RPI anomalies, a.k.a. a look beyond wins and losses
February 10, 2008 1:47 pm by Dan'l BIf bracketologists were marine biologists, poll voters would have their assistants report on goldfish in bowls, RPI enthusiasts would hang out at Sea World, and Sagarin and Pomeroy would only be found by the Coast Guard. I’ve wasted time looking at the polls, and I guess I’ll continue to do so here and there. The RPI is definitely better, but it does a very good job at telling us nothing special. There are two animals that the RPI simply can’t handle: the ridiculously good record against moderate or worse competition and the moderate or worse record against ridiculously good competition.
Naturally, the great records and mediocre schedules come out of the secondary conferences. Drake, Butler, and St. Mary’s, I’m talking about you. Not only is your RPI inflated, you’re shooting up the polls too. “Wait a minute! Give those guys a break, they’re screwed by the BCS bullies!” Well, I think the RPI and polls actually do these schools the bigger injustice. Name the non-BCS schools that have gotten a 5th seed or better in the last decade and done something with it. That list is awfully short. These teams are good, but they don’t belong in the top 25. There’s nothing wrong with being one of the top 40.
And of course there’s the “Izzo-Chaney-Olsen Plan,” in which you schedule half a dozen elite non-conference games before conference play starts. No surprise, it’s Olsen this year with the #8 RPI and #1 schedule. UConn, yes UConn! is a bit overrated by the same method.
What the RPI fails to do is recognize quality basketball and pick out would-be sleepers (if they get to dance at all). Kansas State will obviously dance, and I expect to see them in the Elite Eight. Wisconsin’s very quietly 19-4 and a lot better than their #20 RPI. But the one I’ll be watching will be West Virginia. They’ve got a #48 RPI, a 12-seed from Joe Lunardi, and a resume full of quality play all year. Barring a disaster the rest of the way, they’re going to slip into the tournament in the #8 to #12 range and will have a great shot at the Sweet Sixteen as “underdogs.”
Categories: Rankings, Dan'l B, Connecticut, West Virginia
4 Responses to “RPI anomalies, a.k.a. a look beyond wins and losses”
Besides the win against Marquette, does West Virginia really have a big win? I agree they’ve had a crap load of tough losses, but if they were really that good wouldn’t half (or even 1/3) of those close games be wins?
Just playing devil’s advocate.
The point is to take [i]a look beyond wins and losses[/i]. WVU is 1-4 in close games, easily within random chance and all of which could have gone either way. There isn’t a demonstrable pattern between prior records in close games and future results. That’s what stands out as the difference between these schools (seeds are from ESPN bracketology):
Flip just one of those coins that WVU lost and these markers immediately jump ahead of the others (Marquette and Clemson).
Good point — you’re right. 1-4 is well within random chance.
I think you’re definitely right about Kansas State, but I haven’t seen them that much. On the other hand, I’ve seen West Virginia and never came away impressed.
Well put. It’s hard to tell what the “best” ratings system is. That’s why I just listen to Jay Bilas.
Care to comment?