m-v-c wrote:rlh04d wrote:Using RPI to counter better metrics isn't a good argument -- it's pretty apparent advanced metrics are a better measure of a team than RPI.
That's an opinion, not a fact.
Every formula has its outliers. Last year Ken Pomeroy's ratings claimed Pittsburgh was the 11th-best team in the country. Denver was in the top 40-45 teams before the NCAA Tournament. Pretty sure no one would reasonably group either of those teams in those categories last year.
Every metric formula has its own biases, Pomeroy, Sagarin, the BPI-all of them. The RPI is the simplest and most neutral of them all. It measures 1) are you winning games, 2) who did you beat and 3) how good are they. And then it gives a bonus for winning on the road, which makes perfect sense in college basketball since home teams win 70% of the time. Don't need anything more than that. Margin of victory should not matter, there is too much that happens at the end of a CBB game to influence it (a 1-point game with a minute to play can become a 10-point loss because of free throws, while a 40-point lead can become 20-point win because a team puts all its reserves in).
Well, no, it's not an opinion. Many statistical comparisons have been made between the different measures, and RPI has consistently been one of the weakest statistical measures. We are able to both have different opinions, certainly, but when my opinion is overwhelmingly backed up by statistical evidence, that is now leaving "opinion" terrain and approaching "obvious fact."
Which isn't specifically about KenPom rankings. I think Massey rankings are better, and LRMC better than that. Again, these are "opinions" backed up by statistical evidence. Here's some:
http://www2.isye.gatech.edu/~jsokol/ncaa.pdf
http://www.sloansportsconference.com/wp ... stract.pdf
The flaws with the RPI are numerous and well documented, most popularly the lack of margin of victory, which ranks one point wins just like they were 50 point wins, to keep basketball warm and fluffy and nice to everyone. I'm not going to sit here and re-state what people far smarter than me have already run off against the RPI. I'll just give more links:
http://www.slate.com/articles/sports/sp ... dness.html
http://www.behindthescoreboard.com/2009 ... e-why.html
http://www.basketballprospectus.com/unfiltered/?p=632
http://sports.vaporia.com/bb-fwin.html
http://sports.vaporia.com/pool.html
You're really going to use Pittsburgh at #11 to discredit KenPom rankings? Then what does that make the RPI, which had New Mexico at #2 overall? Didn't that team lose to a 14 seed in the first round? So KenPom's rankings are ridiculous because Pittsburgh lost to a Final Four team, but we should use the RPI, which gave us the #2 overall team in the country losing to an Ivy League school that lost by 23 to a 6 seed in the second game? Please.
Sorry, you don't get to just choose outliers and say everything's just a matter of opinion. The math clearly says there are superior and inferior rankings. Every measure isn't created equal, and just because something is the "simplest" certainly doesn't mean it's the best. It's just not.
Over, and over, and over again the RPI has been proven to be a less accurate predictor of future performance than other statistical measures. It is not as good. That is not an opinion. There's no reason to just bury your head in the sand and pretend like the facts are anything different. We have the numbers to compare the different measures.
