PDA

View Full Version : Rpi is crap



realtydog
02-24-2011, 08:14 AM
# 37. Cleveland St. 23 - 6
SOS =109
vs. top 50 0 - 4
vs. 51-100 4 - 1

# 71. Gonzaga 19 - 9
SOS = 89
vs. top 50 1 - 6
vs. 51-100 5 - 1

how in the hell are they so high----are they really twice the team GU is??? give me a break---next year we need to put the whole big sky on the NOn Con schedule I guess if that will help us have a 37 RPI

willandi
02-24-2011, 08:22 AM
RPI IS crap. What we need to do is schedule all the best DIV 1 teams from the really crappy conferences, at their house, and win all those games. Take care to win the games at home we should, and handle in conference, and our RPI would be in the 20's.

bullzag23
02-24-2011, 08:33 AM
# 37. Cleveland St. 23 - 6
SOS =109
vs. top 50 0 - 4
vs. 51-100 4 - 1

# 71. Gonzaga 19 - 9
SOS = 89
vs. top 50 1 - 6
vs. 51-100 5 - 1

how in the hell are they so high----are they really twice the team GU is??? give me a break---next year we need to put the whole big sky on the NOn Con schedule I guess if that will help us have a 37 RPI

Winning percentage is a huge (1/3 I believe) of RPI, so that makes a fairly big difference (GU is 0.679 vs 0.793 for Cleveland St.). If we had won a couple more of those top 50 games we'd be looking a hell of a lot better right now RPI (and NCAA) wise.

GU69
02-24-2011, 08:41 AM
Ken Pom has GU at 50, Cleveland State at 68.

Angelo Roncalli
02-24-2011, 08:42 AM
Winning percentage is a huge (1/3 I believe) of RPI, so that makes a fairly big difference (GU is 0.679 vs 0.793 for Cleveland St.). If we had won a couple more of those top 50 games we'd be looking a hell of a lot better right now RPI (and NCAA) wise.

The RPI is not a subjective "eye test" ranking like the AP poll or the coaches' poll. It's not a poll at all. The RPI is a relatively simple mathematical formula. It isn't subjective. It doesn't take into account strength of schedule, reputation, injuries, or any "style point" components. It is what it is. It's a tool. It's not the be all, end all for the selection committee.

CDC84
02-24-2011, 08:50 AM
As Coach Few talked about during the coaches show, certain programs and conferences have figured out how to rig the RPI in their favor.

Personally, I look at Cleveland State's numbers and don't really even pay much attention to the fact that they are at 37. What I pay more attention to is what they have done against the RPI top 50 and the RPI top 100. They haven't been anymore impressive than the Zags.

All of these ratings systems have problems, which is why Coach Few thinks they should all be combined and averaged out. Kenpom.com has problems: the fact that Washington is ranked at #10 is all the evidence you need to know. Sagarin has issues because it incorporates margin of victory in a sport where it only takes six one point victories to win the NCAA championship. Not to mention the fact that margin of victory involves so many other factors like the size of your bench, a head coach's willingness to run up the score, etc.

They are really just tools for evaluating teams.

Once and Future Zag
02-24-2011, 09:04 AM
From an earlier thread:


RPI is a very clumsy and imprecise tool as it doesn't take into account the in-game performance, but just captures a "snapshot" when the clock runs out.

Let's say (for simplicity and clarity of explanation sake) CBB played a 5 game season.

Let's say the Little Sisters of the Poor University and Sports Bar plays the following teams

LSOPaSB 64 - Duke 68
LSOPaSB 54 - Ohio State 55
LSOPaSB 103 - Pitt 105 (2ot)
LSOPaSB 85 - Southern - 32
LSOPaSB 90 - Centenary - 54

vs

LSOPaSB 44 - Duke 73
LSOPaSB 54 - Ohio State 93
LSOPaSB 64 - Pitt 82
LSOPaSB 54 - Southern - 51
LSOPaSB 70 - Centenary - 62

In both cases their RPI would be the same - but one is clearly a better team.

The good thing is that the committee knows this and doesn't even have a given team's RPI on their info-sheet, just their records against top 50, top 100, and sub 200 losses.

It may not be precise, but it gives a reasonable breakdown to the committee of W/L against good teams, and bad teams.

I'd prefer they use some sort of variation on KenPom's numbers as it's actually a better metric of how good a team is - normalized for SoS.



All of these ratings systems have problems, which is why Coach Few thinks they should all be combined and averaged out. Kenpom.com has problems: the fact that Washington is ranked at #10 is all the evidence you need to know.

Why does that indicate problems per se? There aren't a lot of great teams this year - so a team that's inconsistent on a per-game level, but overall puts together a decent season will rank high - especially if they're gaudy at one end of the court, and decent on the other.

webspinnre
02-24-2011, 09:07 AM
They are really just tools for evaluating teams.

