Joe Craig is a huge college football fan – Ohio State, in fact – and a huge fan of numbers and statistics.
So over a couple of beers with some other really smart, statistics minded, college football fans, the assistant economics professor at UCCS and his buddies developed a new mathematic algorithm to create a better college football playoff ranking. And he has the formula to prove it.
Craig said it’s a “general equilibrium modeling system,” after the same design from Nevin Winchester, a principal researcher at MIT that has been used for everything from climate change to rugby and international trade.
You can have a look at the original posting at this link: http://communique.uccs.edu/?p=36621
He assigns rating points based on game scores after adjusting for the strength of the opposing team. From that they measure the wins on a neutral field. And margin of victory is very important. The University of Georgia accumulated 30.40 points in Craig’s system, third overall. Notre Dame had 18.55 points, 12th overall. But as we all know, Notre Dame finished undefeated (12-0) and Georgia finished 11-3 and lost to Alabama in the SEC title game. Notre Dame was picked for the playoff and Georgia was left out. That didn’t sit well with Craig.
“If you look at who they played and how much they beat them by; it really wasn’t that impressive,” said Craig. “How much you win by is extremely important in predicting future success. They say the great teams win close. That’s true but that’s because the team is so good anyway, they can have a terrible night and recover from it, so I think people are missing the causality there.”
Craig’s ranking system had Alabama first overall and Clemson second, same as the College Football Playoff committee, and Oklahoma fourth. After that, the swings were wild in rankings between his system and the actual rankings. The biggest swing is Notre Dame – outside of Central Florida – and Craig believes they committee got it wrong.
“I was talking to someone about how we shouldn’t have Notre Dame in there and they kept saying, ‘but but but…’ and I said, ‘if the goal is to find the best team in college football, we know that Alabama just beat Georgia, so it’s not Georgia, we don’t really know what Notre Dame is?’ I think it differs on what you think is the goal: if the goal is to get the four best teams, they flubbed it this year. I think Notre Dame shouldn’t have been in, I think Georgia should have been in.”
A big part of the system was ranking the teams by how they performed on neutral territory.
“We found that home advantage was about 3.5 points and that’s an average home advantage because Penn State will have a different home advantage than Alabama-Birmingham.”
As the Power 5 Conferences (Big 12, Big 10, Pac 12, SEC, ACC) continue to dominate the conversation for the College Football Playoff and expansion being taken off the table – for now – will Air Force, CSU, Wyoming and other Mountain West teams ever get a shot at the Final Four?
“One of our strongest team this year was Fresno State,” Craig said. “At one point we had them as high as eight and they weren’t getting nearly the press they would have otherwise. What we both learned through out this was no matter what our eyes tell us, trust the data… usually like, ‘no, no, the eye test says….’ but that’s usually when we are wrong.”