Clearing Up March Madness with Predictive Analytics

Bookmark and Share

For those whose brackets took a blow after perhaps just the first day of the 2016 NCAA men’s basketball championships, this year’s tournament may provide a case in point on the value of analytics over sentiment and gut feelings.

After just the first day, the New York Times described the tournament as "one upset after another."

To showcase the value of predictive analytics and hard data over intuition or hunches, prior to the start of the 2016 championship, once again, the SAP Data Viz team used SAP Lumira and SAP Predictive Analytics to analyze data from the top 68 teams in order to fill out a bracket and determine which ones will reach the Final Four. 

The Data Viz and Predictive Analytics team relied on publicly available data from multiple sources such as, Wikipedia, GPS Visualizer, and key statistics from sources such as the College Basketball Power Index, Defensive/Offensive Quotient, Strength of Schedule deviation, and other stats. Each matchup was evaluated on a team by team basis. You can check out how your bracket is faring against the SAP team’s predictions here.

And, in a recent FiveThirtyEight blog post, Nate Silver and Jay Boice offered their 2016 forecast for March Madness. Silver, a statistician and author of “The Signal and the Noise: Why So Many Predictions Fail - but Some Don't,” and Boice, a computational journalist, relied on what they describe as “relatively simple information” including final scores from games, home court advantage and the location of games, as well as a team’s conference, as well as other factors.

Still while predictive analytics can provide greater insight, at least in this situation the risk of a bad decision is relatively low, and it’s good to remember that anything can happen and “that’s why they play the games.” Or, to switch sports and quote the wisdom of baseball great Yogi Berra, “it ain’t over til it’s over.”