Can Razorbacks Upset Bulldogs?
NCAA Week 4 - The Arkansas Razorbacks have finished with the worst conference record in a loaded SEC West division every year since 2017 and have not produced a winning conference record since 2015. Few outside of the most diehard super fans expected head coach Sam Pittman's second season in Fayetteville to produce anything more than a modest bowl berth. My FEI preseason ratings were very pessimistic and projected the Razorbacks to finish the regular season with only 3.1 mean wins against FBS opponents. They have already cleared that threshold in September, with a 4-0 start and a pair of impressive pelts on the wall: a 40-21 gashing of Texas in Week 2 and a 20-10 smothering of Texas A&M this past weekend.
The Razorbacks defense has been particularly suffocating in the second half this season. On 20 non-garbage opponent second-half possessions, Arkansas has surrendered only three touchdowns. Only one of the other 17 drives in that non-garbage second-half set managed to cross midfield, a turnover on downs forced against Rice in Week 1. After tying the game up on the next possession against the Owls, Arkansas has not trailed in the first or second half of a game since. Last weekend, the Razorbacks allowed one 67-yard Isaiah Spiller touchdown run in the third quarter, but surrendered less than 50 yards combined on all other second-half possessions against the Aggies.
My ratings missed the mark to start the season and are still catching up on Arkansas, but the Razorbacks have outperformed the betting markets as well. They were an underdog in their victories over the Longhorns and Aggies, and they covered the spread in wins over Rice and Georgia Southern. They are the only team with four wins against FBS opponents against the spread to date. When we compare expected game margins of victory or defeat against actual game outcomes, Arkansas has outperformed the closing line to date by 13.1 points per game. Through four games, only the Georgia Bulldogs have been better (14.5 points per game above closing line margin). Notably, Arkansas travels to Athens this Saturday—more on that later.
I phase preseason projection out progressively throughout the season, and I adjusted the weights this year to react more aggressively to early-season game results than I had in previous years. This was an informed but speculative adjustment to my methodology based on the anticipated unreliability of 2020 data and its impact on my preseason projection model. FEI ratings for Arkansas and others that have played four FBS opponents have a 43% weight applied to preseason projections this week; that is, 57% of Arkansas' opponent-adjusted team, offense, and defense ratings come from 2021 results to date, and 43% of those ratings represent preseason projections.
This preseason projection phase-out strategy has produced some awkward results. Clemson has done nothing to suggest it belongs near the top 10 after its early-season struggles, but the Tigers had a lofty preseason rating that continues to buoy them in the ratings; they are tumbling, but only down to No. 7 this week. Wake Forest, like Arkansas, is outperforming expectations by a wide margin but has merely cracked the FEI top 25. The rankings do not necessarily look right, but how are they performing?
We are not on a 2007-level trajectory of upsets (yet), but the college football season appears to be more wide open than any other of the College Football Playoff era. Given the general unpredictability to date, I have been satisfied with the weekly ratings output and game projection results so far. Week 4 FEI projections had a solid 12.7 absolute error and record against the spread. For the season, FEI projections have actually picked more game-winners than the closing lines, and that has translated into a 56% record (14-11) against the spread in games in which FEI projects the Vegas underdog to win outright. Projections against the spread for all games have been disappointing (46%), but over/under game totals have been solid (55%).
Finding the balance between ratings adjustments and pulling the right levers and weights on preseason priors will always be a challenge. Arkansas has proven more to date than FEI recognizes this week, but the relative ratings of Arkansas and Georgia are the more critical benchmarks. For their part, the Bulldogs have the most impressive team efficiency metrics by far this season, and they would be a clear FEI No. 1 team this week if they were not also held back somewhat by more modest preseason projection numbers. FEI projects Georgia to win by 18.8 points over Arkansas, consistent with the performance expectations of an elite team against a merely good team.
Is Arkansas a merely good team, or are they poised to be a conference championship contender? The betting markets seem to agree with FEI—the line is Georgia by 18.5. We will see if FEI falls on the right side of the number on Saturday, or if Arkansas will continue to defy expectations.
2021 FEI Ratings (through Week 4)
FEI ratings (FEI) represent the per-possession scoring advantage each team would be expected to have on a neutral field against an average opponent. Offense ratings (OFEI) and defense ratings (DFEI) represent the per-possession scoring advantages for each team unit against an average opponent.
Preseason projected ratings are progressively phased out over the course of the season. Expanded ratings for all teams include overall, offense, defense, and special teams efficiency ratings. Ratings and supporting data are calculated from the results of non-garbage possessions in FBS vs. FBS games.