Week 5 saw the model go 4-2 ATS (against the spread). As predicted last week, Tennessee is officially out of luck. It takes a special kind of coach to try out his 7-3-3 defense on the final play of regulation with the game on the line. With the combined incompetence of Les Miles and Jordan Jefferson, this 5-0 LSU team could easily be 2-3 with loses to UNC, West Virginia and Tennessee, but their defense has kept them undefeated despite their offense’s best intentions. My lock of the week (Alabama) prevailed in an even more one-sided manner than I had imagined, but I feel it was more a function of Florida’s incompetence than Alabama’s dominance. As I feared, the model lumped in Vandy with the SEC big boys in coming up with an estimate for the Connecticut game and they subsequently failed to cover. Auburn exceeded my model’s expectations as the UL-Monroe Indians/Warbirds/Warhawks seemed to determined to turn the ball over and not cover a very generous five touchdown line. Georgia again managed to lose as the favorite, despite the impressive return of wideout AJ Green. For the second straight week, Ole Miss put up some big numbers on the scoreboard, and for the second straight week Kentucky’s defense failed to show up.
What’s new this week
First, I have introduced a new feature: my “not a lock” game of the week. Basically, this is a game where I suspect the model has some fundamental failing, a prediction I don’t trust. In retrospect, I should have introduced this feature last week when I warned that Vanderbilt might not show up against Connecticut. I have also made one important improvement to the model — I now take into account each team’s strength of schedule as measured by their opponents’ cumulative win percentage against all other teams. Although I calculate this strength of schedule metric for each team, I hesitate to use it prior to this week because the values tend not to be nicely distributed.
For example, going into Week 4, most teams have had three opponents who have each played two games against “other” teams. So, the only values we can observe are win percentages of 100 percent, 50 percent or zero percent. While these win percentages are certainly valid, because all of the games in prior years are modeled together, including these early week games would incorrectly weight these strength of schedule metrics because a strength of schedule of say 66 percent means one thing in Week 4 when you’ve only played three opponents, but a completely different thing in Week 10. I’ve grouped each of the 39 variables in my model into one of seven effects:
• Streak (4 variables): home/away win/loss streak
• Poll Ranking (4 variables): home/away AP ranking, AP total votes
• Schedule Strength (2 variables): home/away win cumulative percentage of opponents
• Prior Records (4 variables): Home/Away win pct records from 2008 and 2009 seasons
• Game Location (15 variables): Describing conference play, home/away conferences, in conference game, neutral field controls, and stadium size
• Rushing Per Minute (4 variables): Home/Away Rushing Offense/Defense per minute
• Passing per Minute (4 variables): Home/Away passing offense, defense, passing efficiency
• Special Teams (4 variables): Home/Away punt return and kickoff return yards per kick
The following predictions make use of dozens of carefully selected measures used to describe both the margin of victory and the total points we’re likely to observe in a given game. If one was so inclined, he could use my predictions to place bets. In fact, through the first five weeks of games, this newly revised model is 128-103 against the spread in all lined games and 22-16 against the spread in games involving SEC teams. Just a reminder, the margin of victory is calculated by summing the total “effects” listed above in the charts below.
Arkansas vs. Texas A&M (-5.5, 62) — 2:30 p.m. CDT
Ryan Mallett and Co. take the nation’s No. 2 passing offense (365 yards/game) to meet a 3-1 Texas A&M team that is a late Thursday night field goal away from being undefeated. The poll rankings and strength of schedule both favor the Hogs, though Arkansas’ road win against UGA seems to mean less and less each week. Arkansas gets a game location bump because of how SEC vs. Big 12 matchups have historically gone, though the neutral field game moderates this effect by about 4 points. The Aggies have the Big 12’s No. 3 rushing offense (185 yards/game) and No. 2 passing offense (325 yards/game), but these numbers were largely put up against Stephen F. Austin, Louisiana Tech, and Florida International — not exactly SEC caliber competition. Arkansas mostly managed to contain Alabama’s offense two weeks ago and now leads the SEC in sacks (3.75/game), and is No. 2 in passing defense (157 yards/game), and No. 3 in scoring defense (14.5 points/game). If Arkansas can improve upon their SEC worst turnover margin (-1 per game), the Razorbacks should be able to comfortably handle Texas A&M and their balanced attack. Prediction: Arkansas 32 — Texas A&M 21.
