Student, professor use sports analytics to discover NCAA ranking patterns
Does conference size impact conference rankings in NCAA men’s basketball?
According to research and analysis by one Cornell College student, it does.
Drew Klocke ’18 worked with Cornell College Professor of Math Tyler Skorczewski on the research. Klocke recently finished and presented the work at the Midwest Sports Analytics Meeting at Central College in Pella, Iowa.
“We explored how traditional win-loss rankings differed from other ranking systems in conferences of varying sizes,” Klocke said. “Based on our findings, it would appear that conferences where their teams play an unbalanced schedule—conferences with more than 10 teams that are not playing every other team an equal amount—are more likely to be inaccurately ranked based on win-loss ranks than conferences that have balanced schedules—teams that play every other team an equal amount of times.”
Professor Skorczewski said he’s often inspired to get students working on projects when he’s simply watching television.
“My inspiration for these projects comes from being a fan,” Professor Skorczewski said. “When I am watching games I often find myself asking questions the production hasn’t answered. Then, I wonder if I could answer it myself with data publicly available. Sometimes I can, and if it is interesting I keep going and ask if a student wants to work on it with me.”
This particular topic surfaced while Professor Skorczewski was watching the NCAA conference basketball tournaments in March of last year.
So, the professor and his student set out to better understand the issue using statistics and numbers to map it all out by focusing on collecting data, discussing how to compare the data, and figuring out the best ranking methods.
“We were curious if teams were seeded unfairly for those tournaments based on wins and losses since each team could not play every other team in their conference the same number of times,” Klocke said. “I did some preliminary research for a project in my linear algebra course in Block 7 last year. Then, I spent the summer completing the data collection and doing some analysis.”
“Drew’s results show that as conferences increase to 13 teams or more, an individual team’s rank with the traditional win-loss scheme can be over four places different from alternative schemes,” Professor Skorczewski said. “This is enough to have a significant impact when determining tournament participation or seeding.”
Klocke said he enjoyed analyzing the sports statistics and hopes to get the ball rolling on more projects like this in the future.
“The one thing that I have taken away from this project is that just because an idea or concept may seem obvious does not necessarily mean that there is not research that can be done to better understand the concept and potentially even show that some changes to that concept may be beneficial,” Klocke said.