This is the key. All of these are useful tools, but the important part is you need a human to be able to interpret them, because the human can, in many of these cases, adjust for variables the statistical model isn't equipped to handle.

CDC84
02-24-2011, 09:37 AM
Why does that indicate problems per se? There aren't a lot of great teams this year - so a team that's inconsistent on a per-game level, but overall puts together a decent season will rank high - especially if they're gaudy at one end of the court, and decent on the other.

There are problems with these ratings systems every single year. It doesn't matter if college basketball is experiencing a down season. Perhaps the flaws are more pronounced in certain years....

I don't have an issue with the ratings systems having flaws. I've known that they've had flaws for years, and yet I still pay attention to them. I have issue with anyone who feels they are the be all, end all when it comes to evaluating the goodness of teams. And there are those people......

They are just tools.

bullzag23
02-24-2011, 09:38 AM
The RPI is not a subjective "eye test" ranking like the AP poll or the coaches' poll. It's not a poll at all. The RPI is a relatively simple mathematical formula. It isn't subjective. It doesn't take into account strength of schedule, reputation, injuries, or any "style point" components. It is what it is. It's a tool. It's not the be all, end all for the selection committee.

I wasn't suggesting that RPI is subjective by any means, just trying to say that if we had won a couple of our top 50 games the RPI would jump up by a decent amount because our winning percentage would be higher than it is currently. I also realize that 2-3 more wins wouldn't get us anywhere near #37 from #71 but we would be higher without a doubt.

Shanachie
02-24-2011, 10:57 AM
CDC84 is right - RPI is one of several ratings systems that can be useful, and it is particularly important because it is used by the selection committee. While there are and always will be flaws with any rating system, remember that RPI is generally good for teams like Gonzaga. RPI through the years has been an important factor in the ongoing argument between so-called "mid-majors" and the big boys. It gives an objective way of measuring an 18-14 ACC team vs. a 22-7 WCC team and makes it harder to brush aside the smaller conference teams with the old "if they were in the ACC they wouldn't win 3 games" BS we used to hear (and sometimes still do).

And just to follow up on a few other comments from this thread:


The RPI is not a subjective "eye test" ranking like the AP poll or the coaches' poll. It's not a poll at all. The RPI is a relatively simple mathematical formula. It isn't subjective. It doesn't take into account strength of schedule, reputation, injuries, or any "style point" components. It is what it is. It's a tool. It's not the be all, end all for the selection committee.

True on all points, except the strength of schedule. SOS is an important factor in RPI.


... if we had won a couple of our top 50 games the RPI would jump up by a decent amount because our winning percentage would be higher than it is currently. I also realize that 2-3 more wins wouldn't get us anywhere near #37 from #71 but we would be higher without a doubt.

True about the RPI jump if we had 2-3 more wins, regardless of whether they were against top 50 or 100+ teams. The exact RPI scores and rankings vary a bit from site to site based on timing etc., but according to realtimerpi.com, Gonzaga's RPI is .5593 for a rank of 73. This is with a listed record of 18-9. Ignoring the home/away adjustments, if their record was 21-6, all else equal, GU's RPI score would be .0277 higher, or .5870. This would put them at #43 on the list and in an entirely different position in the at large discussion.

Angelo Roncalli
02-24-2011, 11:08 AM
I wasn't suggesting that RPI is subjective by any means, just trying to say that if we had won a couple of our top 50 games the RPI would jump up by a decent amount because our winning percentage would be higher than it is currently. I also realize that 2-3 more wins wouldn't get us anywhere near #37 from #71 but we would be higher without a doubt.]


You and I agree. I was attempting to give background to what it really is.

Ezag
02-24-2011, 11:39 AM
I reiterate "RPI is Crap!"

rijman
02-24-2011, 11:51 AM
CDC84 is right - RPI is one of several ratings systems that can be useful, and it is particularly important because it is used by the selection committee. While there are and always will be flaws with any rating system, remember that RPI is generally good for teams like Gonzaga. RPI through the years has been an important factor in the ongoing argument between so-called "mid-majors" and the big boys.

But, as coach Few discusses in his show, which RPI? They can vary significantly from source to source. Few suggests averaging some RPI's to get the overall RPI rating. Few also opined that the offensive and defensive efficiency ratings of each team might provide a more accurate ranking system than RPI's.

jake
02-24-2011, 12:21 PM
But, as coach Few discusses in his show, which RPI? They can vary significantly from source to source. Few suggests averaging some RPI's to get the overall RPI rating. Few also opined that the offensive and defensive efficiency ratings of each team might provide a more accurate ranking system than RPI's.

