Great post, thanks for the examples. I'll take a shot at answering, but can't say I'll be 100% on this. Sorry for really long post.
#1, remember my basic RPI comment, which is that RPI is just math. It doesn't have an agenda or a bias, it's just math. The formula is (Win%*0.25)+(OppWin%*0.5)+OppOppWin%*0.25). The win% aren't just the standard (in our case) 24/38=0.632, but rather home wins are discounted by a 0.7 multiplier and road wins are inflated by a 1.3 multiplier. Mathematically this causes problems because a win by the home team results in a combined record for the two teams of 0.7 wins and 0.7 losses = 1.4 games played while a win by the road team is 1.3 wins +1.3 losses = 2.6 games played. If at the end of the season the win% of the home team is not exactly .500 then the total number of apparent games played at the end of the season will not equal the total number of games actually played. That is my main complaint about the formula.
To your question about what matters more, the answer is #2, #3, and opponents win %. #1 is totally irrelevant, RPI isn't aware conferences even exist. #4 is an output from the RPI formula, not an input.
To your examples....
San Diego: They have a better basic record than us (26-12 v 24-14), but more importantly they have played 24 of 38 games on the road. And they have done well in those 24 games, so their adjusted record is a really good 27.5-10.2. In essence, take 1 or 2 of their losses and make them wins just because they have played a bunch of true road games. Our adjusted record is 21.0-13.7, or in other words take 3 of our wins at random and pretend they never happened as a penalty for us playing over 60% of our games at home.
Notre Dame: As San Diego does, they take advantage of the road win inflation, also playing 24 of their games on the road. They haven't been quite as successful, but that does adjust their record to 20.0 - 15.0, so randomly toss out 3 of their losses. Our adjusted record is still a game better than theirs. Of the top 7 teams in the ACC they have played 5 (going 0-15), whereas we have played 4, so I have to assume that this paired with playing most of their non-conference games on the road help them in the calculation.
Auburn: Just what you said, play a really tough schedule and get credited for at least winning a couple of them. They have played RPI #1, #2, #3, and #6, posting a 2-10 record in those games. I agree with you that there shouldn't be any such thing as a good loss, but since 75% of the formula is not about you but rather your opponent, they are taking advantage of this. Over time and enough games this should even out, but a 60 game college schedule may not be enough for that to happen. Still, to my knowledge no SEC team has made the NCAA tournament without making the SEC tournament, so I think they will be left out unless they can rally in the next month.
Ohio St: See the explanation above about road win inflation, their adjusted record is 19.6-15.2, basically the same as Notre Dame. They also managed to avoid scheduling the REALLY, REALLY BAD teams. They only have 1 game against sub-200 RPI schools (Cal St - Fullerton!!!!) and only 2 other games against sub-150 RPI teams. IOW, nearly all their games have been against schools respected by the RPI. By comparison, we have played 8 games against sub-200 schools and 3 more against sub-150 RPI schools.
Kennesaw State: They also take advantage of the road win inflation. Their adjusted record is 23.2-13.6, which is 2 full wins better than ours. Plus they took a series @ Kentucky. That is probably the single biggest series win by any school in the country this year.
If we go 6-6 the rest of the way, yes, I expect our RPI will rise unless we couple that with losing all our mid-week games and get swept in the ACC tourney.
My advice to Coach Hall for next season would be to #1, avoid scheduling 3 schools the caliber of Youngstown St, Radford, and Cornell. One is probably OK, but not three. #2, look for a way to schedule one of the pre-conference weekend series on the road. I'm not sure exactly how to do this, and it may not work financially, but what if we played an early season series at Troy or App State or Western Carolina. Schools we should be able to beat, but that would not kill our RPI and would give us some road inflated wins.
