Way to go RPI!!!! ( sarcasm )

[dcZONAfan]

Soon to be 6 loss Kansas can do no wrong in the RPI....

no worries, they are 9th in kenpom and will be ranked outside the top 10 come next week.

 

[Jack_John_Mark]

All I know is which ever one ranks my team the highest is definitely doing it right

 

[element1286]

It's 25/50/25 (W%/OP W%/ OP OP W%)

You're trying to turn the RPI into something it's not. If you want to measure relative team strength, there are already dozens of rating systems that do that. That's not what the RPI is for.

The RPI measures resumes. A resume is Who you played, where you played, and who won. Thats it.

Your resume is where you worked, what position and how long you worked there. That's it. It isn't how many TPS reports you filed while you were working for Inatech. It isn't the number of inter-office memos you sent while you were at Intertrode. It's where you worked, what position, and how long. If that stuff is good enough they'll bring you in for an interview and you can tell them all about your TPS reports.

If your teams resume is good enough, then the committee will take a closer look and you can tell them all about how you kicked the crap out of Duke by 25. But that isn't what goes on a resume.

So you are suggesting that a 'resume' based on MOV, SOS, wins/losses, home/road/neutral doesn't exist?

I know what it is measuring, it doesn't want MOV bias, but I still don't know how that makes it the best 'resume' measurement. There is nothing supporting that assertion.

 

[TrollyMcTroller]

So you are suggesting that a 'resume' based on MOV, SOS, wins/losses, home/road/neutral doesn't exist?

I know what it is measuring, it doesn't want MOV bias, but I still don't know how that makes it the best 'resume' measurement. There is nothing supporting that assertion.

I'm saying that MoV doesn't belong on a resume.

A resume isn't every detail of your life's work. A resume is the highligts. The bullet points.

It isn't any different from a job resume. It's just a quick picture to give the decision makers something to go on to see if they need to dig deeper. You're making the "But if you guys would just bring me in for an interview, you'd see that I'm a great worker and..." excuse. That isn't the way the world works, and that's not the way the selection committee works.

If your gripe is that the selection committee doesn't look at MoV, then you'd be wrong. They have access to that information and plenty more. They have access to pretty much every metric imaginable in the war room. And when necessary they'll use it. They'll dig deeper. They'll look to see how much you beat teams by. They'll look to see if players were injured. They'll look at anything of consequence.

But I've yet to hear a compelling argument as to why MoV should be included in the RPI. Pretty much the only argument that I've heard boils down to "but then it wouldn't deviate so much from all the other metrics." Which quite frankly is a stupid reason. If you want something like the other metrics, then just use the other metrics. It's like complaining that a McDouble doesn't taste as good as Chick-Fil-A sandwich, so McDonals should try and make their burgers more like Chick-Fil-A sandwiches. If you want a Chick-Fil-A sandwich, go get one at Chick-Fil-A. Plenty of people like the McDouble, and like only paying a buck for it. They're two different things, they serve two different purposes, for two different groups. Arguing that they should converge into one thing just show a lack of understanding about what either of them are targeted towards.

That said, in the context of what the tournament Selection committee has said they are looking for in a resume, the RPI is the best that I'm aware of as it's the only thing that actually measures what the selection committee has said they are looking for on SS. You might be able to argue that Sagarin's Elo_Chess is better, but since he won't release his formula, or go into much detail about his methodology, it's difficult to say if it's better or worse. The RPI was put together with input from the selection committee, and I think they have a pretty good idea, what they are looking for.

 

[element1286]

A resume isn't every detail of your life's work. A resume is the highligts. The bullet points.

What? Add a bullet point then, this is a meaningless definition.

But I've yet to hear a compelling argument as to why MoV should be included in the RPI.

You would never just look through winning percentages of various teams and pick the highest ones as the best teams. There is the SoS component included through the use of Opponents Win% and Opponents Opponents Win% but how can we determine those are the most accurate representations of the SoS? Why would a multiplication of select winning percentages be accurate if the original value 'winning percentage' can't be relied upon?

 

[TrollyMcTroller]

What? Add a bullet point then, this is a meaningless definition.

Fine wise ass. Then you add in MoV, and let us know how how the results look.

You would never just look through winning percentages of various teams and pick the highest ones as the best teams. There is the SoS component included through the use of Opponents Win% and Opponents Opponents Win% but how can we determine those are the most accurate representations of the SoS? Why would a multiplication of select winning percentages be accurate if the original value 'winning percentage' can't be relied upon?

