Bracketology

[cincrulz11]

Originally Posted by sedziobs

It's interesting that some bracketologists are ignoring the NCAA bracket preview. For instance, Lunardi has Florida instead of Oklahoma as a 4 seed. That doesn't make much sense to me, if the goal is to predict what the selection committee will do. Some have an automated algorithm, or predict results for the rest of the year. But if you're predicting the bracket using current resumes (as Lunardi does), why would you not match the committee's own bracket?

he figured his stuff doesn't really match the committee at the end of the year, so why start now!

[lumberjack]

https://www.aseaofblue.com/2018/2/11...tion-committee

Look who thinks we are a favorable match up. It does not look like this link will work. You can access the article through the site.

Typical

[jacobkdoyle]

Originally Posted by lumberjack

https://www.aseaofblue.com/2018/2/11...tion-committee

Look who thinks we are a favorable match up. It does not look like this link will work. You can access the article through the site.

Typical

Kentucky would be lucky to score 50.

[jacobkdoyle]

Originally Posted by lumberjack

https://www.aseaofblue.com/2018/2/11...tion-committee

Look who thinks we are a favorable match up. It does not look like this link will work. You can access the article through the site.

Typical

I don't think these teams realize what they're up against...until we're up by 20 and they're left reeling and wondering what's happening.

[Jon Rycek]

Knox is their only consistent player. We would dominate them defensively.

[sedziobs]

This Jay Bilas comment is getting a lot of coverage:

A road win over Rider is valued the same as a home win over Villanova. Both are Quadrant 1 wins.

Jay is simply wrong. Two teams being in the same quadrant does not mean they are valued the same. It just means they passed the first sorting algorithm together. RPI Groups are meant to be used as aggregates, not to analyze individual games. Over an entire season, Group records generally are a good unbiased way to initially rank teams. If a team has only Group 1 wins that are of similar quality to @Rider, then there is a problem. But realistically, everyone's Group 1 games will be a mixture of teams in the top, middle and bottom of the Group. Complaining about the extremes is shortsighted. The new system is way better than the old one.

[jacobkdoyle]

Originally Posted by sedziobs

This Jay Bilas comment is getting a lot of coverage:

Jay is simply wrong. Two teams being in the same quadrant does not mean they are valued the same. It just means they passed the first sorting algorithm together. RPI Groups are meant to be used as aggregates, not to analyze individual games. Over an entire season, Group records generally are a good unbiased way to initially rank teams. If a team has only Group 1 wins that are of similar quality to @Rider, then there is a problem. But realistically, everyone's Group 1 games will be a mixture of teams in the top, middle and bottom of the Group. Complaining about the extremes is shortsighted. The new system is way better than the old one.

Lots of that on social media. I agree with you the more I think about it. No matter the cutoff, you'll be able to say "How can Villanova count the same as _______".

[cincrulz11]

Originally Posted by sedziobs

This Jay Bilas comment is getting a lot of coverage:

Jay is simply wrong. Two teams being in the same quadrant does not mean they are valued the same. It just means they passed the first sorting algorithm together. RPI Groups are meant to be used as aggregates, not to analyze individual games. Over an entire season, Group records generally are a good unbiased way to initially rank teams. If a team has only Group 1 wins that are of similar quality to @Rider, then there is a problem. But realistically, everyone's Group 1 games will be a mixture of teams in the top, middle and bottom of the Group. Complaining about the extremes is shortsighted. The new system is way better than the old one.

but to his point, using the rpi for the quadrants is a problem because the rpi is such a bad metric at determining how good your team is. buffalo and rider have no business being quadrant 1 wins.


if you applied the same rules for each quadrant and applied it to kenpom, then i think that tells a much clearer picture of how good your wins are.


his problem is with the rpi.

[sedziobs]

Originally Posted by cincrulz11

his problem is with the rpi.

I understand his main position, and I agree that RPI is vastly inferior to Kenpom. I think he's overreacting and misleading people though. If you replaced Rider with Fresno St, you have the same problem with Kenpom.

[cincrulz11]

Originally Posted by sedziobs

I understand his main position, and I agree that RPI is vastly inferior to Kenpom. I think he's overreacting and misleading people though. If you replaced Rider with Fresno St, you have the same problem with Kenpom.

but didn't the committee show that they kinda just go by the overall group 1 and 2 wins and didn't really dig too far into where within the group they were?