Researchers explore accuracy of NCAA men's basketball tournament seeding
(Phys.org) —As the annual NCAA men's basketball tournament—commonly known as "March Madness"—ramps up, fans wonder if their team will be included in the tournament and, if so, where they will be seeded.
Some teams will automatically make "The Big Dance" thanks to winning their conference tournament. Of the 64 teams in the NCAA tournament, 31 will be conference champions and the other 33 "at large" teams are chosen by a selection committee made up of university and conference administrators.
It is the same selection committee that chooses how the tournament teams are ranked (seeded) which determines who plays who as the teams advance to the "Final Four."
The committee has never revealed how they decide the team seeding or if there is even an established seeding procedure.
Exploring the selection process
Management information systems professors Bruce Reinig and Jim Lackritz of San Diego State University and information systems and operations management professor Ira Horowitz of the University of Florida researched the seeding of the teams based on their conferences to determine if mid-major conference teams were given unfavorable seedings over major conference teams.
They also wanted to see how the seeding of the mid-major teams impacted the point spreads and betting lines for gamblers.
Each year, the tournament is parsed out in brackets where 64 teams are divided by four and lumped in the Mid-West, West, East and South regional tournaments. The team seeded first in each regional tournament plays the 16th seed, the second seed plays the 15th seed, etc., until the four champions of each regional tournament emerge to play for the national championship.
The researchers wanted to see if the seeding was fair to teams from the nine mid-major conferences (such as the Missouri Valley, the Mountain West and the Atlantic 10 conferences) compared to teams from the six major conferences (such as the Pac-12, the Big 10 and the Atlantic Coast Conferences).
They also wanted to see how the seeding of the mid-major conference teams impacted gamblers for both the point spread (defined as the point deferential between the winning and losing teams which determines the payout) and the betting line (defined as how much a gambler needs to bet on each team to win $100).
The three professors gathered the data for the tournament from the years 2000 – 2012 and computed how many of the four number one seeds, four number two seeds and so on from each year (13 years X four numerical seeds for 52 teams) were distributed to teams from the majors, mid-majors and other conferences (such as the Ivy League or Ohio Valley Conference).
The results were published in The Journal of Gambling Business and Economics. Of the 52 number one seeds between 2000 and 2012, 48 were distributed to teams from major conferences, four were from mid-majors and none were from other conferences.
In fact, of those 832 teams that participated during the time frame in question, 420 hailed from the majors, 212 from mid-majors and 200 were from other conferences. Additionally, of the 443 teams awarded at-large bids, 342 were from the majors, 95 were from the mid-majors and only 6 were from other conferences.
Upon examining final scores, the researchers concluded that when a mid-major was conferred with a higher seed, they tended to not live up to expectations, but when given a lower seed than their early opponents, they tended to exceed expectations. "Since most mid-majors were seeded at the middle or lower-middle part of the bracket, they tended to surpass the committee's expectations," Lackritz said.
This was also true of the betting market when gamblers bet on the opponents of lower-seeded mid-majors. The betting line moved, on average, more than one point for a game between a higher seeded major versus a lower seeded mid-major. Additionally, the researchers concluded that the spread became more difficult to predict with each passing round of the tournament.
"The conclusions indicated that the committee selecting the at-large bids and then seeding the tournament statistically tended to overwhelmingly favor teams from major conferences and, judging from the games' outcomes, the mid-majors tended to perform better than expected when placed at a lower seed," Lackritz observed.
"This also accounts for frequent wide-ranging differences when bettors gamble on the spread or the betting line for the same game."