Bracketology and Team Resumes

[IUHoosierJoe]

5 minutes ago, RaceToTheTop said:

I don't think teams should ever move up after a loss (with the exception of wins or losses in other games that day which would effect a shuffling).  

I'm simply not a big fan of efficiency metrics being used as selection criteria.  It's why I'm such a big fan of WAB.  

Let's say that two teams play the exact same schedule -- one goes 22-8 and the other goes 19-11.  The 22-8 team won a lot of close games and were blown out a few times while the 19-11 team lost a lot of close games and blew a few teams out, the 19-11 team likely have a better efficiency rating and potentially a better NET.

At the end of the day if I were to be selecting teams I would only care about wins/losses and who you played (SOS rankings, who you played and where).  It's not that I don't like the predictive measures like Pom and Torvik -- I love them -- but I don't like them being used as selection measures.  And I do believe by the end of the season that wins and losses are better predictive measures of future success than are Pom and Torvik.  I'm sure you are familiar with Bill James' Pythagareon Theorem for predicting wins....but even he said that there does reach a certain point of the season where previous win/loss totals become better predictors of future success than does net runs.  So a team that is coached poorly in late game situations -- like IU is -- those close losses aren't an accident (or, using Pom's term 'luck').

That said, I know each committee member weighs different factors differently.  Just stating my preference and what I would be looking at.

I need to study WAB more, so forgive me if this is a dumb question: How does WAB determine who the bubble teams are?

[Uspshoosier]

3 minutes ago, go iu bb said:

And Purdue found out what IU fans already know from the past few seasons: you can drop in the NET after a win if that win is closer than expected against a bad team.

Purdue stayed at 9 after this game.   

[go iu bb]

1 minute ago, Uspshoosier said:

Purdue stayed at 9 after this game.   

Ah. I saw someone say they dropped a place. Guess they were wrong.

[Uspshoosier]

On 1/28/2025 at 8:32 PM, Uspshoosier said:

https://x.com/TheAndyKatz/status/1884254185543389403
 

a little more on the WAB and NET.   Always nice to get more info from the people the help create it.   If you listen to it you will notice some notes that I say all the time when dealing with the NET, Quads and such 

@IUHoosierJoe

[RaceToTheTop]

4 minutes ago, IUHoosierJoe said:

The bigger question with NET is not necessarily if they win, but if they perform better than the NET expected--i.e., who did they play, where did they play, and what was the score margin.  There is a small portion of result-based metric in the NET, but not a whole lot.

I disagree that it's just a small portion.

Last season Indiana was Michigan State were both 5 games over .500 entering selection Sunday.  IU's NET strength of schedule was 23, Michigan State was 14.  Yet Michigan State had a NET ranking of 24 and Indiana's was 98.  Using just the two results based Matrix results, IU would have been 61 and Michigan State would have been 47 (average values of the two predictive s).  Using the two predictive measures, IU was 97 and Michigan State was 18.

So, in essence, IU's 61 results based and 97 predictive based gave them a NET 98.  Michigan State's result based of 47 and predictive of 18 gave them a NET of 24.  The NCAA hasn't released the formula they use, but you can see that it isn't a little predictive based.....it's predictive driven first.

[IUHoosierJoe]

1 minute ago, RaceToTheTop said:

I disagree that it's just a small portion.

Last season Indiana was Michigan State were both 5 games over .500 entering selection Sunday.  IU's NET strength of schedule was 23, Michigan State was 14.  Yet Michigan State had a NET ranking of 24 and Indiana's was 98.  Using just the two results based Matrix results, IU would have been 61 and Michigan State would have been 47 (average values of the two predictive s).  Using the two predictive measures, IU was 97 and Michigan State was 18.

So, in essence, IU's 61 results based and 97 predictive based gave them a NET 98.  Michigan State's result based of 47 and predictive of 18 gave them a NET of 24.  The NCAA hasn't released the formula they use, but you can see that it isn't a little predictive based.....it's predictive driven first.

Yes, I said the result-based is the small portion.  I can basically reproduce the formula.  The result-based is about 10%.

[LIHoosier]

14 minutes ago, IUHoosierJoe said:

The bigger question with NET is not necessarily if they win, but if they perform better than the NET expected--i.e., who did they play, where did they play, and what was the score margin.  There is a small portion of result-based metric in the NET, but not a whole lot.

I know that, watching the NET daily the past few years, road wins are given a heavy weight even if its within 5 pts and esp. if you win what they consider a Q1 or Q1-a away win.

[Uspshoosier]

1 minute ago, go iu bb said:

Ok. I saw someone say they dropped a place. Guess they were wrong.

