RPI is meaningless - ours dropped this week!

[seaweed]

Otherwise how do you explain our RPI DROPPING from last week to this week despite the two major conference wins? Take care of business on the road against Cincy and then beat a top 10 (RPI #6) team at home in one week and DROP in the ratings? Somehow, despite another dominating week , we go down from 16 to 17 in RPI?

If this is not final proof that RPI is meaningless, then nothing will ever convince the kenpom sycophants. Try all you want to explain how this is 'statistically consistent' or whatever other defense you care to make, I don't care - it is B*** S***.

I do not believe that my failure to understand this phenomenon is at all tied to my avoiding Stats, Macro or Micro Econ while at GU.

(couldn't wait for official updated NCAA, used thier officials rankings from last week and the latest up to the minute from the link up top of hoyasaxa.com (kenpom)

[FOTP]

Actually it's not too complicated. We get kille because 3 out of our 5 loses were at home. The RPI count home loses (no matter who it's against...which is insane) as 1.4 losses. Road losses (again against anyone) only count as .6 losses.

This is part of the reason why teams liek ND and Nova have huge RPIs...the system is set up for really good home teams.

The RPI is good directionally, but you really can't say one teams being a #14 is really worse than a #7. See KY so high up. We'll never be seeded behind them, but they play a great schedule.

Winning against Cincy actually hurts because they have a terrible RPI. WInning at home against Pitt was kind of expected so it doesn't help as much as one would think.

Don't worry too much about the RPI. The committee will see through the basic problem.

[MCIGuy]

I know maryland was one spot behing GU going into the UNC game (more proof of how useless the RPI can be at indicating the strength of teams) and I'm sure the Terps leapfrogged the Hoyas after last night's collapse by the Tar Heels.

[VictorSkyPage]

I don't think its meaningless.. its just a supplementary stat -- helps indicate the strength of the team.. not an exact science by any means, but not meaningless... can't get too caught up

ken pom is the man, his rankings are what I would say are the best indicator of team strength (and Parrish's rankings aren't bad either, simply because I usually agree with them)

[Cambridge]

seaweed said:

Otherwise how do you explain our RPI DROPPING from last week to this week despite the two major conference wins? Take care of business on the road against Cincy and then beat a top 10 (RPI #6) team at home in one week and DROP in the ratings? Somehow, despite another dominating week , we go down from 16 to 17 in RPI?

If this is not final proof that RPI is meaningless, then nothing will ever convince the kenpom sycophants. Try all you want to explain how this is 'statistically consistent' or whatever other defense you care to make, I don't care - it is B*** S***.

I do not believe that my failure to understand this phenomenon is at all tied to my avoiding Stats, Macro or Micro Econ while at GU.

(couldn't wait for official updated NCAA, used thier officials rankings from last week and the latest up to the minute from the link up top of hoyasaxa.com (kenpom)

Ken Pomeroy hates the RPI and thinks it is a horrible statistical evaluation. He provides the number, because it is what the committee uses to determine the invitees to the Tourney, but he has his own (more accurate) statistical formula.

If you are curious, we are #7 in his own rankings, only behind UNC, Kansas, TX A&M, Florida, UCLA and Ohio St.

kenpom.com/rate.php

[FOTP]

Pomeroy is pretty dead on IMO.

He does have MD around 12 or so which is interesting and ND is pretty high.

I actually think his is the most accurate and it's what I focus on. RPI is ok, but not the end all be all. When you look at Pomeroy's rankings you kind of wind up nodding your head because it makes sense in terms of the best teams right now.

[DoubleOhHoya]

seaweed said:

Otherwise how do you explain our RPI DROPPING from last week to this week despite the two major conference wins? Take care of business on the road against Cincy and then beat a top 10 (RPI #6) team at home in one week and DROP in the ratings? Somehow, despite another dominating week , we go down from 16 to 17 in RPI?

If this is not final proof that RPI is meaningless, then nothing will ever convince the kenpom sycophants. Try all you want to explain how this is 'statistically consistent' or whatever other defense you care to make, I don't care - it is B*** S***.

I do not believe that my failure to understand this phenomenon is at all tied to my avoiding Stats, Macro or Micro Econ while at GU.

(couldn't wait for official updated NCAA, used thier officials rankings from last week and the latest up to the minute from the link up top of hoyasaxa.com (kenpom)

Don't blame Pomeroy for that mess. His ratings are there BECAUSE the RPI sucks, and he has us appropriately ranked.

