It appears the RPI and KPI significantly undervalue Cincy and UConn compared to the predictive power ratings (the other four: BPI, Sagarin, Pomeroy, LRMC).
The RPI and KPI significantly overvalue Temple compared to the predictive power ratings.
And the RPI/KPI are in general agreement on SMU, Tulsa and Houston compared to the predictive power ratings.
(This post was last modified: 02-15-2016 08:06 AM by CougarRed.)
Those 'predictors' would predict Temple to lose to SMU, UCONN x2, Cinci x2, and Tulsa as well as beat Houston at home. I'd say the predictors are overvalued compared to the ratings index known as reality.
(02-15-2016 08:08 AM)Cecil_B_DeMoore Wrote: Those 'predictors' would predict Temple to lose to SMU, UCONN x2, Cinci x2, and Tulsa as well as beat Houston at home. I'd say the predictors are overvalued compared to the ratings index known as reality.
The committee is the only group that would disagree with you.
Not sure why the committee would disagree with me since by the time they make their decisions the predictors don't matter because there are no more games to predict. Predictors have nothing to do with the selection committee's job.
(02-15-2016 09:41 AM)Cecil_B_DeMoore Wrote: Not sure why the committee would disagree with me since by the time they make their decisions the predictors don't matter because there are no more games to predict. Predictors have nothing to do with the selection committee's job.
They actually have everything to do with the committee's job. You just don't like the results.
As a whole, these rankings systems have been shown to be a more true representation of teams and their relative standing than the outdated RPI system, which is oversimplified and struggles to adapt. That's useful information to the committee.
But as you point out, there is a reason we have humans making the final decisions. None of these systems are perfect, and having actual people to watch the games and look at the results matters. But, whether you like it or not, the predictive analytic systems are mattering more and more to the committee.
(02-15-2016 09:41 AM)Cecil_B_DeMoore Wrote: Not sure why the committee would disagree with me since by the time they make their decisions the predictors don't matter because there are no more games to predict. Predictors have nothing to do with the selection committee's job.
They actually have everything to do with the committee's job. You just don't like the results.
As a whole, these rankings systems have been shown to be a more true representation of teams and their relative standing than the outdated RPI system, which is oversimplified and struggles to adapt. That's useful information to the committee.
But as you point out, there is a reason we have humans making the final decisions. None of these systems are perfect, and having actual people to watch the games and look at the results matters. But, whether you like it or not, the predictive analytic systems are mattering more and more to the committee.
Simple Question: Do you believe UC is the 28th or 30th best team and should get a 6 or 7 seed ? That's what the predictors are saying. Yet every bracketologists has UC down in the 11,12 seed or out, which is in line with the RPI.
(02-15-2016 09:41 AM)Cecil_B_DeMoore Wrote: Not sure why the committee would disagree with me since by the time they make their decisions the predictors don't matter because there are no more games to predict. Predictors have nothing to do with the selection committee's job.
They actually have everything to do with the committee's job. You just don't like the results.
As a whole, these rankings systems have been shown to be a more true representation of teams and their relative standing than the outdated RPI system, which is oversimplified and struggles to adapt. That's useful information to the committee.
But as you point out, there is a reason we have humans making the final decisions. None of these systems are perfect, and having actual people to watch the games and look at the results matters. But, whether you like it or not, the predictive analytic systems are mattering more and more to the committee.
Simple Question: Do you believe UC is the 28th or 30th best team and should get a 6 or 7 seed ? That's what the predictors are saying. Yet every bracketologists has UC down in the 11,12 seed or out, which is in line with the RPI.
Those are 2 different questions. No I don't think they deserve a 6 or 7 seed. Do I think they are the 28th-30th best team? Personally, I think they are pretty close to that. You don't use the KenPom, Sagarin, etc as straight ranking systems any more than we use the RPI as a straight ranking system. You balance out who you think are the best teams vs what teams have earned during the regular season.
If you use UC's RPI at present, UC isn't even on the bubble. They are about 15 spots away from it. The bracketologists are using more than just the RPI, which is why UC is where it is.
Question for you: Do you think we should just roll straight down the RPI list and select the tournament teams starting from #1 there until we run out of spots?
(02-15-2016 09:41 AM)Cecil_B_DeMoore Wrote: Not sure why the committee would disagree with me since by the time they make their decisions the predictors don't matter because there are no more games to predict. Predictors have nothing to do with the selection committee's job.
They actually have everything to do with the committee's job. You just don't like the results.
As a whole, these rankings systems have been shown to be a more true representation of teams and their relative standing than the outdated RPI system, which is oversimplified and struggles to adapt. That's useful information to the committee.
But as you point out, there is a reason we have humans making the final decisions. None of these systems are perfect, and having actual people to watch the games and look at the results matters. But, whether you like it or not, the predictive analytic systems are mattering more and more to the committee.
Simple Question: Do you believe UC is the 28th or 30th best team and should get a 6 or 7 seed ? That's what the predictors are saying. Yet every bracketologists has UC down in the 11,12 seed or out, which is in line with the RPI.
Those are 2 different questions. No I don't think they deserve a 6 or 7 seed. Do I think they are the 28th-30th best team? Personally, I think they are pretty close to that. You don't use the KenPom, Sagarin, etc as straight ranking systems any more than we use the RPI as a straight ranking system. You balance out who you think are the best teams vs what teams have earned during the regular season.
