The fate of over- and under-seeded teams

I’ve been saying for years that the single greatest determinant of a team’s tourney fate is its seed. If you knew nothing else, you could still fill out a reasonable bracket. But teams don’t always get a seed that reflects their true quality. Some undeserving teams get seeded higher than they should, while other disrespected squads get seeded lower.

We all know this is true, but the fact begs a couple questions: 1) how do we know whether a team is over- or underseeded? and 2) even if we could identify those teams, is the mis-seeding reflected in tourney performance?

To answer the first question, I evaluated the last nine tourneys using KenPom efficiency ratings. I identified two classes of teams:

  • Disrespected teams seeded two through four whose Pythag efficiency values indicated they should’ve gotten a higher seed
  • Undeserving teams seeded one through four whose Pythag efficiency values indicated they should’ve gotten a lower seed.

To answer the second question, I evaluated these two classes of teams by performance against seed expectations (PASE), seed overachievement rate (SOAR) and tourney accomplishments. The results were what I suspected.

Let’s look at the disrespected teams first. These are the two to four seeds whose efficiency numbers suggest that they got jobbed in the seeding.  There were 31 of these teams in the last nine dances. They had an average seed of 3.35, overachieved 45% of the time, and posted an impressive PASE of +.423. More importantly, seven of them made the Final Four and two cut down the nets. Here are the more celebrated teams in this category:

  • Georgia Tech (2004, runner-up) – three seed, #5 in efficiency
  • UConn (2004, championship) two seed, #2 in efficiency
  • Louisville (2005, Final Four) four seed #7 in efficiency
  • Florida (2006, champion) three seed, #8 in efficiency
  • Kentucky (2011, Final Four) four seed, #7 in efficiency
  • Ohio State (2012, Final Four) two seed, #2 in efficiency

The 2004 tourney was a banner year for the disrespected. Both Georgia Tech and UConn deserved better seeds than they got. But the two most criminal mis-seedings were Louisville in 2005 and Kentucky in 2011. By KenPom numbers, both should’ve gotten two seeds.

Of course, not every disrespected team was a success story. The 2006 Kansas Jayhawks got a four seed despite being the second most efficient team in the country. Somebody on the committee must’ve known something, because 13 seed Bradley upset Kansas in the first round. Those Jayhawks built much of their efficiency advantage on defense. They were the top rated team in defensive efficiency and just the 23rd best offensive team—further proof that offense matters in the dance.

So the disrespected teams beat their seed expectations. What about the undeserving teams—those one through four seeds whose efficiency numbers suggested they should’ve been seeded lower? There were 60 teams that fell into this category. Overall, they averaged a 2.55 seed (higher than the disrespected) and only 30% of them beat expectations while posting an underachieving PASE of -.375. What’s more, only five of these squads reached the Final Four and one cut down the nets. Think about it: 23% of the disrespected made the final weekend of the tourney, while just 8% of the undeserving did. Here’s a roll call of the biggest flops:

  • Stanford (2004, round 2 upset victim) one seed, #17 in efficiency
  • Vanderbilt (2008, round 1 upset victim) four seed, #39 in efficiency
  • Vanderbilt (2010 round I upset victim) four seed, #35 in efficiency
  • New Mexico (2010 round 2 upset victim) three seed, #42 in efficiency
  • Duke (2012 round 1 upset victim) two seed, #17 in efficiency

Some of these inequities are patently ridiculous. How does the 42nd most efficient team in the nation grab a three seed? That’s like Baylor or St. Mary’s getting a three seed this year. Uh…no. By all rights, New Mexico should’ve been an 11 seed. Fittingly, the Lobos got “upset” by an actual 11 seed in the second round. Of course, that year Washington had a Pythag value suggesting they were more like an eight seed. So maybe it wasn’t really an upset.

Most people had sniffed out New Mexico as a pretender in 2010; after all, the Lobos are perennial underachievers. But no one foresaw what happened to Duke last year. In hindsight, the Blue Devils were a five seed in disguise—and 15 seed Lehigh was more like a 14. Given those numbers, an upset doesn’t seem quite so shocking.

Some undeserving teams did manage to buck the tendency to underachieve. The most celebrated of them was the 2011 champion. UConn was just the 17th most efficient team that year. They were more like a five seed—and yet Kemba Walker willed them to a title. In addition to having a star with a penchant for pulling out close games, the Huskies also built their efficiency on balance. They had the 17th best offense and the 23rd best defense—so they played both ends of the court with above average quality.

Come Selection Sunday, I’m going to keep a close eye on the disrespected and undeserving seeds. These numbers suggest that it might help me identify the tourney sleepers and pretenders.

This entry was posted in Bracket Tools, Tourney Trends. Bookmark the permalink.

2 Responses to The fate of over- and under-seeded teams

  1. Adam says:

    This is great stuff and an incredible blog for college basketball junkies like myself. Keep up the good work.

    I’m curious to know how you came up with metrics for performance against seed expectations (PASE) and seed overachievement rate (SOAR). Also, how do you define “overachieved” or “underachieved” with respect to this blog post? Do you mean getting to the Final Four/not getting to the Final Four, Elite Eight, etc?

    • ptiernan says:

      Adam – In the right column under “Categories,” pick “Basic Concepts” then scroll down to the bottom of the article list. There’s a primer on PASE and SOAR.

      PASE (Performance Against Seed Expectations) calculates the average degree to which a team, coach or other attribute exceeds seed projected win totals. It’s kind of the “WAR” of NCAA tourney performance analyis.
      SOAR (Seed Overachievement Rate) just calculates the % of times a team, coach or other attribute beats seed expectation in its appearances.

      Hope this–and the primer–help you understand the concept. In the Bracketmaster, you can do queries to your heart’s content and get the PASE values returned. So, for instance, you can ask: “How do teams from the Big Ten perform when they score more than 70 points a game, have an All-American and rely on guards for 50+% of their scoring. The Bracketmaster returns the overall record and the PASE–indicating whether that record reflects over- or underachievement against seed expectations. (In this case, the answer is: not very good. The 17 teams meeting these criteria of -.246 PASE underachievers…essentially 1/4 of a game per dance below projected win totals.)

Leave a Reply

Your email address will not be published. Required fields are marked *

* Copy This Password *

* Type Or Paste Password Here *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>