What's new
By:
USCHO Fan Forum

This is a sample guest message. Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

  • The USCHO Fan Forum has migrated to a new plaform, xenForo. Most of the function of the forum should work in familiar ways. Please note that you can switch between light and dark modes by clicking on the gear icon in the upper right of the main menu bar. We are hoping that this new platform will prove to be faster and more reliable. Please feel free to explore its features.

Bracketology 1.1 - before the conference tournaments. Don't be upset

Fishman'81 said:
Salient post. Nice job.

I didn't realize that the RPI had a top-20 "cliff". I hate cliffs.

All stats should be calculated along a continuum to be considered valid.
Click to expand...

It used to be worse, one of the Pairwise comparisons used to be record against TUC, with TUC arbitrarily defined as teams with an RPI of .5 or better. This QWB has a pretty small impact, comparatively.

Edit: for comparison’s sake, the highest QWB in D1 is .0084 added to the RPI, not very large.

r
 
Last edited:
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

Fishman'81 said:
Salient post. Nice job.

I didn't realize that the RPI had a top-20 "cliff". I hate cliffs.
Click to expand...

It is not a cliff, it is a slope - maximum bonus for beating number 1 taper to 0 bonus for beating number 21. It's not a step function - the bonus varies by the RPI ranking of the teams involved. I can see that. What I would like to see an official description of the procedures for calculating.
 
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

cetihcra said:
It used to be worse, one of the Pairwise comparisons used to be record against TUC, with TUC arbitrarily defined as teams with an RPI of .5 or better. This QWB has a pretty small impact, comparatively.

Edit: for comparison’s sake, the highest QWB in D1 is .0084 added to the RPI, not very large.

r
Click to expand...

I agree.

After looking at the possible impact of the current RPI, the effect is fairly minimal v. the old (and awful) TUC in D-1, which was easily the steepest and most illogical criterion ever employed in college hockey, to my knowledge... (Well, aside from the utterly un-accountable "RNK" in D-3.)

Our game is slowly making much more sense.
 
Last edited:
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

NUProf said:
It is not a cliff, it is a slope - maximum bonus for beating number 1 taper to 0 bonus for beating number 21. It's not a step function - the bonus varies by the RPI ranking of the teams involved. I can see that. What I would like to see an official description of the procedures for calculating.
Click to expand...

Isn't that a cliff, if the bonus points are arbitrarily rescinded at #20?
 
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

Fishman'81 said:
Isn't that a cliff, if the bonus points are arbitrarily rescinded at #20?
Click to expand...

Not a cliff because it's like the bottom of an outrun on a ski slope. Slides down to a flat bottom. A "cliff" is like what we call a step function. f(x) = 1 if x ≤ 20, f(x) = 0 if x > 20. This is g(x) = 0.05 - 0.0025*(21 - x) if x ≤ 20, and 0 if x > 20. It reduces to zero instead of drops off to zero.
 
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

NUProf said:
Not a cliff because it's like the bottom of an outrun on a ski slope. Slides down to a flat bottom. A "cliff" is like what we call a step function. f(x) = 1 if x ≤ 20, f(x) = 0 if x > 20. This is g(x) = 0.05 - 0.0025*(21 - x) if x ≤ 20, and 0 if x > 20. It reduces to zero instead of drops off to zero.
Click to expand...

Well, it's a cliff in the sense that there's an arbitrary cut-off at #20. Why not #37, or #82? That defies logic.
 
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

Russell Jaslow said:
This.

Football does not make the NCAA much money. Basically, all the NCAA gets out of D1 football is a "sanctioning" fee each bowl pays the NCAA for the right to operate a post season game.
Click to expand...

The NCAA pulled in $761 million from the 2017 NCAA tournament. That number is set to rise to $869 million this year. The NCAA also generated $129.4 million in ticket revenue and $60 million in marketing rights for the 2017 fiscal year. The NCAA's expenses were $956 million.
 
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

Fishman'81 said:
Well, it's a cliff in the sense that there's an arbitrary cut-off at #20. Why not #37, or #82? That defies logic.
Click to expand...

A cliff is by definition a step function - this is a tapered reduction.
 
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

sacole13 said:
The NCAA pulled in $761 million from the 2017 NCAA tournament. That number is set to rise to $869 million this year. The NCAA also generated $129.4 million in ticket revenue and $60 million in marketing rights for the 2017 fiscal year. The NCAA's expenses were $956 million.
Click to expand...

Which tournament are you taking about? Basketball or football?
 
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

Why should beating a bad team hurt your RPI?

It doesn't. The RPI adjusts for it.

But the "quality win" thing is a complete gimmick.

The #20 v. #21 distinction is indeed a cliff, all semantics aside.
 
Last edited:
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

One of the things to remember is that this current PWR formula has been arrived at in reaction to situations that have happened in D-I in previous years. Each time an undesired result happened or came close to happening, formulas were adjusted in response. The QWB replaced the TUC. The TUC used to add AQs, which really made it volatile. And the PWR also used to give a boost to last 16 games. Those all changed because of undesirable possible outcomes.

It’s transparent, but it’s also fudged and manipulated from season to season. Perhaps taking the previous D-III criteria and applying them by the book as a PairWise would be the best. We actually did that at USCHO until the powerbrokers of D-III started to play games and it became meaningless to publish.

In any case, getting rid of the smoke-filled rooms and the gaming of the criteria that went on is a great step forward.


Sent from my iPhone using Tapatalk
 
Ed Trefzger said:
One of the things to remember is that this current PWR formula has been arrived at in reaction to situations that have happened in D-I in previous years. Each time an undesired result

In any case, getting rid of the smoke-filled rooms and the gaming of the criteria that went on is a great step forward


Sent from my iPhone using Tapatalk
Click to expand...
It’s going to be interesting to watch the PW rankings change in real time . To get an idea of what could happen , tomorrow’s 1pm semi final game, Amherst/Wels could shift numbers in the top 10,,, so compare the RPI numbers before and after that game,,, should give us a little idea anyway. Logic behind that thought is Wes with a win could jump NEC and NE thus creating quality win points to go up or down accordingly to those who have beat them. Could be an indicator anyway,,,, I believe when all is said and done PW teams 6 though 13 could all move around
 
Re: Bracketology 1.1 - before the conference tournaments. Don't be upset

I was just nosing around on the NCAA site. I went to the bracket and it shows a spot for a third place game.
 
You must log in or register to reply here.
Back
Top