BI Trends

The Metrics of March Madness: In Search of the Perfect Bracket

By Jen Senwoo
Share on LinkedIn Tweet about this on Twitter Share on Facebook

About a year ago, just before March Madness began, 11-time NBA all-star and Basketball Hall of Famer Charles Barkley declared that “analytics is crap.” He said the NBA is all about talent, and that proponents of analytics are “a bunch of guys who have never played the game and they never got the girls in high school.”

Be that as it may, Charles may be alone in this line of thinking, as the NCAA has been leaning on the strength of basketball data now more than ever. Fans and coaches alike have become so dependent on player stats, game performance, and team traction, that some say March Madness has become “America’s most popular exercise in statistical reasoning.

Remember that movie ‘Moneyball?’ Well top NBA teams are now scouting talent based on college statistics, extending the “moneyball” concept beyond baseball and firmly into the NCAA. Duke University has installed SportsVU cameras to better track player movements and deliver NBA-quality stats. ESPN has famously said, “Sooner or later, big data [would come] to college hoops.”

It’s true, being somewhat of a data scientist within the madness of March Madness can get you much closer to a winning bracket, and a big payout for those who have stakes in which team is left dancing after the final game. But the metrics of March Madness is about more than just rankings and player stats. It also takes into account coaches, mascots and even fan sentiment on every social media channel known to man.

Of course, picking a nearly perfect March Madness bracket rarely happens. In fact, Warren Buffett offered a $1 billion prize two years ago to anyone who picked all the winners perfectly (no one did). This year, he lowered the bar, offering Berkshire Hathaway employees $1 million a year for life if their bracket was perfect just up until the Sweet 16.

Kevin Ota, an ESPN spokesman, has said that, “The number of people who get all of the picks correct is astoundingly low.”

That doesn’t stop data scientists from devising some sort of formula to get to a winning bracket. For them, the challenge isn’t simply about predicting the winners, but rather to reward the confidence made by those predictions. People are now predicting the winning percentage based on their level of confidence in that team. This means taking into account superfluous data, like how many players on the teams were All-Americans in high school, or looking at the players that often hit their stride later in the season.

Social sentiment also plays a factor. If a player gets injured, fans will generally chatter about the impact of the loss. And sentiment analyzers often take into account this negative collective feedback and weave it into the prediction, impacting people’s choices for who they think will make it to the final dance.

At the end of the day, you can slice and dice the data in a million different ways to come up with your perfect bracket. Experts will tell you one thing. Analysts will tell you another. But truly making sense of all of this data (and ultimately creating the perfect bracket) is still a confounding mystery.

Your best bet for now? Use the knowledge that you already know. Or if you’re like me, just guess because half the time that gets you closer than anything else!

How else is big data being leveraged across organizations today? Download our “What’s the Big Deal about Big Data?” eBook.


Originally published March 31, 2016; updated on August 9th, 2017

About the Author

Jen Senwoo is the Director of Marketing Demand Generation at Logi Analytics, where she is responsible for developing content as well as creating and measuring integrated marketing/sales campaigns to support lead generation and opportunity goals for the organization. She has previously held marketing positions at American University, BroadSoft, and Chevy Chase Bank. Jen holds a Bachelor’s degree from the Robert H Smith School of Business at University of Maryland College Park.