The science of picking NCAA Tournament basketball games — bracketology — is not a new field of study, but one Central Michigan senior has taken the science a step further.

Delton senior Elizabeth Jackson, an actuarial science major, is receiving national attention for her honors capstone research project, focused on forming the best mathematical NCAA Tournament bracket.

Jackson's formula, which includes over 60 variables based on 30 years of past tournaments, is unlike most other formulas. Jackson said her study changes the variables of the formula as the tournament progresses from round to round, taking new variables into heavier consideration.

Her work has been featured by USA Today article and The Washington Post asked for a copy of her bracket.

"It's definitely nerve racking knowing everyone is watching after (my research) came out," Jackson said. "Everyone wants to know my picks, so if my picks are wrong, it definitely puts a lot more pressure on."

Jackson is a sports fan who plans to work in the insurance industry as an actuary. She said the project was the perfect way for Jackson to apply her area of study to her hobby — sports statistics.

"I grew up in a household that loves sports. My parents and brothers were huge sports fans," she said. "When I had to do my capstone project, I knew I wanted to do something with sports statistics."

With help from professor Felix Famoye, a statistics professor in the College of Science and Engineering, Jackson proposed the project last semester. She began the extensive process of gathering tournament history data from online sources during winter break.

"The number crunching I liked," she said. "Gathering data was stressful. It was hard to find because it's not like all the data can be found in one place, especially finding all the RPI rankings from 30 years ago. (That) could be a bit of a struggle, but luckily I found some good and consistent data sites that gave me a lot of information."

Although her bracket isn't doing as well as she hoped, the experience has garnered more interest than she expected. She predicted 21 of the 32 first round games correctly and nine of her Sweet 16 teams remain after the second round.

"I never really expected (my work) to get this much attention and fame, but it's been really, really great," Jackson said. "It's been really cool to combine something I love and use it academically and also something that can help other people and other people are interested in."

The formula and results

Jackson estimated she spent more than 100 hours since December analyzing every team and round of every NCAA Tournament game from the past 30 years, Jackson concluded which variables were the most significant as the tournament progressed.

The four main variable categories include:

Ranking – Seeding and strength of schedule.
Historical – Team's success in previous years when advancing to Final Four and championship games.
Season performance – Number of wins and field goal accuracy percentage.
Coaching – A team's coach's success in previous tournaments.

Jackson found a team's seeding is typically only relevant in predicting games in the first round of the tournament.

To find which teams will advance to the Elite Eight, the Simple Rating System — which considers teams' average point differential and strength of schedule — and total wins become important variables. As for the Final Four, the most important variable becomes the head coach's previous tournament success.

"Something that's very predictive (for the Sweet 16) is the number of Final Fours a team has made it to in the past, as well as the number of Final Fours the coach has made it to," Jackson said. "The (team's) total number of wins as well as the SRS rating, those were a few of the strongest variables and predictors for the Sweet 16."

Jackson's model gives each team a probability of the likelihood of advancing to the next round. She picks the teams with the higher probability.

But the formula doesn't always pick the higher seeded team. Jackson said her model predicted multiple first-round upsets, including No. 11 Gonzaga over No. 6 Seton Hall and No. 10 Virginia Commonwealth over No. 7 Oregon State.

Her model did, however, show No. 11 Michigan as more likely to beat No. 6 Notre Dame, but the Wolverines lost 70-63.

"That was a very close pick," Jackson said. "I was sad to see it didn't come out quite how I planned, but it was definitely interesting to see that."

As for the second-seeded Spartans loss to No. 15 Middle Tennessee State, Jackson's research showed the Blue Raiders had less than a five percent chance of winning the game. She had the Spartans beating No. 1 Virginia to advance to the Final Four.

"It was pretty incredible," she said. "They actually showed it as the third-biggest upset that's every happened in the tournament. Although Michigan State, they say, is probably the best team that's ever lost in the first round."

No. 15 Norfolk State's 86-84 upset win over No. 2 Missouri in 2012 — which Jackson said had a less than one percent chance of happening — is the biggest NCAA Tournament upset of all time.

Jackson's model didn't predict upsets like No. 13 Hawaii's upset of No. 4 California and No. 12 Yale's upset of No. 5 Baylor.

"I was initially surprised by it, but looking back it seems the Ivy League schools have done well in the tournament, historically," she said.

The first round took its toll on Jackson's bracket, but she still has three-out-of-four teams remaining in her Final Four after two rounds.

"It seemed with a lot of these games, either team could have won — there was no clear favorite," she said. "My bracket was really on the downside of a lot of those (close games) — such as the USC vs. Providence game and the Texas vs. Northern Illinois game — some of the games that came down to the last shot."

Jackson plans to publish her work toward the end of the spring semester.

"I've just been focusing on the data and, right now, it's actually my time to sit back, relax and watch (the tournament) unfold," she said.