Home > Bracketology, Current Updates, S-Curve, Team News > Quick Takes: Composite rankings and the Seed List

Quick Takes: Composite rankings and the Seed List

Dave2015This past week, the NCAA Men’s Basketball Selection Committee hinted that it may consider integrating a wider range of metrics into its selection and seeding process.  Although the Ratings Percentage Index (RPI) has been the NCAAs primary tool for organizing teams, committee members have been free to utilize other metrics for several years – so this isn’t totally new territory.  Additional metrics include KenPom (Ken Pomeroy), Sagarin, and BPI ratings, among others.  Using various analytics, the ratings are designed to provide an objective overview about how teams perform beyond wins/losses, team records, etc.  You can decide for yourself how well the various ratings work.  We all have our preferences.  And the goal here isn’t to discuss the merits and pitfalls of each.

Interestingly, the Selection Committee also hinted that it could end up integrating a type of composite score (rating) for teams based upon the metrics selected for inclusion.  How that might look (and be calculated) remains to be seen.  But in the interest of college hoops fans (and a certain bracket nerd), I took the liberty of developing a composite for the teams at the very top of our seed list, and for some along the proverbial bubble.  Let’s take a look at how the process worked, along with some interesting side notes.  For this story, the composite rankings include the RPI, NPI (Nolan Power Index – Warren Nolen), ELO Chess, KenPom, and Sagarin.

Here’s the Top 10 on our Seed List – in order for Jan. 16, 2017 – with composite ranking in parenthesis:

  1. Villanova (2.8)
  2. Kansas (4.4)
  3. Baylor (8.2)
  4. UCLA (10.6)
  5. Kentucky (3.4)
  6. Gonzaga (4.0)
  7. Florida State (14.8)
  8. Creighton (11.8)
  9. North Carolina (8.4)
  10. Louisville (10.4)

If we re-ordered these same teams based on composite rankings alone (only), the Seed List would look like this …

  1. Villanova
  2. Kentucky
  3. Gonzaga
  4. Kansas
  5. Baylor
  6. North Carolina
  7. UCLA
  8. Louisville
  9. Creighton
  10. Florida State

In the interest of time, I didn’t go beyond these ten to see what other teams may have moved into the Top 10.  It was designed to be an exercise.  But there were a couple of interesting takeaways …

  • While UCLA ranks lower in the RPI, KP, and Sagarin, the Bruins are much higher in the NPI and ELO Chess models. UCLA also owns a win at Kentucky, arguably the best (or one of the best) road wins to date.  Their only loss is by two points against a fully healthy Oregon team on a last-second shot in Eugene; and the Bruins are 8-1 away from home overall.  The composite ranking places Kentucky firmly ahead of UCLA even though the Bruins won at Rupp Arena.  I would imagine this type of situation would create a discussion among Selection Committee members were those two teams under discussion for say a cutline spot (1/2 seed) on the final seed list.  Is UCLA a better team simply because the Bruins won a single game against UK?  No. Not any more than Indiana State is better than Butler because they beat the Bulldogs during a single game in Terre Haute.
  • If you factor in the math, Villanova’s composite score lead (2.8 to 3.4) is actually larger than it looks.  There’s little question that Villanova would be the overall No. 1 seed if the final bracket were released today.
  • Florida State ranks surprising low across the board (double figures in all metrics) despite having a profile that could easily be top-five material.  If Duke had FSU’s profile, most of us would probably have the Blue Devils as a No. 1 seed.
  • Meanwhile, the metrics and composite love Gonzaga. That’s not a knock on the Bulldogs.  They are an excellent team and could rightfully end up in the No. 1 seed discussion. But based on “results,” Gonzaga has beaten four teams currently in the field, FSU has beaten six.  Both have beaten Florida.

As a wrap-up, here are a few notes related to teams along the cutline; those that would today be on the bubble.  We used the same five metrics to compile scores of 16 teams; either currently in or under serious consideration.  Composite score in parenthesis.

  • Miami-FL (38.8) ranked highest (lowest score) among the 16.  This despite an RPI of 68 (this morning) and only one win (at Pittsburgh) over an at-large team in the field.
  • Kansas State (40.0) was second despite having no wins against a team in the Top 100 of the RPI (for what it’s worth).  K-State’s best win is either Texas or Oklahoma.
  • Wichita State (40.2) is third and the Shockers have the largest disparity among the metrics (83 RPI vs. 24 Sagarin).  WSU’s best wins are Tulsa, LSU, and Oklahoma.
  • VCU (42.0) ranks significantly better than Illinois (62.8).  Metrics other than the RPI do not appreciate the Illini’s work.  The sidebar of course, is that Illinois beat VCU by 18 points on a neutral court.  One day, one game.  But something, in and of itself, less relative to metrics than perhaps humans.

Hopefully, you found this interesting.  Metrics have their place and can certainly help organize teams and offer perspective among closely grouped teams beyond wins, losses, scores, and game location.  But there should always be a place for subjectivity and results.  Who you played, who you beat, and where you beat them are important – at least to me.  It would be unfortunate if computer rankings alone determined inclusion and layout for the 68-team bracket.

I’ll do this exercise again at some point and try to include relative composite notes in future Quick Takes, Seed List, and Bracket updates.  Thanks for your interest.


  1. No comments yet.
  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: