Posted by coach_billyg on 12/30/2011 4:24:00 AM (view original):

Posted by isack24 on 12/29/2011 11:29:00 PM (view original):
"All things equal in conference play, the guy going 5-5 with a 15 SOS hurts all 11 of his conference mates more than he is hurt by the guy who goes 10-0."

You can't say that in a vacuum.  I think you're being way too broad.  If someone has a 300 SOS, it's because the teams they play are really bad OOC.  If a team has an SOS of 15, it's because the teams they play have won a lot of games.  

Yes, opponents' opponents only counts for 25% of RPI, but when you're talking about potentially 30 games under .500 by the end of the years versus 30 games over .500, that 60-game swing will be huge, even if only accounting for 25%.  

I don't disagree with your premise or assertion that people simply don't understand RPI, but I think your extreme examples are probably wrong or, at the very least, incorrect as a general rule.

you can say that in a vacuum.

what rails said is correct - going 5-5 is worse than 10-0, no matter what. the best case (impossible) for the 5-5 team is their sos is 1.00, meaning every team they played never lost and all their opponents never lost either - where the 10-0 team played only 0-10 teams who all played 0-10 teams. when you play the 5-5 guy, you get 2/3rds of the value from the 5-5, and 1/3rds from their opponents, .100. this sums to .5 * 2/3 + 1 * 1/3 or .667.

when you play the 10-0 team, you get 2/3rds from their 10-0 and 1/3rd from their opponent's .00 winning percentage. thats 1.00 * 2/3 + 0.00 * 1/3, or .667. this means that in NO scenario is it possible for the 5-5 team to be better for their conference mates.

realistically, the best kind of SOS you see is .7, the worst being about .3. then, the 5-5 team would contribute only .566! the 10-0 team would contribute a whopping and vastly superior .766!

Yeah, that's the best way for you.  The conversation we were having was about what is better for your conference mates.  8-2 or 7-3 (or even potentially 5-5) against a difficult schedule is still better for YOU than 10-0 against a weak one.

EDIT: I should explain further.  Because your own W/L is only 25% of your RPI, your opponents' W/L (50%) is worth twice as much.  So scheduling games against bad teams OOC WILL significantly affect your own RPI.  It's the fundamental problem with RPI, in my opinion, and why people who think seeding should be strictly tied to it are wrong. 

Posted by bhansalid00 on 12/30/2011 10:07:00 AM (view original):
Interesting. So the rule of thumb I was told by a veteran coach many, many seasons ago was to schedule the best OOC opponents you can that you expect to beat, and make as many of those road games as possible. Is that still a valid approach?

Posted by tianyi7886 on 12/30/2011 12:32:00 AM (view original):
I think the guy that schedules weak to a 10-0 is apologizing because he could get to 18-20 wins and still not make the NT, thus  hurting his conf in terms of recruiting cash.