Iowa Basketball

Iowa Basketball

Post by Kyle T. Jone » Sat, 06 Apr 2013 08:26:26



Quote:



>>> I'm unconvinced.  Iowa has an RPI SOS of 103, with losses to 150+ and
>>> two 100+ RPIs.  They were less than .500 against top 50 teams.  So with
>>> less than 50 at large bids, why should they be in?

>>> .500 conference play isn't enough.

>> The committee is asked to select the *best* teams among those who do not earn an
>> auto-bid, not the ones with the best resumes.  That's the committee's charge,
>> however they end up focusing on resumes.  Using the RPI, RPI SOS, or conference
>> records is not the best way to measure team strength, at least not in this day
>> and age.  The RPI was created a long, long time ago and is not based on sound
>> mathematics.  The weights are arbitrarily chosen, and it doesn't consider
>> schedule strength beyond opponents' opponents.  There are much better formulas
>> out there to measure team strength. Also, I would average the results of several
>> of these formulas, instead of using only one.  I don't have a problem with
>> involving the human element as well, since a lot of formulas do not account for
>> things like injuries or suspensions.  However, if you are going to use a formula
>> -- and the committee clearly does use the RPI to measure team strength if you
>> consider some of the past head scratcher selections -- then use a formula that's
>> has a better track record of measuring team strength.  Hell, even if you believe
>> the committee should select based on resumes, then there are better formulas for
>> doing that than the RPI formula.

> I do not think injuries and suspensions should be considered explicitly.
> Everybody has them. A good system for assessing strength will account
> for them. A post hoc consideration of injuries and suspensions will give
> them unjustified extra weight.

Summing a function that calculates player value and multiples by
expected court time?

Cheers.

Quote:
> Your comments are well taken. A 25 game schedule is simply not enough
> to make much finer distinctions than have been made. They should be
> able to do better from what is being said. Better yet a double elimination
> tournament. [Ducks for cover]

--
Too bad. Read the manual. If this stuff were easy
we would not get the big bucks. -- Michael Press, June 1st 2012
 
 
 

Iowa Basketball

Post by JGibso » Sat, 06 Apr 2013 08:32:52

Quote:





> >>> I'm unconvinced.  Iowa has an RPI SOS of 103, with losses to 150+ and

> >>> two 100+ RPIs.  They were less than .500 against top 50 teams.  So with

> >>> less than 50 at large bids, why should they be in?

> >>> .500 conference play isn't enough.

> >> The committee is asked to select the *best* teams among those who do not earn an

> >> auto-bid, not the ones with the best resumes.  That's the committee's charge,

> >> however they end up focusing on resumes.  Using the RPI, RPI SOS, or conference

> >> records is not the best way to measure team strength, at least not in this day

> >> and age.  The RPI was created a long, long time ago and is not based on sound

> >> mathematics.  The weights are arbitrarily chosen, and it doesn't consider

> >> schedule strength beyond opponents' opponents.  There are much better formulas

> >> out there to measure team strength. Also, I would average the results of several

> >> of these formulas, instead of using only one.  I don't have a problem with

> >> involving the human element as well, since a lot of formulas do not account for

> >> things like injuries or suspensions.  However, if you are going to use a formula

> >> -- and the committee clearly does use the RPI to measure team strength if you

> >> consider some of the past head scratcher selections -- then use a formula that's

> >> has a better track record of measuring team strength.  Hell, even if you believe

> >> the committee should select based on resumes, then there are better formulas for

> >> doing that than the RPI formula.

> > I do not think injuries and suspensions should be considered explicitly.

> > Everybody has them. A good system for assessing strength will account

> > for them. A post hoc consideration of injuries and suspensions will give

> > them unjustified extra weight.

> Summing a function that calculates player value and multiples by

> expected court time?

The BPI does something exactly like that.  The BPI is also frozen prior to the NCAA tournament, so we can see where it ranked teams.  It would have given Iowa a #11 seed.  Not as high in Pomeroy, but still in.  Iowa is now #20 in Pomeroy, but a large part of that is their performance in the NIT.  And I don't remember exactly where Iowa was in Pomeroy pre-tournament.  If you click on the links within BPI, you also see the BPI rating with all the best players present, the BPI against teams with all their best players present and so forth.

 
 
 

Iowa Basketball

Post by Michael Pres » Sat, 06 Apr 2013 17:29:11



Quote:




> >>> I'm unconvinced.  Iowa has an RPI SOS of 103, with losses to 150+ and
> >>> two 100+ RPIs.  They were less than .500 against top 50 teams.  So with
> >>> less than 50 at large bids, why should they be in?

> >>> .500 conference play isn't enough.

> >> The committee is asked to select the *best* teams among those who do not earn an
> >> auto-bid, not the ones with the best resumes.  That's the committee's charge,
> >> however they end up focusing on resumes.  Using the RPI, RPI SOS, or conference
> >> records is not the best way to measure team strength, at least not in this day
> >> and age.  The RPI was created a long, long time ago and is not based on sound
> >> mathematics.  The weights are arbitrarily chosen, and it doesn't consider
> >> schedule strength beyond opponents' opponents.  There are much better formulas
> >> out there to measure team strength. Also, I would average the results of several
> >> of these formulas, instead of using only one.  I don't have a problem with
> >> involving the human element as well, since a lot of formulas do not account for
> >> things like injuries or suspensions.  However, if you are going to use a formula
> >> -- and the committee clearly does use the RPI to measure team strength if you
> >> consider some of the past head scratcher selections -- then use a formula that's
> >> has a better track record of measuring team strength.  Hell, even if you believe
> >> the committee should select based on resumes, then there are better formulas for
> >> doing that than the RPI formula.

> > I do not think injuries and suspensions should be considered explicitly.
> > Everybody has them. A good system for assessing strength will account
> > for them. A post hoc consideration of injuries and suspensions will give
> > them unjustified extra weight.

> Summing a function that calculates player value and multiples by
> expected court time?

Do not know what you are getting at and I was not clear enough.
A system that examines win/loss margins, points per what not
and so on will automatically take into account injuries. Putting
them in explicitly is saying "I know what _would_ have happened."
Well, No you don't. Saying you do turns your model into a sock puppet.

--
Michael Press