Saturday, March 04, 2006

My eerily detailed thoughts about the Missouri Valley's RPI

You've just got to be really clever about scheduling your pansies. Make sure that you schedule the 1st or 2nd place finisher out of the Southland conference or the Big Sky instead of #7 or 8.Ok, so generally, (leaving out the home-road distinction for the sake of simplicity) RPI= 1. Your win pct (25%), 2. Your opponents' win pct (50%), and 3. your opponents' opponents' win pct (25%).

Here's Mizzou St.'s nonconference opponents for example
N. Illinois RPI 126---(15-10; 1st in Mid-American Conf.-West)
Ark. St (twice) RPI 224--(10-18; 3rd Sun Belt West)
Ga. Southern RPI 161--(18-9; 1st in Southern Conf.-South)
T A&M C-C--RPI 202--(16-8; Independent)
Oral Roberts--RPI 131--(17-11; Mid Con Conf.T-1st)
Detroit--RPI 180--(14-16; T 3rd Horizon Lg)Arkansas (L) RPI 50--(20-8; 3rd, SEC West)Wisc-Milwaukee--RPI 58 (18-8; 1st Horizon Lg)___________________________________

For a nonconference RPI of #6 in the country. Yet that's all the teams they played. It's based purely on record. They went (8-1) against this schedule, so that's 25%. and then if you add up the records, then that's (128-88) a 60% win pct., which is 50% of the RPI formula.

Let's compare that to a team that did a poor job scheduling their "pansies".

How about Kansas (RPI 41)?
Idaho St. RPI 201--(11-14; T-last in Big Sky)
Arizona (L) RPI 18--(18-10; 4th in PAC-10)
Arkansas (L) RPI 50--(20-8; 3rd, SEC West)
Nevada (L) RPI 27--(22-5; 1st WAC)
W. Illinois RPI 321--(5-20; 8th Mid-Con Conf.)
St. Joseph's (L)RPI 46--(14-12; 5th Atlantic-10)
California RPI 61--(17-9; 3rd Pac-10)
Pepperdine RPI 242 (7-20; last West Coast Conf.)
N. Colorado RPI 327 (3-24; Independent)
N. Orleans RPI 254 (8-18; 5th, Sun Belt West)
Yale RPI 190 (15-13; T-3rd Ivy League)
Kentucky RPI 31 (19-10; T-2nd SEC East)

For a nonconference RPI of #138 in the country. They went 8-4 against this slate for the first 25%. Then adding up thier records, you get (159-163) a 49.3% win pct. for the second 50%. Scheduling a team like N. Colorado (3-24) in this regard kills this stat. If they had just replaced them with Hofstra (22-5) for example, they would be at (178-144) and probably at least be in the top 100 in nonconference RPI. Additionally, for the RPI, scheduling a team like Kentucky at 19-10 gets you the nearly same benefit in the first 2 categories as Tx A& M Corpus Christi, and as long as that team didn't schedule too poorly, then the 3rd category shouldn't hurt you too badly. Not to mention you don't take the risk of losing and hurting the first category (which offsets the 3rd a bit)

So, from what I can tell, If you do a good job of scheduling the right pansies, you can make out like a bandit.

The main thing is don't schedule the big time losers like N. Colo, W. Illinois, and Pepperdine (that one was tough to predict).


  1. That was indeed "eerily detailed," SJ.

    There's something not quite right about the RPI: your ranking is based on the rankings of your opponents, which are based on the rankings of THEIR opponents, which ... Where does it START? It just seems like an endless feedback loop to me.

  2. It's not reflexive as you suggest.

    That's an often-charged criticism of the BCS formula in football, but the RPI is pure, mathematically speaking. It starts when your first game starts. So, when you play your first game, the formula goes (1-0)your record 25% (0-1)your opponents' record 50%, and (0-0) your opponents' opponents' record (assuming neither team has played anyone else. I'm just suggesting that it's flawed as a methodology of finding the strongest teams for the reasons I outline.