Posted by asher413 on 12/4/2011 3:19:00 PM (view original):
Must've been 10 days since the last anti-ACC/Allen thread.

Posted by jtt8355 on 12/5/2011 10:30:00 AM (view original):
if the acc is such a problem, why don't the other coaches just boycott scheduling them? if they are only able to schedule against sims, their collective rpis should drop, causing less appearances in the post season, lessening the damage they do in recruiting...

Posted by cheeznsweet on 12/5/2011 1:23:00 PM (view original):
jtt must be an acc plant...

haha - def not! nothing to do in d1 in either world i'm in. just trying to think outside the box to address a repetitive forum complaint. i guess i don't understand rpi well enough, but i thought that opponents' rpis get weighted into the calculations, so it wouldn't matter that you have 12 10-0 teams playing in conference...if their cumulative non-con opponents' rpis average 150, the acc rpi would hover in the high (60-80) double figure range. any math or rpi whizzes know where my thinking is off?

I was just messing with you, jtt.

Opponents' RPI doesn't get weighted.  Their W/L and SOS get weighted in. 

RPI is three things:

1) 25% your winning %
2) 50% your opponents' winning %
3) 25% your opponents' opponents' winning %


Take an example Team A and let's look at each of the 3, above. 

1) Let's say Team A goes 10-0 in ooc and split 8-8 in conference.  That gives them an 18-8 record or a .6923 winning percentage for calculating RPI.  So 25% of their RPI is locked in at .6923 which is a value of .1730.

2) Let's also say their conference goes 120-0 in ooc.  Then, because conference records are a zero sum game, the total conference has to have 96 wins and 96 losses and therefore the conference portion of Team A's opponents' winning % is 0.6923 also.  Since the conference portion of your opponents' winning % is 16/26ths of the season, and 50% of the total RPI, that contributes .2130 to Team A's RPI

3) Now we're to opponents' opponents' winning %.  And since every team in that conference is playing the rest of the conference, they all get/provide the same benefit to its opponents' opponents' winning % which is (16/26ths * .6923) and receives a 25% weighting.  That adds in .1065.

Add up 1, 2, and 3 and you get:  .1730 + .2130 + .1065 = .4925

In most seasons, in most worlds, that alone would be good for somewhere between 8th and 11th best conference or an individual RPI of 150-170.  But we're still completely ignoring the out of conference winning % and opponents' opponents' out of conference winning percentage.  In essence, you'd still have this RPI even if every team that Team A and their conferencemates played out of conference went 0-26 - The entire conference would still have an RPI of .4925. 

The part we've ignored (the ooc opponents' winning % and opponents' opponents' winning %) has a total potential value of .2885.  So let's throw some numbers in and see what happens.  Let's assume that all of Team A's and their conference's ooc opponents get only 5 wins all year.  Let's also assume that the ooc opponents also play crap schedules and the ooc opponents' opponents also only get 5 wins.  That right there is good enough to bump Team A up by .056 RPI points.  That would put them at .5480 which is generally good for about 90th overall.

But If the ooc teams and their opponents average measly 10-16 records, that would put Team A and that entire conference at an RPI of .6035 which is about a Top 40 RPI.

In practice what happens is that conferences that do this try to target SIM teams for their ooc that can go about .500.  What often happens is that the best teams understand that this occurs and you'll see half or more of a good SIM ooc target have other very good teams on their schedule.  This works out very well since it helps the target SIMs opponents' winning percentage...

Complex but a useful way to think about RPI and see just how important all the little OOC scheduling decisions are.