I’ve notice once teams get into the top 10 of the NET it is extremely hard to move up unless they play other top 10 teams.   And to move down it would have to take an extremely bad performance to move them down after 1 game.  If they string together a bunch of bad games they can fall

[IUHoosierJoe]

Just now, Uspshoosier said:

I’ve notice once teams get into the top 10 of the NET it is extremely hard to move up unless they play other top 10 teams.   And to move down it would have to take an extremely bad performance to move them down after 1 game.  If they string together a bunch of bad games they can fall

Yes.  It would be nice if the NET published the actual numerical rating values.  The reason for what you noticed is because the rating values for the top 10 teams are naturally a little more spread out.  Once you get down into the pack, the numerical rating values are bunched closer together, so you jump over more teams with, say, a 0.5 increase in your rating value 

[Uspshoosier]

It’s already that time of year to trash the NET?  I thought that didn’t start until mid February.   lol.    It’s the system that’s in place so until they change it that’s what will be used.      End of the day it’s still 12 peoples opinion on what they find most important for this year.     We will get a clue when they announce their top 16 which should be coming soon.   No spoiler here but Auburn will number overall.   12 Q1 already 

[IUHoosierJoe]

4 minutes ago, Uspshoosier said:

It’s already that time of year to trash the NET?  I thought that didn’t start until mid February.   lol.    It’s the system that’s in place so until they change it that’s what will be used.      End of the day it’s still 12 peoples opinion on what they find most important for this year.     We will get a clue when they announce their top 16 which should be coming soon.   No spoiler here but Auburn will number overall.   12 Q1 already 

For those who prefer purely result-based metrics, ESPN Resume/SOR and Colley Matrix (I believe it's colleyrankings.com) are two I look at.

[Uspshoosier]

10 minutes ago, IUHoosierJoe said:

Yes.  It would be nice if the NET published the actual numerical rating values.  The reason for what you noticed is because the rating values for the top 10 teams are naturally a little more spread out.  Once you get down into the pack, the numerical rating values are bunched closer together, so you jump over more teams with, say, a 0.5 increase in your rating value 

Definitely same as KenPoms number value 

[Uspshoosier]

4 minutes ago, IUHoosierJoe said:

For those who prefer purely result-based metrics, ESPN Resume/SOR and Colley Matrix (I believe it's colleyrankings.com) are two I look at.

SoR is on the teamsheets 

[LIHoosier]

7 minutes ago, IUHoosierJoe said:

For those who prefer purely result-based metrics, ESPN Resume/SOR and Colley Matrix (I believe it's colleyrankings.com) are two I look at.

I usually check the Massey composite on Mondays, they have probably 50 or so different rankings of all sorts, basically the rankings p**n version of bracket matrix for the bracketeers.

[Uspshoosier]

Still comes down to Who did you play, where did you play and who did you beat.   
 

A teams NET number isnt as important as your opponents NET number.  Think of it as a sorting tool.   For example Sparty NET was 24 last year that suggests a 6 seed.   Once you look through their resume you clearly see they were not close that.   They beat some tourney quality teams away from home and at home.    They ended up a 9 seed I believe.  IU finished with the same conference record as them however when you looked at their team sheet they only had 2 wins against tourney quality teams and they were both at home 

[RaceToTheTop]

42 minutes ago, IUHoosierJoe said:

Yes, I said the result-based is the small portion.  I can basically reproduce the formula.  The result-based is about 10%.

Gotcha, I misread.

[IUHoosierJoe]

28 minutes ago, LIHoosier said:

I usually check the Massey composite on Mondays, they have probably 50 or so different rankings of all sorts, basically the rankings p**n version of bracket matrix for the bracketeers.

Mine used to be part of that, but I don't have a website anymore, lol.

[Class of '66 Old Fart]

What we face the remainder of our season

		NET @ 2.2.25
@ Wisky	Feb 4	16
Michigan	Feb 8	17
@ Sparty	Feb 11	20
UCLA	Feb 14	28
Puke	Feb 23	9
PSU	Feb 26	53
@ Washington	Mar 1	93
@ Oregon	Mar 4	33
OSU	Mar 8	26
IU @ 02.02.25		65

[Stuhoo]

2 minutes ago, Class of '66 Old Fart said:

What we face the remainder of our season

		NET @ 2.2.25
@ Wisky	Feb 4	16
Michigan	Feb 8	17
@ Sparty	Feb 11	20
UCLA	Feb 14	28
Puke	Feb 23	9
PSU	Feb 26	53
@ Washington	Mar 1	93
@ Oregon	Mar 4	33
OSU	Mar 8	26
IU @ 02.02.25		65

And with a current record of 14-8? Oh that’s not gonna work. If we can go 5-4 in those games we will have very much earned that tourney bid. Pretty damn unlikely tho.

[Uspshoosier]

Hoosiers 

NET-65

SOS-39

Non con SOS-166

Results Based metrics 

KPI- 41

SOR-50 

WAB-52

Predictives

KenPom-55

Torvik-60 

BPI-58

Q1-2-8

Q2-2-0

Q3-6-0

IU is 10-8 in the top 3 Quads which is important that they stay about .500

3 wins away from home with one being a road win at a current projected tourney team 

you would like to see at least one more win away from home.  (Unwritten rule tells you at-large teams usually have 4 wins away from) teams have gotten in with less but it’s not common 

IU has 9 games left in regular season.  7 of those are against teams currently in the field with 4 at home and 3 away.  
 

Other road game is a Q2 game

As I said a while ago 10-10 probably puts them in the tourney but near the Last 4 in range which then you have to worry about bid stealers.  11-9 and they are in the tourney.  Only my opinion but the math currently backs that up. 

most important games left ranked 

Michigan

UCLA

Ohio st

Penn St 

@washington 

Purdue 

@Oregon

@Sparty

@Wisky