Can we predict the winner of the March Madness Tournament?

Sanjeev Lingam-Nattamai
Deep Dives with Data
4 min readMar 7, 2022
Gonzaga Center Chet Holmgren and Forward Drew Timme (Courtesy: Young Kwak/Associated Press)

Madness is coming! The biggest single-elimination collegiate sports tournament (March Madness) starts next Tuesday which features 68 college basketball teams from across the country. It’s an exciting time for many fans like myself who hope that their team can make it far in the tournament and also win their bracket league. During this season, I was interested to find if there was a correlation between a team’s regular season performance and how well they would perform in the March Madness tournament.

I came across KenPom which is a college basketball analytics website created by Ken Pomeroy — a statistician and columnist. KenPom utilizes a ranking system which is calculated with a team efficiency-based model. Here are the top 20 teams based on KenPom rankings from games through March 5th:

Pomeroy uses adjusted efficiency margin (AdjEM) to rank every team. AdjEM is the difference between a team’s adjusted offensive and defensive efficiency (AdjO and AdjD) and measures how much that team will outscore the average college basketball team. For example, from the table above, the Gonzaga Bulldogs are projected to outscore the average college basketball team by a little over 32 points on a neutral court. Raw offensive and defensive efficiency measures the number of points scored or allowed per 100 possessions. A team’s AdjO and AdjD is calculated based on raw efficiencies to factor the strength of a team’s opponents.

I wanted to visualize the numbers on the table better so I built a team tier plot based on a team’s AdjO and AdjD. I used kenpompy which is a Python web scraper for the KenPom website. Here is the result with the 20 teams in the table above:

Can the Zags finally get over the last obstacle?

From the plot, we can see that the Gonzaga Bulldogs have been the best basketball team this season. In fact, no team has dominated a conference like Gonzaga has dominated the West Coast Conference during Mark Few’s tenure as head coach since 1999. While many will argue that Gonzaga ranks highly because they play weaker competition, they also have a .780 winning percentage against teams from the Big 12, Big Ten, Pac-12, Big East, ACC and SEC since 2017. For the second consecutive season, Gonzaga is the pre-tournament favorite according to DraftKings.

However, the March Madness tournament is unforgiving. The single-elimination format means that the best regular-season team doesn’t always win the entire thing. The big question will be if Gonzaga can finally get over the last obstacle and capture the program’s first national title or will they fall short like they did in 2017 and 2021.

Gonzaga head coach Mark Few (Courtesy: Isaac Brekken/Associated Press)

Who will win based on past history?

While the March Madness tournament is full of crazy, jaw-dropping results, one thing largely remains the same. The top seeds historically have the best success in making it far in the tournament. In fact, there have only been two times when a № 1 seed hasn’t reached the Final Four since 1985. Since that same year, more than 80% of Final Four’s contain one or two № 1 seeds. Additionally, a № 2 seed has made the Final Four in 30 of the past 36 tournaments.

Outside of the remarkable tournament championship run by the 2013–14 UConn Huskies, 7 of the past 8 tournament winners were in the top 4 in KenPom ratings before New Year’s day. For this season, the top 4 teams before New Year’s were Baylor, Duke, Purdue, and Gonzaga. Will one of these 4 teams end up being victorious or will another team end the year celebrating in confetti instead? That question will be answered in the championship game on April 4th.

A link to the code for this project can be found here.

--

--

Sanjeev Lingam-Nattamai
Deep Dives with Data

SDE @ AWS | Graduate of Computer Science + Statistics @ Purdue University