Monday, March 2, 2015

Bracket Matrix comparison 3/2

As we get closer to the end, it's a good time to study where I differ from the majority of bracktologists, and see why we differ.  The Bracket Matrix has all the data that I'll cite below.  Let's see where I differ from the consensus and find out why everyone else is an idiot why I might be off on some seedings.

Iowa St and Baylor:  Bracketball 4 seeds, Bracket Matrix 3 seeds
Utah and Notre Dame:  Bracketball 3 seeds, Bracket Matrix 4 seeds

I might go down with the ship on Notre Dame.  They have a horrible non-con SoS that we've seen the committee punish heavily in the past.  But those were usually in situations where the team was a bubble team.  With a team like UND?  I dunno.  Compared to UNC and Louisville, they have less bad losses...and they did beat Duke and Carolina.  They showed well enough against the good teams they did play.  It's my estimation that the selection committee will cite their eye test on this one.

Same for Utah.  I think the Big 12 is slightly hurt by having all the teams beat up on each other, and I just think they're going to be reluctant to put all of Iowa St, Baylor, and Oklahoma on the 3 line.  Here's the thing about the voting system the committee uses:  they vote on several teams at once.  For example, for the 3 seeds, they'll vote for 4 teams from a group of, for example, Maryland, Utah, Notre Dame, Iowa St, Baylor, Oklahoma, and the two MVC schools.  If there's no consensus on who is best between ISU, BU, and OU, then teams like Utah and Maryland and Notre Dame can sneak ahead.

And of course Utah's resume is still sterling.  3 losses to teams on the 2 line, 3 other road losses, no bad losses...if you want to argue ISU and OU and Baylor ahead of Utah, I won't stop you, but I think the dynamic of group voting comes into play here.  And of course, if one of these Big 12 teams really separate, then all bets are off.

Providence:  Bracketball 5 seed, Bracket Matrix 7 seed

Part of this is just fluky math:  Providence has an average seed of 6.41.  Still, I have Providence ahead of Butler and Georgetown, the opposite of the consensus.

It's close, but I'll take Providence.  Comparing against Georgetown...it's razor thin.  GU has the signature win over Nova, checkmark there.  Taking away other games against top-8 teams, Provi is 5-3 against the Top 50 and GU is 4-6.  Provi has the worst losses by far, but swept GU, and also has a better non-con signature win (UND vs. Indiana).  Georgetown has a 6.29 average on the Bracket Matrix.  Basically even.

Butler's average is 5.56.  More to explain here.  Butler's non-con SoS is worse (not that 86 is bad, it's just not 19).  They did beat UNC and GU on a neutral court, but the third best win is @St John's, and I think Provi's N-UND, sweep of GU, and win at Butler(!) is a slightly better collection of wins.  Of course, these two teams play again, so this will solve itself.  But I'm not seeing a lot to separate these 3 Big East teams.  7 is too low a seed...I'm okay with 6, but I'm surprised more don't have them as a 5.

Ohio St:  Bracketball 9 seed, Bracket Matrix 7 seed

And up until the weekend, I had Ohio St on the 10 line.  Non-con SoS of 209, best road/neutral win of Minnesota (only road win inside the top 100), no signature win until Maryland.  I thought 9 was high when I did the seeding - 7 is insane.  Even their 7.97 average is too high.  People are going to look bad on this one.

Cincinnati:  Bracketball 7 seed, Bracket Matrix 11 seed

!!!!!!.  Actually, I might look bad on this one.  Here's my argument:  The resumes fall off a cliff at #24, and I decided to put Cincy up near the top of the next tier of teams.  In the end, a top 25 non-con SoS, their beating of 3 teams around them in the S-Curve (SMU 2x and SDSU), and the road win at NC State gaining steam...I moved them up.  With so many teams around them looking so mediocre (SDSU, Michigan St, Indiana, Xavier, etc....it was the best of some bad choices.

However, all that said, they should not have an average seed of 10.21.  That's aggressively low.  .500 on the road, and the 4 wins over tourney teams (5 depending on how you feel about Temple) is enough to clear immediate bubble danger.

Texas A&M:  Bracketball OUT, Bracket Matrix 10 seed

Stop paying attention to the raw RPI number (33).  STOP.  They swept LSU...and those are their only Top 75 wins.  Yes, the bubble is soft, but it's not that soft.  EVERYONE around them offers significantly better signature wins.  And it's not like A&M has a magic elixir elsewhere in their profile.  Their SoS is middling.  Their bad losses aren't that bad (N-K-State, @Alabama), but not great.  They're 7-6 in road/neutral games.  What's so special outside of a SEC record that's inflated because they only play Kentucky and Arkansas once this year?

A&M and Cincy are ranked right next to each other on the Bracket Matrix.  Cincy has 4 wins better than A&M's best win; they have a better non-con SoS; Cincy has a couple worse losses; and everything else is more or less the same.  I don't see how anyone can have A&M above Cincy.  The signature wins mask the bad losses every time here.

No comments: