RPI

[Jerry1714]

Ironically, kenpom is not very good. So I guess it shouldn't surprise me that the NCAA uses it. I'm not saying it is RPI awful. It just isn't very good. Sagarin is better and, for those who want more than just one or two numbers to look at, Mike Greefield does a great job at teamrankings.com. In fact, let's look at those numbers, as of today:

Rating Value Rank Conf Rank
Predictive Rating 14.6 #22 #2
Home Rating 13.4 #30 #3
Away Rating 19.7 #4 #1
Neutral Rating 13.3 #34 #3
Home Advantage 0.3 #257 #7
Strength Of Schedule 6.3 #41 #2
Future SOS 9.3 #47 #2
Season SOS 7.1 #39 #1
SOS - Basic 2.0 #81 #5
In-Conf SOS 5.8 #81 #6
Non-Conf SOS 6.6 #14 #1
Last 5 Rating 19.0 #8 #1
Last 10 Rating 13.8 #31 #1
In-Conf Rating 13.6 #32 #2
Non-Conf Rating 16.1 #22 #3
Luck Rating -1.1 #262 #7
Consistency Rating 10.0 #170 #6
Vs. 1-25 Rating 9.1 #68 #5
Vs. 26-50 Rating 16.5 #24 #3
Vs. 51-100 Rating 14.1 #31 #1
Vs. 101-200 Rating 21.1 #8 #1
Vs. 201+ Rating 14.2 #34 #3
First Half Rating 7.1 #33 #2
Second Half Rating 7.9 #18 #2

The two things that jump out are our crappy defense of home court and our crappy play against Top 25 teams. This team has to win the home games they should win and they HAVE to beat a Top 25 team. The only one left is SMU. I guess one could look at our "luck" rating as bad too. That is just our inability to win close games. That isn't good either and has to change.

Kenpom is considered to the best indicator of all the metrics, not surprising you think it is not very good.

 

[OkaForPrez]

250? 150 would be more useful tool. You have to be pretty bad to be sub 150 and any win by a tourney team over a sub 150 should be worthless.

Yes but sub 150 losses are occurring a bit more than what I would call an anomaly. So you don't want a system which considers a team ranked 155 the same as a team ranked 345, those are significantly different opponents in my opinion. But given there's been only 2 losses to teams +250, I think its safe to say playing the 251st team in the RPI system is = to playing the 350th team. If there were only 2 collective losses from teh at large field to teams ranked 150 and above, I would feel comfortable making the cut there.

 

[pepband99]

devil's advocate - wouldn't "capping" lower numbers just compress the importance of the higher ones? Meaning, you would have the same argument about how much difference there really is between 150 and 140, for example.

It's like adding teams to the tourney - they'll whine as the 33rd team / 65th team / 69th team / 129th team etc.

 

[OkaForPrez]

@pepband99 Lesser evil. I'd rather have a system that is considering that difference to a greater degree than lending a significant greater weight to beating a dog vs. beating a dog with fleas.

 

[tzznandrew]

But in 2012 we had an RPI of 32 and our SOS was #2 (damn, I didn't even know it was that high)

Butler currently has an RPI of 57 and their SOS is 90.

Their record right now 14-6. Don't see how they'd be able to get in with another 4 losses in their conference.

Very curious about Butler and how they'll do.

Two weeks from this, they've gone 3-1.

They are 17-7 (6-6)
RPI: 58
SOS: 91
vs. RPI 25: 1-3
vs. RPI 50: 1-5
vs. RPI 100: 5-5

They've tread water despite going 3-1. Now they have H-Xavier, H-Creighton, @Nova, @Gtown.

They have to win the Creighton/Georgetown games, I'd think, to lose both the Xavier and Nova games.

Regardless, I'm really interested to see what the committee does with this team.

 

[Paesano]

you are always a ray of sunshine

teams never lose at home and beat the same team on the road, it never happens, ever

Any other tidbits of wisdom, Mr. Positive?

But, "let me be clear", I never said never. I only asked why one would expect it. Logic dictates that it is less than 50% probable.

