Skip navigation
 

UPDATED: Calculating March Madness

CMU student crunches the numbers on NCAA tournament Final Four

Contact: Curt Smith


​​​​Final Four, second chances

If your bracket didn't survive the first weekend of March Madness, you weren't alone. A total of 10 double-digit seeds earned wins, a record number in the tournament's history. But in case you're looking for opportunities in the second- or third- chance bracket, Elizabeth Jackson has run an analysis on the remaining teams.

"As a basketball fan, the first two rounds of the tournament were incredibly exciting and entertaining but as a statistician, they were stressful and surprising," Jackson said. "It didn't take long for the dream of the perfect bracket to be shattered."

Several games have made history, including a historic 15-seed Middle Tennessee beating a second-seeded Michigan State University. Also notable, says Jackson, was the astonishing comeback of Texas A&M over Northern Iowa. Statistically, the chances of that game ending the way it did were one in 3,000. In addition, a 10-seeded Syracuse has made it to a formidable Final Four.​​

Jackson's analysis of the Final Four includes:

  • North Carolina has a team with a lot of experience, talent and depth. In her initial analysis and again in her second-chance bracket, Jackson has chosen UNC to come out on top.
  • Syracuse's defense is something to keep an eye on, said Jackson. The Orange have already made it deep into a tournament in which many pundits did not predict, with less than one percent of all brackets picking Syracuse to make it to the Final Four.
  • Though she had originally selected Villanova to fall to Kansas, Jackson admits the Wildcats have played increasingly good offense as the tournament has progressed. Additionally, Villanova's defense also showed up to play against the Jayhawks as they forced 10 first-half turnovers. 
  • Oklahoma's explosive offense, led by Buddy Hield, got them to an early lead in the Elite Eight, and the Ducks were never able to bounce back. However, Jackson thinks that the Wildcats will be more effective at limiting Hield and the Sooner offense than other teams they have met in the tournament.

CMU student crunches the numbers on NCAA tournament​

​​​​​​​​​​​​​​​Central Michigan University actuarial science major Elizabeth Jackson is pairing her knowledge of statistics and love of basketball to calculate what it takes for a team to succeed in the NCAA Championship Tournament.

Jackson, an honors student from Delton, evaluated every tournament fr​om 1985 to 2015 and analyzed many of the variables that affe​ct a team's chances of advancing from round to round. She separated the variables into four different categories, each with multiple c​omponents.

Categories and variables affecting a team's success included:

  • Rankin​g – examples include seed and strength of schedule;
  • Historical – data focused on the amount of success a team has had in previous years when advancing to Final Four and championship games;
  • Season performance – variables focused on statistics such as number of wins and field goal accuracy percentage; and
  • Coaching – the amount of success a team's coach has had in previous tournaments.

"Although some are decent predictors, the difficult part is that there isn't one variable that can accurately predict every game or even 80 percent of games," she said. "My advisor and I had to sort through many statistics trying to find the combination of variables that gave us the highest chance of success."​

A few highlights of Jackson's original findings:

  • Seed is only significant in the first round of the tournament;
  • A team's simple rating system – a ranking that accounts for a team's average point differential and strength of schedule – and total number of wins are significant in predicting whether a team will advance from the first round through the Elite Eight round;​
  • The number of times a team has appeared in the Sweet 16 and Final Four is significant in predicting whether they will make it there again this year; and
  • Variables related to the coach's success in previous tournaments are the strongest predictors in the Final Four and championship games.

Jackson's faculty advisor, CMU professor Felix Famoye, is an internationally renowned statistician and fellow of the American Statistical Association. He and Jackson worked closely together to develo​p the database she used for the analysis of the tournament. Jackson's capstone honors project will be presented at CMU's Student Research and Creative Endeavors Exhibition in April.

The most commonly reviewed variables pundits look at when creating a bracket are seed, simple rating system, rating percentage index, total games played in a season and total wins in a season. Below are graphics showing how certain variables affect a team's chance of winning in different rounds of the tournament. The labels on the vertical axis show the percentage of time that variable correctly selects the winner of a game.​​

Seed

1985-2015 Seed Graph 

Simple Rating System

1985-2015 SRS Graph 

Rating Percentage Index

1985-2015 RPI Graph 

Total Games Played

1985-2015 Total Games Graph 
​​

Total Wins

1985-2015 Total Wins Graph 


Jackson will apply her framework of past games to the upcoming 2016 NCAA men's basketball tournament. We will use her variables and background analysis of previous tournaments to illustrate the math behind the chance each team has to ad​vance. 


Photo Associator

Article Photo Title

Photo Title required.

Photo for News Home

Select File
{{vm.homeFile.fileName}}
Upload
Use This One

Photo for News Feeds

Select File
{{vm.feedFile.fileName}}
Upload
Use This One