Great argument. Then why stop at MoV? Why not add tempo free stats? Why not throw in offensive and defensive efficiency too? How bout' rebounding margin? Nothing says "tournament team" like a +10 on the boards. You can add stats all day long, and get a better representation of how good the team is. And then you end up with just another Sagarin or KenPom or Massey, or any of the other dozens of systems out there designed to evaluate teams.

The RPI does what it is designed to do and it does it very well. I think it could use some tweaking as some of the "we don't want you tinkering with your schedule to manipulate the RPI" measures have proven counter productive, but beyond that, it appears to be serving its purpose just fine. Just keep in mind, if the committee wanted to get a revision to the RPI, all they have to do is say the word and it's done. The fact that it's changed very little since inception should tell you something.

I understand your complaint. More detail is better than less detail. More accurate is better than less accurate. In general that's typically the case, but when something is designed specifically to look at a handfull of metrics, because that is what they want measured, then adding other things simply goes outside of its intended purpose.

As stated earlier, its just a tool used by the committee. It isn't set in stone. It doesn't have to be the most accurate thing in the world. That's why there is a committee. To actually pour over the data and decide based upon all the evidence. If this was the BCS, then I would completely agree with you. Throw in every metric you can think of to paint the most accurate picture possible. But this isn't the BCS. It's college basketball, and the committee is smart enough to include MoV when they think it is relevant.

 

[Jack_John_Mark]

If you want a real edumacation go to kenpom.com and pay the $19.95 for the 1 year subscription.

You'll understand exactly why each team is succeeding the way they are.

If a player eats a candy bar 5 minutes before the game you know. It's pretty ridiculous how far it goes.

 

[TrollyMcTroller]

If you want a real edumacation go to kenpom.com and pay the $19.95 for the 1 year subscription.

You'll understand exactly why each team is succeeding the way they are.

If a player eats a candy bar 5 minutes before the game you know. It's pretty ridiculous how far it goes.

I know all about KenPom. I think KenPom is great. But KenPom is evaluating teams not resumes. It's a subtle but important distinction.

 

[Jack_John_Mark]

I know all about KenPom. I think KenPom is great. But KenPom is evaluating teams not resumes. It's a subtle but important distinction.

Part of his formula is evaluating quality and non-quality wins and losses and how they come at home vs on the road.

Now, of course, the way he determines "quality" is basically as you say.

 

[element1286]

Fine wise ass. Then you add in MoV, and let us know how how the results look.

Great argument. Then why stop at MoV? Why not add tempo free stats? Why not throw in offensive and defensive efficiency too? How bout' rebounding margin? Nothing says "tournament team" like a +10 on the boards. You can add stats all day long, and get a better representation of how good the team is. And then you end up with just another Sagarin or KenPom or Massey, or any of the other dozens of systems out there designed to evaluate teams.

You can add anything you want, I don't see how any of those are a measure of how a team has done. I'm sure rebounding margin correlates to wins, but I doubt the relationship is completely independent, and probably built into any other measure of efficiency. I'd rather they just use one of the already created metrics.

The RPI does what it is designed to do and it does it very well. I think it could use some tweaking as some of the "we don't want you tinkering with your schedule to manipulate the RPI" measures have proven counter productive, but beyond that, it appears to be serving its purpose just fine. Just keep in mind, if the committee wanted to get a revision to the RPI, all they have to do is say the word and it's done. The fact that it's changed very little since inception should tell you something.

I understand your complaint. More detail is better than less detail. More accurate is better than less accurate. In general that's typically the case, but when something is designed specifically to look at a handfull of metrics, because that is what they want measured, then adding other things simply goes outside of its intended purpose.

Agree to disagree on the RPI, it seems arbitrary to me. For example, is there a theoretical basis for 25/50/25 weighting, or did they just pick them because they are nice round numbers? Same with the home/away weights, just nice round numbers.

As stated earlier, its just a tool used by the committee. It isn't set in stone. It doesn't have to be the most accurate thing in the world. That's why there is a committee. To actually pour over the data and decide based upon all the evidence. If this was the BCS, then I would completely agree with you. Throw in every metric you can think of to paint the most accurate picture possible. But this isn't the BCS. It's college basketball, and the committee is smart enough to include MoV when they think it is relevant.

