It is not uncommon to see at least one Heisman trophy contender from outside the six automatic-qualifying (AQ) BCS conferences. Unfortunately, since the BCS was introduced in 1998, no player from a non-AQ conference has finished higher than third in Heisman voting (Hawaii QB Colt Brennan in 2007) and the last winner from a non-AQ conference was Ty Detmer of BYU in 1990.
Three major hindrances to the Heisman aspirations of non-AQ players are the lack of marquee games, strength-of-schedule and the perceptions about the systems in which they play. We have already looked at how the Heisman race is largely determined by a few key games which rarely feature non-AQ teams. Regarding strength of schedule, the average Sagarin strength-of-schedule ratings for non-AQ Heisman finalists since 1998 is 65.46 compared to 75.27 for Heisman finalists from AQ conferences. Finally, most non-AQ players who end up as Heisman finalists play in systems that are designed produce gaudy numbers regardless of the talent of the player in question and voters take that into consideration .
So, what would it take for a player from a non-AQ conference to win the Heisman? I developed a statistical model to forecast Heisman voting based on a number of factors, including player position, conference, strength of schedule and on-field statistics. It’s clear that, as much as some fans wish it were the case, Heisman voting is not determined by a simple formula and voter behavior is influenced by several factors. The model can be used, however, to capture overall voting trends and estimate what kind of numbers a non-AQ player would have to put up to win the Heisman. Let’s look at a few finalists from recent years and see how much more it would have taken for them to win.
Hawaii QB Colt Brennan finished sixth in 2006 and the was a Heisman finalist when he finished third in 2007. In 2007, he passed for 4,174 yards, 38 TDs and 14 INT, for the 12-0 Rainbow Warriors. From model estimates, Brennan would have needed 5,100 yards, 62 TDs and a passer rating of 197 to beat out Tim Tebow for the Heisman.
Boise State QB Kellen Moore placed seventh in 2009, fourth in 2010 and eighth in 2011. Looking at the 2010 season, Moore passed for 3,506 yards, 33 TDs and 5 INTs for the 11-1 Broncos. Moore would have needed 4,900 yards, 58 TDs and a passer rating of 199 to win the trophy over Cam Newton.
Houston QB Case Keenum finished eighth in 2010 and seventh in 2011, the year he passed for 5,099 yards and 45 TDs with 5 INTs for the 12-1 Cougars. To top Robert Griffin III, Keenum would have needed an eye-popping 5900 yards, 74 TDs and a passer rating of 214.
All of these estimates are based on the candidates’ actual win-loss record and strength-of-schedule from that particular year. From these cases, it’s clear that non-AQ players would have to produce mind-boggling stats to take home the Heisman trophy.
Northern Illinois QB Jordan Lynch (who finished seventh in Heisman voting in 2012) and Fresno State QB Derek Carr have been the only two non-AQ players in the Heisman discussion this year; neither is projected to place in the top five finishers in Heisman voting (currently projected as Jameis Winston, Johnny Manziel, Marcus Mariota, Bryce Petty and AJ McCarron). Assuming all of the top candidates win their remaining games and put up numbers consistent with their season averages, what would Lynch and Carr have to do to win?
Lynch, as a dual threat QB, would have to combine for 8,000 yards of total offense and 77 total touchdowns to take home the trophy, while Carr would have to pass for 5,200 yards, 69 touchdowns and a QB rating of 216 to win. It would take consecutive weeks of record-setting performances for either player to achieve these numbers.
There’s certainly a chance for both of these players to gain a decent amount of Heisman support, but their chances of winning are demonstrably slim.
* * *
The author of this post, Daniel Heard, is a PhD candidate in Statistical Science at Duke University. He has dedicated a significant portion of his research to examining trends in Heisman Trophy voting and developing a model to forecast the voting each year.
You can contact Daniel at firstname.lastname@example.org.