. This makes it very clear that losing to a top rpi team now carries huge value. 

Agreed, games vs top 50 win or loose seem to be magnified.

Basicaly what I would like to see is a simple modifier on RPI, based on who you play (not changing actual RPI, I'm talking about seeding).  Right now, RPI's flaws are rooted in (1) a failure to properly distinguish who you are beating/losing to, and (2) arbitrarily-assigned home/road modifiers that have no real correlative value in HD.

I think if you took RPI, reduced the home/road modifier and added a sliding-scale modifier for the actual team you played against (not just based on W/L record) in which wins against top-50 or so teams counted more and losses against 100+ RPI teams counted more, I think someone (someone smarter than me, anyway), could come up with a pretty simple formula for seeding that would be a nice compromise between RPI and what we have now.

Posted by tianyi7886 on 12/13/2011 2:53:00 PM (view original):
I thought your point was that playing the 300 rpi team hurts your rpi, which gives you a lower seed, while losing to a top rpi team boosts your SOS, and thus, your rpi, helping you with a better seed.

The example I provided above shows that the 300 rpi team hurting your rpi isn't the problem right now, the problem is that the value of playing a top 50 rpi team (even losing to one) is too great. Losing to an rpi 1-50 team now boosts your seeding.

In the above example, before the CT championship game was simmed, I was sitting at #5 in the projection report, with Greensboro a few spots below. After I won the CT game and Greensboro losing it, I get dropped to #11 while Greensboro overtakes me. This makes it very clear that losing to a top rpi team now carries huge value. 

Agree with that part.  Just think both are a problem,