[thornski]

Also agree that Pomeroy is very good, although a few little outliers...mentioned MD at 12, also had Georgia Tech at 17.

Another example, I know how much we've been ripping on Nevada (and no way should they be ahead of us in the coaches poll), but Pomeroy has them at 63!! I mean I know they're overrated, but are they THAT overrated?

[DoubleOhHoya]

FWIW, Nevada's Pomeroy rating is a big 63. I smell a fraud. Other frauds include Butler (23), Washington State (27), Southern Illinois (34), and heck, I'll say it, Oregon (44).

[DoubleOhHoya]

thornski said:

I mean I know they're overrated, but are they THAT overrated?

Yes.

[FOTP]

Agreed. Nevada is the real GW this year. They've played no one and the league is terrible.

Ken Pom is dead on. Trust his rankings and if you like to gamble bet on his rankings and predictions. Works pretty well.

[seaweed]

OK, so maybe the swipe at Pomeroy was unfair. What I am getting at is the general obsession with statistical models that don't accurately tell what happens on the floor.

I agree Pomeroys ratings are MORE accurate, but really, he has UNC on top after loosing just yesterday. I have a problem with that and I don't buy his Luck stats either.

[KHoyaNYC]

Pomeroy has Memphis in the top 10. Just saying.....

[SFHoya99]

RPI is kind of a joke stat...

But with any ranking, how can you say that a team that wins 2 games can't move down? We beat a crappy team at home and a good team at home. If someone was ranked right behind us, and say, beat the #1 and #2 teams on the road, wouldn't they deserve to move up higher than us?

The original posts' logic -- we can't move down because we didn't lose and ignore what others did -- is the same logic that give us the polls we bitch about every day. You automatically move down if you lose -- regardless of who against, and you don't get jumped unless you lose.

[Cambridge]

seaweed said:

OK, so maybe the swipe at Pomeroy was unfair. What I am getting at is the general obsession with statistical models that don't accurately tell what happens on the floor.

I agree Pomeroys ratings are MORE accurate, but really, he has UNC on top after loosing just yesterday. I have a problem with that and I don't buy his Luck stats either.

The luck stat (which pomeroy wrote an entire blog on, and stated his apprehnsion vis a vis the name) is meant to measure the number of close wins and losses that were determined by one or two possessions.

In his opinion, the number of close games that came down to one or two bounces this way or that is a reflection on the element of chance in college basketball. Therefore, that stat is meant to show whether the ball has either bounced more in their favor (see OSU with a lot of close wins) or against (see Georgetown with a few close losses and very, very few close wins; mostly blowouts) in close games.

He isn't saying that those teams didn't deserve to win or lose those close games, he's merely creating a statistical measure of how often they have been at the mercy of late game officiating, late game heroics, late game clutch performances, etc.

This allows you to then look at the team's record and discount or reconsider their record. For example, OSU and UCLA have been the luckiest teams in the top 10, while Georgetown and UNC have benefited from the least amount of luck. You could read that as OSU and UCLA getting it done in crunch time...or them being only a few lucky breaks from 2-3 extra losses this year.

On the flip side, Gtown and UNC have had very few close wins and have suffered a couple very close losses. Therefore, a couple bad breaks have cost them 1-2 wins.

[ExcitableBoy]

The main value of the RPI as far as I'm concerned is the ability to judge how teams have fared against top tier, second tier, third tier etc competition.

A ranking system that puts Kentucky above us at this point in the season clearly has some major issues. Where it is valuable is for getting a rough estimate of how teams have done versus top tier competition (record against RPI top 50 teams - we're 6-5, Kentucky is 3-8, Nevada is 0-1) and also exposing bad loses (BC is only 2-1 against RPI 201+ teams).

The actual number that a team is ranked is important, sure, but it's far more valuable to me to be able to assess relative strength of schedules and quantify good wins and bad losses.

[hoyalove4ever]

RPI is the worst statistic ever. Period. It is not a valid "tool"; rather, it is used by tools who are too lazy to give even a cursory evaluation of different teams and want to base a ranking on some "objective" number.

The home/ road differential is awful, but it is not the worst aspect of the RPI. The worst thing is that the RPI tries to quantify wins over cupcakes teams. It makes a huge difference to the RPI if you beat the 209 RPI team instead of the 290 RPI team...when it really makes no difference. That's a cupcake game. You won. End of story. Wins like that should not be a factor in the NCAA analysis, but RPI has dragged them in. What a disgrace.

Everyone speaks out about the BCS. It's time for fans to speak up against the RPI.