CHAPEL HILL, N.C. – North Carolina’s breakthrough season in 2015 has many wondering if the Tar Heels can reach the 10-win plateau in back-to-back seasons for the third time in program history.
The vast void that is college football’s offseason is often filled with a slew of prognostications intent on providing a road map to conference championship games and ultimately the College Football Playoff. Random sportswriter guesses aside, there are plenty of computer databases based on algorithms of various sorts that attempt the all but impossible task of converting past performance into tangible future results.
FootballStudyHall.com’s Bill Connelly relies on his S&P+ ratings to forecast coming seasons, while ESPN leans on its FPI formula and USA Today makes use of Jeff Sagarin’s ratings. Connelly’s S&P+ ratings are based on four statistical factors (explosiveness, efficiency, field position and finishing drives). ESPN’s FPI relies heavily on the expected points added (EPA) statistic, which determines a point value for each play of a game. Sagarin combines three different score-based methods to arrive at his ratings.
Kenneth Massey’s calculations are a bit simpler than most in that he takes each game’s result and adjusts it for the quality of competition. His approach ranked UNC No. 23 nationally after the completion of the 2015 season, combining a top-10 offensive rating (No. 9) with average defensive (No. 62) and strength of schedule (No. 56) ratings.
Massey’s final calculation of UNC’s ranking is in line with the majority of computer models attempting the same task. According to Massey’s composite of 117 ranking systems, UNC’s average rating was 20.95, while its median was nearly identical at 21.0.
The relevance of last season’s final stat crunch lies in the fact those rankings serve as the starting point for projecting the 2016 season. For example, ESPN’s preseason FPI ranks UNC at No. 20 and projects an 8.4-4.0 record with an 8.9 percent chance of winning the ACC. Connelly’s projected S&P+ rating for the Tar Heels is 27th, and he considers the boys in blue a third-tier ACC team with 7.5 projected wins.
The large variety of variables in play – outgoing lettermen, incoming recruits, coaching turnover – for every team makes its impossible for objective computer models to obtain perfection. ESPN acknowledged in December that its 2015 FPI preseason projections were within one win of the actual win totals for roughly a third of FBS teams.
In attempting to reconcile UNC’s 11-win season in 2015 with projections in the 8-win range for 2016, it’s important to understand the difficulty in winning 10 games in a season. Only seven elite programs have managed to average 10 wins over the past decade – Ohio State, Alabama, Oregon, Oklahoma, LSU, TCU and Boise State. Twenty-nine teams won 10 or more games in 2015.
A simple way of explaining the challenge in winning double-digit games is to look at point spreads and the percentages involved with wins and losses. One 12-year study of point spreads and game results – utilizing 9,626 data points - found that a touchdown favorite only won 69 percent of the time. A team favored by a field goal won less than 56 percent of the time. It wasn’t until the point spread reached 17 that the favorite won 90 percent of the time.
This is where Massey’s data of past performance based on adjusting for competition becomes relevant. UNC was 10-0 during the regular season in games in which it was favored, including five as a touchdown favorite or less. In adjusting for the point spread data, the Tar Heels were more likely to go 10-4 than 11-3.
Credit UNC for outworking those odds, and it may take a similar effort this fall to eclipse the 10-win threshold once again.