Determining the odds on football games is not a simple task, but calculating the odds of college football bowl games is even more difficult.
The regular season presents a challenge for numerous reasons, not the least of which is how a season is laid out. Teams largely play mismatched games for the first three or so weeks, making bad teams look decent and decent teams look great. Then they dive into a conference slate with teams that are more compatible and 3-0 or 2-1 starts against lesser opposition turn into losing skids against higher quality foes. Six or seven games can go by before a 'true reading' emerges, and once you think you have a teams' ability nailed down, a key injury occurs and that 'true reading' goes back out the window.
Quite often, a team's 5-1 start and 1-5 finish is merely explained by the schedule itself. Not to pick on Navy and Virginia, but the two participants in this year's Military Bowl are perfect examples of how scheduling can have a direct impact on result.
Navy began the year with 5 straight wins versus teams that finished 26-35, four of which had losing records. The Midshipmen finished the year going 1-6 against seven teams that all became bowl eligible and have a cumulative record of 60-23.
Virginia opened 5-1. They started with a FCS school before facing 5 FBS foes who are now a combined 27-34. The Cavaliers were 1-5 versus a back-half schedule of teams that are now a cumulative 44-27.
It's highly possible that neither Navy or Virginia would have made it to the postseason if just one opponent on each teams' schedule had appeared later in the season.
The Midshipmen were the first to play Lane Kiffin's FAU team and squashed the Owls in the season opener 42-19 at Boca Raton. FAU looked better in losses at Wisconsin and Buffalo, sandwiched around a home win over FCS member Bethune-Cookman, but were just 1-3 through four games. The Owls then ripped off 9 straight wins and won the Conference USA title.
Virginia possibly caught Boise State sulking on its own blue turf in week 4, just two games after the Broncos lost a triple-overtime heartbreaker at Washington State, and a week after sleep-walking through a win over a bad New Mexico team. BSU was 2-2 after 4 games, and wound up 10-3 as champs of the Mountain West.
For the bowl season, several new dimensions can be added to the difficulty of determining who is going to win.
In most cases, the opponents are more evenly matched. A 12-1 Clemson is playing an 11-1 Alabama, a 9-3 Oklahoma State is playing a 9-3 Virginia Tech, and a 6-6 Navy is playing a 6-6 Virginia.
The opponents are from different conferences and there's very little data on which to base a prediction.
For instance, USC and Ohio State is a fantastic matchup in the Cotton Bowl, but only two games were played this season between PAC-12 and Big Ten teams - Washington at Rutgers, and Minnesota at Oregon State. Each conference went 1-1. Ohio State hasn't played a PAC-12 team since the 2015 playoff win over Oregon. USC hasn't played a Big Ten school since the 2015 Holiday Bowl loss to Wisconsin.
In several instances, the head coach is gone. Kevin Sumlin's not at Texas A&M. Jimbo Fisher isn't at Florida State. Willie Taggart's not at Oregon. Arizona State and UCF have their own weirdness added to the situation. Todd Graham has been fired from the Sun Devils but is still coaching in the bowl game. Scott Frost is juggling the tasks of assembling a staff for his new gig (Nebraska), while simultaneously helping to prepare the team he is leaving (UCF) for a Big 6 bowl showdown with Auburn.
There is also the recent craze of players opting out of the bowl game to avoid jeopardizing their NFL draft status. Christian McCaffrey and Leonard Fournette chose that route last year, and yet their respective teams - Stanford and LSU - won their bowl games. Florida State's Derwin James is among the notables taking a pass this year.
Perhaps the Congrove Computer Rankings are as good as any evidence that picking bowl winners is a historically difficult task. Since the beginning of those rankings in 1993, the computer's formula has a .615 winning percentage in bowl games, compared to .750 in the regular season.
Oddly enough, none of these factors appear to have an impact on projecting the outcome of games against the spread. After 25 seasons, the computer's winning clip is .533 in bowl games, the exact same as it is in regular season games.