2-1 to start B1G play is very good

[GatoLouco]

This is heavily informed by quality of wins, the unevenness of schedules, nonconference performance, etc. building a resume is the ticket, you can’t just stack up conference win percentages and assume everybody makes the dance in that order and then allocate bids based on conference strength. Everybody always seems to want to do the last of those in particular, and it simply isn’t how it works.

I understand that's not how it works. I know that much. Though conference strength plays a role in the quality of the wins within conference, or it does not?

We have to generalize to simple terms to make guesses, like how many wins it will take, while knowing Nebraska was left out a few years ago with a 13-5 record. Otherwise there are no guesses till pretty much till the last day of play.

 

[PurpleWhiteBoy]

I understand that's not how it works. I know that much. Though conference strength plays a role in the quality of the wins within conference, or it does not?

We have to generalize to simple terms to make guesses, like how many wins it will take, while knowing Nebraska was left out a few years ago with a 13-5 record. Otherwise there are no guesses till pretty much till the last day of play.

NET is pretty secretive. There is no published document to tell us how 2 Quad 2 wins compares to 1 Quad 1 win.

But they start with KenPom ratings to tell them who the best teams are - so you want your conference to do really well in the non-conference. After you start conference play, the conference can't really move up or down relative to other conferences. Individual teams can, of course.

Most of this works against the mid-majors, whose teams don't get many opportunities to earn Quad 1 points ,but have plenty of chances to get Quad 3 or Quad 4 losses.

 

[AdamOnFirst]

Though conference strength plays a role in the quality of the wins within conference, or it does not?

You’ve got it backwards basically. The strength of a conference in based on how good the teams are within it. In tourney terms, we’re most interested in the distribution of Quad 1-4 teams (yes, that’s an oversimplification) and how many teams are tourney quality, roughly top 50 or so. A conference doesn’t offer quality wins because it is a good conference, it is a good conference because it allows quality wins. This may seem pedantic, but it’s important because a tourney team isn’t just trying to rack up a good record in a good conference, they are trying to collect quality wins. It’s the games that matter, not the conference they take place in.

And I don’t think using conference quality and record is a good heuristic at all. The easiest thing to do is go glance at what Joe Lunardi or the other major bracketologists are saying at any given time. Their projections are accurate enough to essentially be a correct projection under current conditions who is going to be in. If you want to go a step deeper, look at the sites that are comparing quad wins and losses vs the other teams on the bubble and look at KenPom.

Looking at conference record is messy though to subtract information from this approach. If the Big Ten has 5 top 50 teams who will make the tournament and 8 teams in the 51-90 range who will miss it, it doesn’t make it a better conference if one team goes from 52 to 48 and makes it a 6 bid league while the other 7 teams all crash to 150+. It also doesn’t make it really a worse league if a tourney bubble team slips and falls just out of the tourney because all the bad teams turned out to be pretty good.

This is a long and pretty lazily articulated rant. Basically, thinking about conference quality determining quality wins is backwards and has a poor signal to noise ratio compared to these other easier methods to track your team’s progress.

 

[PurpleWhiteBoy]

My best guess at what the NET ratings do is as follows...

Take Ken Pom's ratings as is. (or average a few ratings)
Apply some semi-arbitrary "quadrants" to your games (Road game vs Top 50 is "Quad 1" etc)
Apply some semi-arbitrary weights to the Quad 1-4 wins and losses.
Total it up for each team.
Team with most points is #1.

 

[AdamOnFirst]

My best guess at what the NET ratings do is as follows...

Take Ken Pom's ratings as is. (or average a few ratings)
Apply some semi-arbitrary "quadrants" to your games (Road game vs Top 50 is "Quad 1" etc)
Apply some semi-arbitrary weights to the Quad 1-4 wins and losses.
Total it up for each team.
Team with most points is #1.

Probably not, they probably utilize an ELO style formula to determine the underlying rankings and then proceed to do it as you say with the weights based on past fitting to some criteria like overfitting to past tournament brackets, etc.

 

[PurpleWhiteBoy]

Probably not, they probably utilize an ELO style formula to determine the underlying rankings and then proceed to do it as you say with the weights based on past fitting to some criteria like overfitting to past tournament brackets, etc.

Looks like they used to use Ken Pom but recently changed to do their own version of KenPom.
Lets face it, these rankings are quite similar.
There are always simplifying assumptions and limits imposed. The marginal improvements from one system to the next are fleeting. Torvik apparently is using the score of the game at various points, not just the final score.
Its probably better.

Regardless, the NET uses this and it looks pretty arbitrary....

Quadrant 1: Home 1-30, Neutral 1-50, Away 1-75

Quadrant 2: Home 31-75, Neutral 51-100, Away 76-135

Quadrant 3: Home 76-160, Neutral 101-200, Away 135-240

Quadrant 4: Home 161-353, Neutral 201-353, Away 241-353

 

[PURPLECAT88]

Probably not, they probably utilize an ELO style formula to determine the underlying rankings and then proceed to do it as you say with the weights based on past fitting to some criteria like overfitting to past tournament brackets, etc.

ELO style? You mean they’re using Strange Magic?

 

[CappyNU]

ELO style? You mean they’re using Strange Magic?

I think it means that it's more of a Livin' Thing.

 

[PurpleWhiteBoy]

I think it means that it's more of a Livin' Thing.

Thats why I Can't Get it Out of My Head.

 

[Purple Pile Driver]

ELO style? You mean they’re using Strange Magic?

Don’t bring me down 88.

 

[ricko654321]

I understand that's not how it works. I know that much. Though conference strength plays a role in the quality of the wins within conference, or it does not?

We have to generalize to simple terms to make guesses, like how many wins it will take, while knowing Nebraska was left out a few years ago with a 13-5 record. Otherwise there are no guesses till pretty much till the last day of play.

In most years I think a 10-10 B1G record will get a team in that has a decent OOC performance. Some of those teams quoted as missing with winning conference records had weak OOC slates and/or with multiple bad loss(es).

This year the B1 is weaker, and we lost to Chicago St. But we did get quality wins against Purdue, MSU, Dayton already. As Adam says it's far too early to call, but if I had to guess 11-9 is our target conference record to make the dance. I think 12-8 would be pretty much a shoo-in barring some bizarre path we take to get there (e.g. late season injuries and we totally tank at the end or something).

 

[PurpleWhiteBoy]

This year the B1 is weaker,

Team	2022-23 Rating (End of Conf Tournaments)	2023-24 Rating so far
Purdue	6	1
Maryland	27	109
Indiana	28	95
Michigan St	32	20
Michigan	33	69
Illinois	34	5
Iowa	36	54
Rutgers	38	74
Penn St	40	106
Northwestern	42	50
Ohio St	45	48
Wisconsin	56	12
Nebraska	94	56
Minnesota	204	85

Last year there were 11 Big Ten teams that were very closely bunched. Only 3 teams sat outside the top 50.
This year Wisconsin and Illinois are significantly improved, but there are 7 teams currently outside the top 50.