Fair enough, I'm sure they know about the other stats, I just don't see why the stats can't do the MOV mitigation and the committee mitigate injuries and the really close bubble teams.

 

[TrollyMcTroller]

You can add anything you want, I don't see how any of those are a measure of how a team has done. I'm sure rebounding margin correlates to wins, but I doubt the relationship is completely independent, and probably built into any other measure of efficiency. I'd rather they just use one of the already created metrics.

My point was that adding those types of stats just turns RPI into another KenPon/Sagarin/Massey clone.

Agree to disagree on the RPI, it seems arbitrary to me. For example, is there a theoretical basis for 25/50/25 weighting, or did they just pick them because they are nice round numbers? Same with the home/away weights, just nice round numbers.

I'm not aware of any statistical basis for the weighting. I think it's what the committee deemed was "appropriate" My understanding on the 25/50/25 weighting is that the devalued the team's SOS and inflated their opponent's SOS in order to dissuade teams from scheduling cupcakes and running up a gawdy record.

As mentioned above, I think that the home/away weighting is terrible.

Fair enough, I'm sure they know about the other stats, I just don't see why the stats can't do the MOV mitigation and the committee mitigate injuries and the really close bubble teams.

I think it falls into a weird category. It's a stat that can help the accuracy of a rating, but at the same time MoV can be misleading. I think it's something the selection committee wants to examine in more detail than a formula can provide.

Not all "10 point wins" are the same. Sometimes a team was dominant for the entire game, and by the time the walk-on brigade was done, the gap had narrowed in a contest that was never in question. It could have gone down to the wire, and the winner pulled away in the final 35 seconds, just by hitting free throws. It could have been a 10 point game since half time. Those are three very different situations that aren't decipherable from MoV alone.

 

[ericd7633]

It's not really all that great, though...if this was LAST year's big east they would have a great schedule, but other than KU IOWA and 'Cuse (the first two in the same tournament) and Creighton in conference their schedule has been hideous. I guess we can call SJU good after the win against Creighton, but that's still a stretch...

Nova has the 15th ranked SOS. That is a really really good schedule. Cuse has the 68th, but they have a tough stretch the last month. By the end I'm sure it will be about even.

 

[dcZONAfan]

Nova has the 15th ranked SOS. That is a really really good schedule. Cuse has the 68th, but they have a tough stretch the last month. By the end I'm sure it will be about even.

Look at their schedule, not their schedule rank. It's not that good, and it's because the Big East isn't very good this year.

It's better than ours because they got to play Kansas and Iowa in that tournament, so I'm not saying that, but @ Syracuse was their only true good OOC road game. And they got smoked.

 

[ericd7633]

Look at their schedule, not their schedule rank. It's not that good, and it's because the Big East isn't very good this year.

It's better than ours because they got to play Kansas and Iowa in that tournament, so I'm not saying that, but @ Syracuse was their only true good OOC road game. And they got smoked.

I did look at it and IMO Nova's is better at the current moment. Not to mention they played a significantly harder OOC schedule, and yeah the Big East isn't that good, not going to disagree with you there, but it's not as though the ACC is much better. Only 5 teams are in the top 50 of the RPI and there are 5 teams below 100 in the RPI. It's better than the big east at the top but also has worse teams at the bottom.

 

[element1286]

I'm not aware of any statistical basis for the weighting. I think it's what the committee deemed was "appropriate" My understanding on the 25/50/25 weighting is that the devalued the team's SOS and inflated their opponent's SOS in order to dissuade teams from scheduling cupcakes and running up a gawdy record.

As mentioned above, I think that the home/away weighting is terrible.

Yeah, I kind of figured, but it's another reason I don't like RPI, it's a made up measure. The fact they can just manipulate the formula because their formula was being manipulated by others doesn't inspire confidence to me.

I think it falls into a weird category. It's a stat that can help the accuracy of a rating, but at the same time MoV can be misleading. I think it's something the selection committee wants to examine in more detail than a formula can provide.

Not all "10 point wins" are the same. Sometimes a team was dominant for the entire game, and by the time the walk-on brigade was done, the gap had narrowed in a contest that was never in question. It could have gone down to the wire, and the winner pulled away in the final 35 seconds, just by hitting free throws. It could have been a 10 point game since half time. Those are three very different situations that aren't decipherable from MoV alone.

That's fair, I don't disagree, and it's impossible to be completely accurate, but I'd rather be as accurate as possible, which is really the genesis of my argument.