NCAA Men's Basketball Bracketology: Data Viz

Director of Marketing

It’s here again—the time of year that college basketball nuts go a little crazy. March Madness is as unpredictable as it gets. Teams that should win sometimes don’t. And teams that we don’t expect to win sometimes do! That’s the exciting part. They’re the Cinderellas. I can’t wait for the field of 64 to be complete. As I type this, the “play-in” games have yet to be played. So, right now, I have 68 teams to work with, and I’m going to use Oracle Data Visualization to see who I’m favoring in my brackets. Here we go...

First off, I need a data set. I looked all over the internet for metrics that I liked and finally ran across a website called Sports Reference. Here’s their link.

I like this website because it offers the basic stats I need to be able to “see” the situation as well as a very special stat called SRS. Sports Reference defines this as “a rating that takes into account average point differential and strength of schedule. The rating is denominated in points above/below average, where zero is average. Non-Division I games are excluded from the ratings.” I absolutely love this stat because it factors in the strength of the schedule, and, with the diversity that exists in college sports, you must factor this in.

So, I took those metrics from the website and put them in a spreadsheet. I then filtered out all the teams that are not in the dance, and voila! I had my data set.

I then imported the data set into Oracle Data Visualization and decided to focus on three key stats. SRS (explained above), net rating (the delta between the average number of points scored per 100 possessions and the number of points allowed per the opponent’s 100 possessions), and average point differential per game. Pretty straightforward.

Let’s look at SRS first.

I chose a chord diagram because it's an interesting way to see data. These 20 teams represent the best performers with regard to strength of schedule and point differential combined. That's a powerful metric.

Now, let's look at net rating.

As you can see, many of the same teams pop up. This graphic shows the 20 teams that produce the most points per 100 possessions minus how many points their opponents put up per 100 possessions. Also, it’s a very good measure of a team.

Finally, let’s look at average point differential per game.

This is a tag cloud, and it’s a very interesting way to view data because the bigger the text, the bigger the number it represents. For instance, Gonzaga averages a larger point differential per game than Virginia. These 20 teams are the best in the tournament when it comes to this metric.

After I completed these visualizations, I felt a bit confused. I had not yet decided on which metric I liked best. Also, there were several teams that didn’t make all three lists. Therefore, I decided to combine all three into one visualization.

Each team shown here is in the top 20 of all three metrics I like best. Now, here’s a visualization I can get behind! It gives me a great start on filling out my brackets.

However, there is one problem. If you look at the brackets, you’ll see Virginia and Florida can’t possibly both make the sweet 16. Same goes for Wichita State and Kentucky. So, I will have to do some guesswork there. But that’s the name of the game when it comes to March Madness. You have to make some leaps of faith.

As an Oracle employee, I enjoy using Oracle Data Visualization software to put these kinds of visualizations together. In a business context, savvy executives use this technology to improve their strategies with facts and get action around their ideas. The important thing here is that using data visualization can help us see information differently and make better decisions in a way that spreadsheets or traditional static business intelligence cannot.

Also - if you already have Oracle Data Visualization, here is the .dva file for this analysis:

2017 NCAA Bracketology

Password is "NCAA" (without quotes) all uppercase. Enjoy!