LSU @ Florida (-7, 42.5) — 6:30 p.m. CDT
LSU’s luck will run out when the Tigers visit the Gators on Saturday night. Although LSU showed signs of life on offense last week against Tennessee, the Tigers will not be so fortunate this weekend against a rebounding Florida team. Florida’s No. 9 nationally ranked passing efficiency defense should be able to shutdown LSU’s passing offense (ranked No. 112 nationally at 131 yards/game, and ranked No. 116 nationally at 97.9 efficiency) with their eyes closed (and their hands tied behind their backs in Jefferson starts). How bad is LSU’s passing game? LSU quarterback Jordan Jefferson is ranked No. 13 in the conference, despite their only being 12 teams. For every 7 passes he completes to his own team, he throws one to his opponent. His numbers have been so bad this year that he is not even ranked among the top 100 quarterbacks from 120 FBS schools. LSU’s defense should be able to keep them in this game, however. Despite scoring 31 points per game, Florida’s rushing (151 yards/game) and passing offense (189 yards/game) rank near the bottom of the SEC. This would seem to play right into the hands of an LSU defense giving up 12 points per game (No. 5 national rank), and 79 yards of offense on the ground (No. 6 national rank) and 168 yards through the air (No. 22 national rank). The turnover margin (Florida No. 2 in the SEC vs. LSU No. 9 in the SEC) heavily favors the Gators. If LSU’s defensive line can get penetration and hurry Florida QB John Brantley, LSU could win another tight game. Neither I nor the model is convinced that will happen though, so look for LSU to lose but cover the spread. Prediction: Florida 26 – LSU 21.
Tennessee @ Georgia (-11, 48) — 11:21 a.m. CDT
Georgia may be the best one-win team in the country this year. I thought the same thing last week. Unfortunately for the Bulldogs they played well enough to win at Colorado but still managed to lose. This week I’m 50 percent certain that they will rebound and beat Tennessee. Yes, Tennessee played LSU close, but they needed a +4 turnover margin to do it as they allowed a feeble Tiger offense to put up 400+ yards on 69 plays. Georgia’s offense and defense are just slightly better than Tennessee’s, though Tennessee has played a tougher schedule to date. Where this game will really be decided is on 3rd and 4th down. Tennessee is last in the SEC in converting on these downs while Georgia is middle of the pack. If Tennessee loses this game, they could easily finish the season 3-9 with their only additional win coming at Memphis. A Georgia loss all but guarantees a losing season for the Bulldogs as they’d only likely be favored in two of their remaining games — both home games against Vanderbilt and Big Sky cellar dweller Idaho State. Look for another late Tennessee rally, but much like Volunteer rallies against Florida and LSU, this one will also come up short. Prediction: Georgia 29 – Tennessee 23.
Mississippi State @ Houston (+5, 56) — 7:00 p.m. CDT
Mississippi State takes their middle of the league ho-hum team to Houston on Saturday night to face the 2-0 Houston Cougars. The model is predicting a 4-point Houston victory, though I hope it is wrong this week; I hate to see any SEC school lose to a non-BCS conference opponent. I’ll tell you a little bit about what the model doesn’t know: Houston QB Case Keenum is out for the year and Houston star wide receiver James Cleveland will miss this game. It’s perhaps a significant failing of my model, and one that is not easily remedied. So take this pick with grain of salt as my model works under the fundamental assumption that the players who put up the year-to-date statistics are the same players who take the field this week. Houston lost to the one decent team they played this year (UCLA), so their offensive numbers, as good as they are, are largely against CUSA opponents. Scoring 44 points per game and averaging 465 yards of total offense per game will force Dan Mullen to take the Cougars seriously. If Mississippi State loses the turnover battle, they could very well also lose this game. That said, this is a better than average Mississippi State team that played Auburn very close and managed to play LSU to a draw on offense and defense. The Bulldogs threw 5 INTs in the LSU game and lost 29-7, but it was as close a game as a 22-point loss can be. This game is my “not a lock” game of the week, meaning follow the model only if you’ve got money to burn. Prediction: Houston 28 – Mississippi St. 24.
