Do Seed Numbers Predict NCAA Tournament Winners?


Last night, I filled in my brackets for the NCAA Men’s Basketball Tournament. I spent more time trying to decide whether VCU would beat Wichita State in the first round than I did actually watching college basketball this year. But I’m not going to let my ignorance stop me from having a good time. To aide in my research, I did what any narcissistic blogger does: consulted the blog post I wrote last year on this subject.

I decided to update the pie chart I made last year since I still had the STATISTICA spreadsheet and merely needed to add the 2011 data. The first thing I noticed was something I had forgotten from last year – the seed numbers in the Final Four were 3, 4, 8, and 11. Remember that each team in the tournament is ranked with a “seed” from 1 to 16, with 1s supposedly being the “best,” 2s being the second best, and so on. None of the supposedly top 8 teams in the field of 64 teams made it to the last two rounds of games last year. Is the seeding system becoming less accurate?

To visualize this, I averaged the seed numbers of each of the Final Four groupings since 1979, and put this in a Line Plot (in STATISTICA, select Graphs > 2D Graphs > Line Plot. On the Options 1 tab, select the Year variable to display as the Case Label).



The red line shows the linear trend. The line is pretty flat going back to 1979, with the average seed value in the Final Four slightly less than 3. There have been a few anomalies the past few years. In 2008, all of the teams in the Final Four were 1 seeds and last year, the average was 6.5. On average, though, there is no clear trend yet (linear or otherwise) that leads me to believe that I should discard my old data.


So Here is the same chart from last year, updated with 2011 data. The outer part of the pie chart shows the makeup of team seeds in the Final Four from 1979 to 2011. The inner part of the pie chart shows the team winner’s seed number for the same time period.

The chart is labeled with the team seed number and the percentage of total teams at each stage having that seed number. For example, 41% of the teams who have made it to the Final Four from 1979 to 2011 have been #1 seeded teams, 21% were #2 seeded teams, and so on. About ¾ of the teams participating in the Final Four have been a 1, 2, or 3 seed, with the remainder being comprised of 4th seeds or lower (that is, with a seed number of 4 or greater). No teams seeded 10th or lower than 11th have made it to the Final Four.

The middle of the pie chart shows the seed makeup of the tournament winner. Here, the #1 seeds end up winning more than half of the tournaments, and about ¾ of the tournaments are won by 1 or 2 seeds. No #5 seeds or #7 seeds have ever won. The lowest seed – again, meaning “highest seed number” – to ever win the tournament was 8th seed Villanova in 1985.

Keep these data in mind as you pick your winning teams. Don’t choose Purdue for the Final Four or Southern Miss to win unless you want to completely ignore history. You have been warned.

Photo Attribution:


About statsoftsa

StatSoft, Inc. was founded in 1984 and is now one of the largest global providers of analytic software worldwide. StatSoft is also the largest manufacturer of enterprise-wide quality control and improvement software systems in the world, and the only company capable of supporting its QC products worldwide, with wholly owned subsidiaries in all major markets (StatSoft has 23 full-service offices, on all continents), and its software is available in more than 10 languages.

Posted on March 15, 2012, in Uncategorized and tagged , , . Bookmark the permalink. Leave a comment.

Leave a Reply

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

You are commenting using your 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 )

Google+ photo

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

Connecting to %s

%d bloggers like this: