skip to main content

Jacobson's BracketOdds Research Takes the Madness Out of March

3/3/2021 4:16:29 PM Aaron Seidlitz, Illinois CS

Illinois CS professor Sheldon Jacobson has worked with students since 2007 to implement mathematical and computational research that offset the unpredictability of March Madness, resulting in probabilistic analysis findings and his BracketOdds website.

Sheldon H. Jacobson
Sheldon H. Jacobson

This year, Jacobson and his students addressed the very essence of why the NCAA Men’s Basketball Tournament, perhaps more than any other in American sport, provides a sense of excitement and unpredictability.

The first full round of play includes 64 teams squaring off over a few days on national television. Several of these contests take place at once, and, typically, include a few wild finishes or upsets.

Since so many people fill out their brackets prior to the start of the tournament, much of the uncertainty lies in selecting that first round. There are many lesser known teams, creating more chaos and wreaking havoc on people’s bracket selections.

Jacobson’s group focused on developing two new bracket simulators, which prove why it’s most beneficial to begin selecting winners with either the Final Four or the Elite Eight round.

“The genesis of the work behind these simulators goes back four years. It culminated last year with a published article through the American Statistical Association,” Jacobson said. “We did not publish the simulators on the BracketOdds website last year, though, because the COVID-19 pandemic hit in the United States prior to the tournament, halting all college basketball activity.

“It’ll be great to populate the site this year with the correct teams and seeds just before the start of the tournament. It’s then great fun for our students when pandemonium occurs and the site is swarmed by thousands of college basketball fans for the next 84-96 hours.”

In total, Jacobson’s student research groups over the years have produced six published papers on bracketology. Topics covered range from the breakdown of seeding in the tournament to the most recent piece regarding the methodology to select a pool of high scoring brackets.

In 2012, they launched the BracketOdds website, taking their academic information public for the first time.

One person who has been with Jacobson throughout the last few years, guiding progress behind the new bracket simulators, is fourth year PhD student Ian Ludden.

Ludden’s research falls within the department’s Theory and Algorithms focus area. Jacobson serves as his adviser, assisting him develop other research projects – like one focused on political redistricting.

However, Ludden said the BracketOdds work has been his most fulfilling research experience thus far.

Together with Jacobson and Doug King, professor with Industrial & Enterprise Systems Engineering, Ludden honed a theory about how best to pick your bracket into a research paper published in the Journal of Quantitative Analysis in Sports.

Ian Ludden
Ian Ludden

“For me, this project is one of the things that first attracted me to Professor Jacobson’s work, because it’s clear he emphasizes work with practical impact. At the same time, though, it’s based on a rich theoretical foundation, which I love because I also love math,” Ludden said. “It’s great to see the two new simulators come to fruition this year, because I’ve been working with really talented undergraduate students Kaahan Motwani and Andrea Roy. It’s exciting to see all the work they’ve done progressing toward the beginning of the tournament in a few weeks.”

They also simplified this concept into a table on the site, which clearly shows what they believe will be the true benefit of picking the Elite Eight or Final Four first.

This table shows that by picking just eight teams as your Elite Eight, the picker fixes the result of 16 total games. If the picker chooses the Final Four members first, that creates fixed results for 12 total games.

That means starting with picks in either round gives the selector eight “free games” or already predetermined results when filling the bracket out backwards from that round.

To Jacobson, the great joy in this work comes from viewing the engagement levels between students.

Over the years, this group has rotated between several students, but their experience has been similar. The inspiration is clear that they are working on a project that proves the value of computing by influencing one of the nation’s most passionately followed sporting events.

“Our approach to BracketOdds represents the overarching goal of research at Illinois CS well. As I tell the students, we’re not here to follow hot topics. We’re here to create hot topics. People follow us, we don’t follow them. That’s part of our culture,” Jacobson said. “Projects like this are why I became an academic. It’s an opportunity to instill excitement and dedication in computing for a number of young students who love the work.”