If you use UC's RPI at present, UC isn't even on the bubble. They are about 15 spots away from it. The bracketologists are using more than just the RPI, which is why UC is where it is.
Question for you: Do you think we should just roll straight down the RPI list and select the tournament teams starting from #1 there until we run out of spots?
No, what I really believe, is they use the eye test and what other committee members tell each other way more than they let on.
(02-15-2016 09:41 AM)Cecil_B_DeMoore Wrote: Not sure why the committee would disagree with me since by the time they make their decisions the predictors don't matter because there are no more games to predict. Predictors have nothing to do with the selection committee's job.
They actually have everything to do with the committee's job. You just don't like the results.
As a whole, these rankings systems have been shown to be a more true representation of teams and their relative standing than the outdated RPI system, which is oversimplified and struggles to adapt. That's useful information to the committee.
But as you point out, there is a reason we have humans making the final decisions. None of these systems are perfect, and having actual people to watch the games and look at the results matters. But, whether you like it or not, the predictive analytic systems are mattering more and more to the committee.
Simple Question: Do you believe UC is the 28th or 30th best team and should get a 6 or 7 seed ? That's what the predictors are saying. Yet every bracketologists has UC down in the 11,12 seed or out, which is in line with the RPI.
Yes I believe UC is one of the 30 best teams in the country as the predictive rankings suggest.
No I don't think UC has a 6 or 7 seed resume at this time, which is an entirely different question. UC's RPI is not in line with what bracketologists are saying either. Bracketologists are putting UC somewhere between the predictors and the RPI ranking and that seems about right. UC's big problem is they lost too many close games, so they don't have the wins the committee looks for. Had UC say held on against Butler and SMU (Still keeping losses to Temple, Iowa State and Memphis)... their predictor rankings wouldn't be much different, but I think you'd see UC in that 6 or 7 seed range.
Your tournament seed shouldn't just be a recitation of your kenpom / sagarin rankings. I cite them often, not because they reflect where you will be placed in the tournament, but because they are a good measure of how good a team you actually are. The committee has consistently valued who you beat, where you beat them and who you lose to / where you lose (and they should). UC has lost too many of it's big games (in agonizing fashion) to be a 6 or 7 seed at this point, but it's not surprising that most predictors have them as a top 30 team, because the play on the Court does suggest that, despite those losses.
(This post was last modified: 02-15-2016 10:54 AM by bearcatmark.)
The committee doesn't use RPI as their guide for whether teams get in or not.
The committee uses RPI for seeding purposes after the teams are selected...and also have to factor in splitting up conference teams, look at sites, etc...
The committee though used the advanced metrics more last year than ever before. Really don't think that's a 1 year thing quite frankly. It's why UConn and Cincy are in a lot better shape than folks think- and why Temple really has a ton of work to do to get in the hunt for an at large.....
(02-15-2016 11:04 AM)HP-TBDPITL Wrote: The committee doesn't use RPI as their guide for whether teams get in or not.
The committee uses RPI for seeding purposes after the teams are selected...and also have to factor in splitting up conference teams, look at sites, etc...
It really is as simple as that.
This just isn't true. Committee has always looked to RPI when it tries to figure out who you beat and who you lost to. Record vs. top 25 RPI, top 50RPI, and top 100 RPI are consistently talked about by committee in terms of selection. Your actual RPI number is less important though I think you'll find that teams get dropped a seed line or two if their RPI is significantly lower than the rest of their resume implies they should be seeded.
That said they aren't limited by that and can make quality judgments independent of these things (and often when they do it ends up being among their most baffling decisions). I only take cursory looks at any of the other measures for selection purposes because I almost never hear the committee talk about them.
(02-15-2016 11:07 AM)stever20 Wrote: The committee though used the advanced metrics more last year than ever before. Really don't think that's a 1 year thing quite frankly. It's why UConn and Cincy are in a lot better shape than folks think- and why Temple really has a ton of work to do to get in the hunt for an at large.....
I hope this is true, but really did not get that impression last year.
I love how fans most cinci and UCONN fans won't simply acknowledge the flaws behind Kenpom and Sagarin. It's just as flawed of a system as RPI. I understand that you love seeing your team that high and will fight for the ratings until the bitter end. I would think that the committee is smart enough to see the results that temple has accumulated in actual game action, not some silly metric the values how much you win by.
(02-15-2016 11:07 AM)stever20 Wrote: The committee though used the advanced metrics more last year than ever before. Really don't think that's a 1 year thing quite frankly. It's why UConn and Cincy are in a lot better shape than folks think- and why Temple really has a ton of work to do to get in the hunt for an at large.....
I hope this is true, but really did not get that impression last year.
They said as much last year. It's why Temple and Colorado St didn't make the tourney, and teams like I think it was Indiana did.
(02-15-2016 11:40 AM)TempleU22 Wrote: I love how fans most cinci and UCONN fans won't simply acknowledge the flaws behind Kenpom and Sagarin. It's just as flawed of a system as RPI. I understand that you love seeing your team that high and will fight for the ratings until the bitter end. I would think that the committee is smart enough to see the results that temple has accumulated in actual game action, not some silly metric the values how much you win by.
Kenpom is why you guys missed the tourney last year when your RPI was in the 30's. KP was in the 70's though- and thus you were NIT.
.gif)
