Milligan model picked Gonzaga as champion, ETSU in Sweet 16

Jeff Birchfield • May 5, 2020 at 10:00 AM

Seth Nicol knew the perfect project when he and other Milligan College students took part in a Zoom-hosted series of meetings about how sports analytics can predict the future.

With the NCAA tournament canceled because of the COVID-19 pandemic, Nicol hosted a session titled, “Who Would Have Won March Madness in 2020?”

Through the leadership of David Campbell, an associate professor of economics, Nicol found the model that worked best to determine the final outcome. With college basketball such a changing dynamic, Nicol used data from the last three seasons, looking at the statistics and teams which ended up winning the national championship.

Through his research, Nicol found Gonzaga, Kansas, Baylor and Duke as the top four teams with the Zags emerging as the likely champion.

“I planned on just picking a winner,” Nicol said. “Since the NCAA never released a bracket, I ranked the teams based on my results. The same four teams kept showing up every single time. Repeatedly, Gonzaga showed up as the team to win it all.”

Campbell pointed out that Nicol faced a challenge with the bracket never set. It didn’t account for the different matchup of styles, which can lead to upsets like No. 16-seed UMBC beating No. 1 Virginia in the 2018 tournament.

“One of his challenges is that head-to-head matters,” Campbell said. “Seth’s project was difficult in that he couldn’t point to who wins specific head-to-head outcomes. He had to take the field as a whole and did the best he could with what he had.”


Nicol was questioned how East Tennessee State ranked during his presentation. The Bucs looked to be a surprise team, appearing to be in prime position to make a Sweet 16 run.

“They placed inside the top 16 most of the time,” Nicol said. “Obviously, a matchup could change that, but they looked to be a Sweet 16 team consistently. You do get a bracket picked subjectively and ETSU could be a Sweet 16-caliber team, but end up with a 12 seed and have to play a really good team in the round of 32. It’s disappointing we didn’t get to see that play out.”


Nicol used raw per game statistics with the major emphasis on points per game, points per game allowed, turnovers versus assists, personal fouls and field-goal percentage.

Gonzaga benefitted from playing in the West Coast Conference instead of one of the Power 6 leagues. They put up “magnificent raw statistics” according to Nicol. Still, the Bulldogs are a proven commodity, playing in the 2017 national championship game against North Carolina.


Since there is such a changing dynamic with college basketball, Nicol looked at only national championship teams from the last three seasons as part of his research. It worked well since the three teams had such different styles of play.

Defending champion Virginia emphasized lock-down defense, while 2018 champion Villanova featured a high-powered offense. North Carolina, which defeated Gonzaga 71-65 in the 2017 title game, had a combination of both.

“I ran a regression built off the 2019 tournament,” Nicol said. “I built a model on the last three teams which ultimately got the win. The reason I only went three years back is because the people who were seniors in 2020 were freshman then. I didn’t want to go further back because the game is constantly changing.”


Teams from the Big Ten and Southeastern Conferences fared poorly using Nicol’s rankings. He noted the Big Ten hasn’t had a national champion since Michigan State in 2000, although current league member Maryland won the 2002 title as a member of the ACC.

“Point differential showed the most on teams that shot up to the top in my model,” Nicol said. “That’s why a notable missing conference was the Big Ten. I know that’s a conference that normally has Sweet 16 teams in the tournament, although they’ve not had a winner in a long time. The Big Ten is so deep they play a lot of close games, so you could argue the model didn’t give them a fair chance.”

Kentucky was the only SEC team projected to go deep into the tournament. Nicol was a little surprised that Auburn placed in the middle of the pack since the Tigers were 25-6 entering the SEC tournament.

Alabama, whose SEC tournament game with Tennessee was canceled an hour before tip-off, was projected to be among the final at-large teams, along with Richmond, to make the tournament.

Nicole said the model shouldn’t be taken as fact.

“The thing I learned in building this model is there is no way to truly predict a winner,” Nicol said. “If we could use the econometrics to predict a winner, we would have done it by now and have a perfect bracket by now. Obviously, we don’t.”

The site administrator has disabled comments for this story.