The various ranking systems vary, but unless they are using a different date for their information, RPI shouldn't vary. It's just a formula.

webspinnre
02-24-2011, 12:44 PM
Different places have different RPI formulas, weighting, etc.

bullzag23
02-24-2011, 12:51 PM
But, as coach Few discusses in his show, which RPI? They can vary significantly from source to source. Few suggests averaging some RPI's to get the overall RPI rating. Few also opined that the offensive and defensive efficiency ratings of each team might provide a more accurate ranking system than RPI's.

If you simply averaged out our offensive and defensive efficiency we're roughly 60th in the nation, at least according to KenPom. This is using a very simple metric though (O Eff + D Eff)/2, nothing complicated by any means. Here's the shocker......GU's Offensive Efficiency is 70th, and our Defensive Efficiency is......49th despite very poor 3 point D (291st). Here's the best part: Our 2 point FG% Defense is ranked 10th. Force the Gaels to beat us inside!!!!

gamagin
02-24-2011, 01:31 PM
is like a beauty pageant. It is used, like the Miss Congeniality "award", to explain to all her loyal fans why she, perhaps, did not advance with the rest of the pack for a shot at the crown.

However, so as not to completely ignore her beauty and overall widespread popularity (or in our case long term presence), out of respect, she is officially noticed and given a pat on the head just before she is escorted off stage and out the door while the real contest continues.

229SintoZag
02-24-2011, 01:35 PM
contrasting this thread to the threads on the RPI back when we were a team that finished in the top 25 of the RPI. Back then, RPI was the gold standard.

AK457
02-24-2011, 02:30 PM
rpi isn't crap, it's just a way of quantifying something that's often viewed subjectively (e.g., ap poll going "oh it's duke! name recognition! they must be top 10!). sure there's ways to exploit it because you can look at the formulas and try and optimize your schedule to win the most games with the least effort and risk, but it does a good job of identifying top performing schools. it's analogous to the whole "the USA system of democracy sucks but it's the best system of government we've come up with" deal.

titopoet
02-24-2011, 03:24 PM
What is funny is that Cleveland St is not even on the bubble. The RPI is used only when it favors the big school and tossed when it is the small school. The reality is Baylor and Marquette are both above GU in the bubblewatch; even as GU has beaten them and has similar RPI (though the true RPI that the NCAA uses is suppose to be secret and the others just approximations.)

The Zags should not worry about Cleveland ST's numbers but how teams like Virgina Tech, Maryland, Penn State, Washington, and host of middling big 6 schools finish.

Hextall7388
02-24-2011, 05:38 PM
On the topic of RPI and SOS, I am confused as to why Coach Few and Co. continue to schedule RPI/SOS killer games against the likes of Southern University (RPI 343), EWU (RPI 314) and Cal-State Bakersfield (RPI 313) each year.

If you include San Diego's gruesome RPI of 305 into the mix, by the end of the season the Zags will have played 5 games against teams with an RPI over 300 (granted we have no choice in scheduling SD). My point is, these 5 games have a lot to do with why the Zags' RPI is hovering around 70, and their SOS ranking is around 100. As I understand it, 50% of the RPI formula is based on the winning percentage of a team's opponents. These 5 teams (I've counted SD twice) have a combined win/loss record of 33-99. Just using some simple math, it appears that all of the hard work that goes into scheduling games against the likes of Illinois, Baylor, Marquette, Xavier, OK State, Kansas State etc., and the boost that the Zags receive in the RPI and SOS standings from this games is negated by scheduling so many games against 300+ RPI teams. The historical stats show us that each year there will be at least 1 WCC team with an RPI above 300. So, why do we need schedule more RPI/SOS killer games? I understand that the Zags need to fill our their schedule, and the teams needs a few "easy" games in the mix, but surely there are a couple 100-200 RPI teams out there that the Zags can schedule.

Oh....before people jump to Coach Few's defense to argue that it's impossible to predict an opponent's RPI when scheduling games months if not years ahead of time, I suggest you take a quick look at the historical stats of these three programs. Cal-State and Southern have had an RPI above 300 in each of the last 5 years. EWU is not much better (2 out of the last 4 years). You do not have to see the future to know that these teams are going to stink year in and year out.


Of interest - here are a few of the teams that the Zags are competing with for a possible at-large bid with their RPI, SOS and #of games against +300 RPI teams:

Butler: 46, 86, 0
Cleveland State: 37, 116, 0
Memphis: 38, 53, 1
Old Dominion: 29, 68, 1
St. Marys: 45, 53, 3

I admit that I did not search every single team in the RPI top 75, but of the teams that I looked at, the Zags had the most games against 300+ RPI teams (St. Marys was second). Knowing the schedule that the Zags have played this year, I was shocked to see the list of teams that have a better SOS ranking. I haven't crunched all the numbers, but I am sure these 5 games against 300+ RPI teams have a lot to do with the Zags underwhelming RPI and SOS numbers.

Am I missing something....why are these games on the schedule?