So Kenpom was wrong, again, I guess. But, "let me be clear", I don't hate Ken. His ratings aren't awful. They just aren't the best.

 

[bmo22]

RPI: 46
BPI: 26
KenPom: 22
SOS: 57

 

[Jerry1714]

Any other tidbits of wisdom, Mr. Positive?

But, "let me be clear", I never said never. I only asked why one would expect it. Logic dictates that it is less than 50% probable.

So Kenpom was wrong, again, I guess. But, "let me be clear", I don't hate Ken. His ratings aren't awful. They just aren't the best.

Maybe your logic, which we have all seen is clearly flawed, but not reality.

 

[methodology]

Any other tidbits of wisdom, Mr. Positive?

But, "let me be clear", I never said never. I only asked why one would expect it. Logic dictates that it is less than 50% probable.

So Kenpom was wrong, again, I guess. But, "let me be clear", I don't hate Ken. His ratings aren't awful. They just aren't the best.

Kenpom ratings are the best actually.

 

[Paesano]

Kenpom ratings are the best actually.

Why? Because you say so? I have actually spoken to most of these guys (Sagarin, Greenfield, Pomeroy and Massey). I have analyzed their systems in multiple ways. Pomeroy provides some nice metrics that no one else does but his overall ratings are not the best. Not even close. Sagarin and I have something in common that is relevant to this discussion. I will leave it to you try and figure out what it is.

 

[auror]

For those saying the committee use things other than RPI like KenPom: that is true.

But look at how EVERYTHING is related to RPI on the committee's nitty gritty sheet. It's not just a ranking, it's their organization tool.

 

[Rocktheworld]

Why? Because you say so? I have actually spoken to most of these guys (Sagarin, Greenfield, Pomeroy and Massey). I have analyzed their systems in multiple ways. Pomeroy provides some nice metrics that no one else does but his overall ratings are not the best. Not even close. Sagarin and I have something in common that is relevant to this discussion. I will leave it to you try and figure out what it is.

 

[4in16]

Currently... 43 RPI 22 BPI 23 KenPom SOS 51, 2-3 top 50 and 6-7 top 100. "Worst" loss Temple 68 Best win Texas 21.

 

[auror]

Currently... 43 RPI 22 BPI 23 KenPom SOS 51, 2-3 top 50 and 6-7 top 100. "Worst" loss Temple 68 Best win Texas 21.

Pretty huge if Tulsa can stay top 50 RPI.

 

[caw]

Kenpom ratings are the best actually.

In what way? I mean what are they the best at? Predicting results? Analyzing which team is the best? Analyzing which team has performed the best? I may have jumped into this conversation in the middle but I just want to know what best you all are talking about?

 

[SubbaBub]

Committee Chair gave an interview saying they use all these ratings and more. Including the BPI and the RPI.

 

[BUConn10]

Committee Chair gave an interview saying they use all these ratings and more. Including the BPI and the RPI.

Do they use Kenpom or Sagarin?

 

[VAHuskyFan]

The NCAA started using RPI in 1981 because there had been point-shaving scandals and margin of victory is associated with gambling spreads. The NCAA wanted to distance itself from all that. The NCAA was also concerned about teams running up the score to improve their ranking.

So enter RPI to provide an "objective" measure of a team's strength without factoring in margin of victory. RPI solved a PR problem for the NCAA -- they didn't choose it because it was the best system.

Today, the Selection Committee doesn't rely on RPI -- they basically just pretend to. RPI is a nice safe system. So the NCAA will continue to give lip service to using it (as "one part" or "an important part") in their analysis, even though they know it's garbage.

 

[SubbaBub]

Do they use Kenpom or Sagarin?

My guess is they use whatever they want.

 

[auror]

Committee Chair gave an interview saying they use all these ratings and more. Including the BPI and the RPI.

See my post above about yes, they do, but it's all still framed by the RPI.

Today, the Selection Committee doesn't rely on RPI -- they basically just pretend to. RPI is a nice safe system. So the NCAA will continue to give lip service to using it (as "one part" or "an important part") in their analysis, even though they know it's garbage.

