*Updated: March 16, 2014*

Last year, Florida Gulf Coast upset #2 Georgetown and #7 San Diego State to become the first #15 seed to advance to the Sweet Sixteen since the NCAA Tournament expanded to 64 teams in 1985. During that time, 115 other teams failed to survive the first weekend of tournament play. In fact, only six other #15 seeds have won their first NCAA tournament game. Florida Gulf Coast was the not the only Cinderella story last year. #13 LaSalle and #14 Harvard were two of eight teams with double-digit seeds to win their first game and Wichita State became the first #9 seed to advance to the Final Four. Was 2013 the most upset-crazy year in the tournment's history (at least since 1985). According to the numbers, Yes.

The following graph shows the sum of the seeds for all of the winning teams, starting with the round of 64 (the Tuesday/Wednesday games that have been added in recent years are not included). For example, Florida Gulf Coast, a 15 seed, won twice last year so they would contribute 30 (15 x 2) to the total. #1 seed Louisville won the six games necessary to take home the title so they contribute 6 (1 x 6) to the total. Though not perfect, I think that summing the winning seeds is a pretty good indicator of the performance of the underdogs from year to year. The blue line is the sum of all winning seeds. The red line is a three-year moving average of the sum of winning seeds.

Based on a Sum of Seeds total of 341, the 2013 tournament was the most upset-minded (upsetty? upsetacular?) in the 29 years of 64 team brackets. Coming in a close second, was 2011 with a Sum of Seeds total of 335. The average Sum of Seeds total is 298 so clearly we've seen an uptick in upsets in the past four years. The big question is: will this trend continue? The moving average line suggests that upsets have been cyclical. From 1993-1996, the Sum of Seeds total was in the 270s, well below the average. From 1999-2002, we had above average upset activity (316/year) before a long period without a lot of big upsets (Sum of Seeds averaged 279 between 2003 and 2009 with only one year above 300). I suspect that we will continue to see more upsets than we've seen in the past though maybe not to the extent that we did in 2013. Here are a few possible reasons:

First, the mid-major conferences (leagues not among the power conferences like the Big Ten and ACC nor among the small conferences that rarely win an NCAA tournament game) seem to be catching up to the power conferences and playing tougher non-conference schedules than ever before. Playing teams from the power conferences in the regular season not only gives the mid-majors a taste for the type of tougher competition they will face in the NCAA Tournament but may also build confidence if they win or are highly-competitive in those games.

Second, star college basketball players continue to take the "one and done" approach before heading to the NBA so mini-dynasties are rare. When the tournament expanded to 64 teams in 1985, high school basketball standouts were not going directly to the NBA and college basketball stars often played through their senior seasons. This was good news for the top seeds who were loaded with talent and experience. The top programs still have little problem recruiting talent, but rarely can they mix that talent with experience (Think about how dominant Kevin Durant would have been had he played three or four years at Texas). High school to NBA and "one and done" has been increasing since the mid-1990s so it is hard to explain the relatively uspet-free period between 2003 and 2009 but generally speaking, the lure of the NBA should result in a higher number of upsets.

Another factor is the expansion to a 68 team tournament with four games feeding into the 64-team brackets. This improves the quality of the teams seeded 12 and higher. In the past, the four weakest automatic qualifiers would play the #1 seeds. Now, those four teams play each other, with two being knocked out before the Round of 64. This creates a domino effect in the 12 to 16 seeds. Two teams that would have been 15 seeds before are now 16 seeds, two teams that would have been 14 seeds are now 15 seeds, and so on. This means that the top four seeds are facing slightly tougher competition. Four at-large bid teams are also required to play a "First Round" game in order to advance to the Round of 64. The Tournament Selection Committee has some discretion when it comes to seeding these teams so it can create some pretty tough matchups for some top four seeds. Last year, LaSalle (a 21-win team with a fairly solid RPI ranking of 46) won their First Round game to grab a 13 seed (they then beat Kansas State). The year before, BYU (23-8, RPI ranking of 45) became the toughest 14 seed in tournament history after beating Iona to advance to the 64. Even this small expansion should make the early round games more competitive. In fact, I am a proponent of expanding to 70 or 72 teams. For nearly 30 years, the #1 seeds have essentially had a bye in the first round (116-0, only seven games decided by six points or less). Expansion to 70 or 72 teams would make things at least a little tougher for the top seeds.

**Actual Number of Upsets by Year**

The Sum of Winning Seeds Analysis described above gives us a good look at the overall performance of the weaker seeds on an annual basis but it doesn't really tell us how many uspets we've seen. For example, if the #12 and #13 seeds in the same region win in the Round of 64, they play each other in the Round of 32. If the #13 beats the #12, it shouldn't really be considered an upset. To count the number of upsets, I first needed to define an upset. I settled on the following criteria:

- Any victory by the weaker seed when the seed differential is 5 or more. Teams with a seed disadvantage of 5 (for example, #11 vs #6 or #7 vs #2) win 29% of the time. The numbers are similar for seed disadvantages of 6 and 7 then decrease as seed differential increases. These are certainly notable upsets.
- Any victory by #5 seed over a #1 seed (underdogs wins 18% of the time) or a #6 over a #2 (underdog wins 24%).
- Any victory by a #4 seed over a #1 seed (#4 seeds win only 31% of these matchups).

The following table shows the number of upsets, based on this criteria, per year.

Using this method, the most upsets (13) occurred in the years 1985, 2011 and 2013. The least (3) in 2007. For the most part, this method and the Sum of Winning Seeds are in synch.

NCAA Trends Part II

ESPN College Hoops

CBS Sports College Hoops

StatFox.com

StatSheet.com