Auburn @ Kentucky (+6, 58) — 6:30 p.m. CDT
My model is predicting a big Auburn win, though the Vegas line suggests things may be a little closer. Just for the record, there is no pro-Auburn bias built into the model. What the model understands is that the SEC’s No. 1 rushing offense (269 yards/game) is facing the SEC’s No. 11 rushing defense. While Kentucky does lead the SEC in passing defense (allowing 142 yards per game), Auburn’s passing attack is the most efficient Kentucky will see this year, in fact, Auburn’s passing efficiency is ranked No. 1 in the nation. On defense Auburn is allowing just under 19 points/game (No. 37 nationally) while Kentucky is allowing 29 points/game (No. 87 nationally). Kentucky is also coming off two awful defensive efforts, giving up 48 points to Florida and 42 to Ole Miss. Granted, both those games were on the road, but neither Florida (1.0) nor Ole Miss (1.23) are close to Auburn’s offensive production and points per minute (1.33). Auburn’s weakness in defending the pass (No. 11 in the SEC giving up 242 yards/game) could come into play against a Wildcat team ranked No. 2 in the league (264 passing yards/game). If Auburn can clean things up on defense, the Tigers should win easily. The predicted margin of victory for Auburn at 18 points, on the road, makes this game my “Lock of the Week.” Prediction: Auburn 32 – Kentucky 14.
Alabama @ South Carolina (+7, 48) — 2:30 p.m. CDT
Alabama has played remarkably well on the road over the last two-and-a-half seasons, with only a couple of close calls against Auburn in 2009 and LSU in 2008. In both of those games, Alabama was held to less than 30 points, something South Carolina has a chance to do at Williams Brice Stadium. South Carolina’s defense began the season holding Southern Miss to 13 and Georgia to 6, but has since given up 19 to the mighty Furman Paladins and 35 in a second half collapse to Auburn. It is a little too early in the season for my model to give any serious merit to the notion that South Carolina will compete for the SEC East title, but if Spurrier can get both QB Stephen Garcia and RB Marcus Lattimore going, he’ll have a shot at the upset. From a statistics perspective, it just doesn’t look that likely. Alabama is better than South Carolina in just about every statistical category that counts. South Carolina ranks No. 6 in the SEC in scoring defense, allowing 18 points per game, and ranks No. 7 in scoring offense with 31 points per game. Alabama ranks No. 1 in scoring defense, allowing 9 points per game, and ranks No. 1 in scoring offense with 38 points per game. Alabama’s RBs Mark Ingram and Trent Richardson shouldn’t have any problem rushing for 200+ yards on the same South Carolina defense that allowed 334 yards rushing against Auburn. If the reigning Heisman Trophy winner and play of undefeated QB Greg McElroy aren’t enough, Alabama also has the advantage in special teams play and turnover margin. My personal wishes aside, I don’t see the Tide losing to South Carolina. My model predicts this game won’t be nearly as close as the Vegas line suggests. Prediction: Alabama 35 – South Carolina 19.
Eastern Michigan @ Vanderbilt (-26, 53) — 6:00 p.m. CDT
Perennial SEC doormat Vanderbilt gets a chance this week to take a switch to the red-headed stepchild of the Mid-American Conference: Eastern Michigan. Losers of 17 straight games, Eastern Michigan’s last win was over fellow directional university Central Michigan back in November 2008. Anything short of a four-touchdown blowout would be a “win” for the Eagles who gave up 73-points to Ohio State a few weeks back. As bad as Vanderbilt has looked on offense this year (18 points and 298 yards of total offense per game), Eastern Michigan is going to make them look like world beaters. Eastern Michigan ranks last or near last nationally in every major defensive statistical category. In fact, the only part of Eastern Michigan’s game that is better than Vanderbilt’s is, ironically, the Eagles punting game (No. 17 nationally, netting 39.5 yards/punt). No doubt there’s something to the old expression about practice making perfect. And as you might expect from a team with three wins in as many seasons, for years Eastern Michigan has been allowing anyone and everyone to run or pass on them at will. Nothing short of the Commodores somehow getting lost on their way to their own stadium is going to keep Vandy from enjoying their second (and most likely final) win of the season. Prediction: Vanderbilt 40 – Eastern Michigan 10.
Mac Mirabile is a 2002 graduate of Auburn’s economics and journalism departments. During his time at Auburn, he was a copy and photo editor with The Auburn Plainsman. He has a master’s in economics from UNC-Chapel Hill and has written numerous academic publications on college football, the NFL, and gambling markets. His previous columns can be found here. He can be reached at firstname.lastname@example.org