Look at that nitty gritty sheet posted above and see how every single thing on there is framed by RPI. They might not rely on the single number, but they look at how team's perform against teams in those RPI brackets, etc.

 

[VAHuskyFan]

I agree with you that the NCAA Selection Committee frames its communications in terms of RPI.

My point is that it's important to understand why they do that when everyone knows there are much better systems than RPI out there. It's for PR purposes, and you can see that historically.

So the NCAA talks about RPI a lot publicly, but the Committee's decisions over the last several years show that they know it sucks and place greater emphasis on other factors behind closed doors.

See my post above about yes, they do, but it's all still framed by the RPI.



Look at that nitty gritty sheet posted above and see how every single thing on there is framed by RPI. They might not rely on the single number, but they look at how team's perform against teams in those RPI brackets, etc.

 

[Paesano]

I agree with you that the NCAA Selection Committee frames its communications in terms of RPI.

My point is that it's important to understand why they do that when everyone knows there are much better systems than RPI out there. It's for PR purposes, and you can see that historically.

So the NCAA talks about RPI a lot publicly, but the Committee's decisions over the last several years show that they know it sucks and place greater emphasis on other factors behind closed doors.

I would love to believe that but the NCAA is so stupid and corrupt that I am skeptical. Do you have some examples?

 

[Hi-Low]

Reasons #138 & #139 why the Rpi is a joke......

- Princeton has an Rpi of 42....top 100 wins - O....best wins Columbia (128) & Hampton (188)
- Akron has an Rpi of 41....top 100 wins - Ohio (77) & Santa Barbara (94)

& UConn has an Rpi of 43....top 100 wins. - TX, Tulsa, Mich, G-Town, OH St, Houston

 

[VAHuskyFan]

I would love to believe that but the NCAA is so stupid and corrupt that I am skeptical. Do you have some examples?

Folks aren't going to like this, but the AAC last year was an example.

SMU finished the season #13 RPI (4-seed) but #24 BPI (6-seed). They got a 6-seed.

Temple finished the season #34 RPI (9-seed) but #66 BPI (no bid). They didn't get a bid.

The AAC's RPI numbers were inflated last year (Tulsa lost to a bad Div II school, which RPI doesn't count - just another example of it being a garbage ranking). Obv the Committee looked past that. And there have been studies showing BPI is a better predictor than RPI.

That doesn't mean the NCAA isn't corrupt! Pretending to do one thing for PR purposes and actually doing something different behind the scenes is standard NCAA operating procedure. And of course human bias comes in as well in terms of odd bubble decisions that tend to favor P5 conference mates of Selection Committee members.

 

[HooperScooper]

Folks aren't going to like this, but the AAC last year was an example.

SMU finished the season #13 RPI (4-seed) but #24 BPI (6-seed). They got a 6-seed.

Temple finished the season #34 RPI (9-seed) but #66 BPI (no bid). They didn't get a bid.

The AAC's RPI numbers were inflated last year (Tulsa lost to a bad Div II school, which RPI doesn't count - just another example of it being a garbage ranking). Obv the Committee looked past that. And there have been studies showing BPI is a better predictor than RPI.

That doesn't mean the NCAA isn't corrupt! Pretending to do one thing for PR purposes and actually doing something different behind the scenes is standard NCAA operating procedure. And of course human bias comes in as well in terms of odd bubble decisions that tend to favor P5 conference mates of Selection Committee members.

Actually I think we will like this. UConn's BPI is 22 (http://espn.go.com/mens-college-basketball/bpi). So following your examples, a 6 seed. I'll take it.

I've said this a million times, the RPI of your own team is not a good indicator of getting into the tournament or the seeding. The committee uses the RPI as a tool to see who you've beaten and lost to. They need some kind of ranking for teams to define the level of teams you've beaten and lost to. Obviously teams who have great records and beaten some good teams have high RPIs. But I've seen it written a bunch of times that the AP and Coaches Polls are actually better indicators of seeding than the RPI. Just win baby!