Dr Bob
Auburn (-6 ½) 33 Wisconsin 26
Thu Jan-01-2015 at 09:00 AM Pacific Rotation: 256 Over/Under 63.5
Another head coach has left Wisconsin and once again it is athletic director and former head coach Barry Alvarez that will lead the team through the bowl season. Alvarez did the same thing at the end of the 2012 season after Bret Bielema left for Arkansas and the Badgers played reasonably well in a 14-20 loss as a 5 ½ point dog to Stanford in the Rose Bowl. But, how do the Badgers rebound from the 0-59 beating they were giving by Ohio State, especially when they lost their head coach shortly afterwards? Mediocre teams can often bounce back from a bad beating at the end of the regular season but better teams have struggled with a loss of swagger after a resounding beat down. In fact, bowl teams with 4 or fewer losses on the season are just 19-40-2 ATS if lost their previous game by 24 points or more, including 13-37-2 ATS they lost to a conference opponent (0-9-1 ATS getting less than 10 points in their bowl after losing by 24 or more in their conference championship game). I do have a 7-30 ATS situation that applies to Auburn, so both teams are in bad situations. With that being the case let’s take a look at the numbers.
Auburn’s offense was once again among the best in the nation, averaging 6.8 yards per play against a schedule that would allow 5.0 yppl to an average team, although that unit would be worse if star WR D’Haquille Williams does not play. Williams is listed as questionable with an undisclosed illness and his 10.7 yards per pass thrown to him would be tough to replace. In fact, Williams missed 3 games late in the season (weeks 11 through 13) and the Auburn pass attack averaged a modest 6.4 yards per pass play, which is well below the 8.3 yppp that they averaged for the season. It wasn’t the competition that contributed to the drop, as the teams they faced in those 3 weeks were a bit worse than the average pass defenses that the Tigers faced over the course of the season. Williams isn’t accountable for a 1.9 yppp drop but replacing his numbers with the numbers from the rest of the receivers would results in an expected decrease of 0.7 yppp, which is worth 1.8 points in the case of Auburn. For now I’ll assume that Williams is out and the math projects the Tigers with 422 yards at 6.4 yppl against a normally very good Wisconsin defense that was 0.8 yppl better than average for the season despite the 10.0 yards per play they allowed to Ohio State.
Wisconsin’s offense is all about RB Melvin Gordon, who has run for 2336 yards at 7.6 ypr and 26 touchdowns. Williams also had a bad game against Ohio State and perhaps the absence of their starting center had something to do with that. However, one offensive lineman is not responsible for a team that averages 7.4 yards per rushing play to suddenly average only 3.0 yprp, as the Badgers did against the Buckeyes, and I’ll chalk it up to a bad day. Wisconsin C Dan Voltz is questionable for this game but reports are that he’ll probably play. The Badgers are 1.4 yards per play better than average offensively and are projected to gain 407 yards at 6.2 yppl against an Auburn defense that is 0.5 yppl better than average (and 0.6 yards per rushing play better than average).
The projected yards are pretty close but Auburn has an advantage in projected turnovers that is worth about 1.8 points and the Tigers are significantly better in special teams. Overall the math favors Auburn by 6.5 points (with a total of 58.9 points) if Williams doesn’t play and by 8.3 points (and 60.5 total points) if Williams is 100%. I’ll call for a 7 point win and I have no opinion on the side and I’ll lean Under 63 points or higher.
*UNDER (71 ½) - Michigan State (+2 ½) 33 Baylor 30
Thu Jan-01-2015 at 09:30 AM Pacific Rotation: 257 Over/Under 72.0
Baylor was campaigning hard to get a spot in the playoffs and even hired a PR firm to make their case. However, Ohio State’s romp over Wisconsin trumped Baylor’s solid win over Kansas State and the Bears are left with disappointment and potentially may not have been fully motivated while preparing for this game. Michigan State, meanwhile, is excited about playing a highly ranked Baylor team and the Spartans match up pretty well given their good defense and a quarterback that can exploit Baylor’s weakness in the secondary.
Michigan State is known as being a good defensive team but the Spartans are very good offensively this season, averaging 6.8 yards per play when starting quarterback Connor Cook is in the game (against teams that would allow 5.5 yppl to an average team). The Spartans run the ball very well (242 yards at 5.6 yards per rushing play) and Cook has one of the highest compensated pass efficiency ratings in the nation with 8.6 yards per pass play against teams that would combine to allow just 5.7 yppp to an average quarterback. Baylor should defend the run pretty well (the Bears are 1.0 yprp better than average against the run) but Baylor’s starting defense has allowed an average of 6.9 yards per pass play in their last 10 games (I excluded their first two games against an impotent SMU offense and against FCS teams Northwestern State) to quarterbacks that would combine to average 6.9 yppp against an average defensive team. Baylor dominated weaker passing teams Iowa State and Texas and they had the fortune of playing Oklahoma with big play WR Shepard out. However, the Bears mostly had problems with good quarterbacks and allowed 7.7 yards per pass play or more in 5 of their last 6 games and allowed 9.0 yppp or more in each of their final 3 games. My math model projects Cook to average 8.9 yppp in this game and the fact that the rushing attack isn’t likely to be as successful as usual (4.5 yprp predicted) should mean a few more pass plays than normal from Cook, which is a positive. That likelihood is built into the model, which projects 470 yards at 6.5 yppl for Michigan State in this game.
While I fully expect Michigan State to move the ball well through the air the matchup between a good Baylor pass attack and a good Michigan State pass defense is less predictable. Bryce Petty struggled in the opener against SMU but I tossed that game out (just as I tossed out the Bears’ defensive effort in that game) and Petty’s numbers from week 3 on were stellar, as he threw for 8.1 yards per pass play against teams that would allow 5.7 yppp to an average quarterback. Petty will be up against a Michigan State pass defense that is the second best that they’ve seen this season (MSU allowed 4.9 yppp to quarterbacks that would combine to average 6.1 yppp against an average defense). The best pass defense that Baylor faced was Texas and Petty completed only 7 of 22 passes and averaged just 3.7 yppp in that game. The next best pass defense that Petty faced was West Virginia and he also struggled in that game (16 of 36 for 4.8 yppp). Petty did play well in some games against good pass defenses but not against the best two that he faced and overall there was a strong tendency to play relatively worse against better defensive teams and relatively better against bad defensive teams (like the 12.2 yppp he averaged against Buffalo). The linear equation to predict Petty’s compensated yppp as a function of the opposing pass defense has a slope of 1.93, which means he was 1.93 yppp better/worse for every yard worse/better than average in pass defense his opponent was. A quarterback that plays at the same relative level regardless of opposition would have a slope of 1 and most quarterbacks are near that slope, so there is strong evidence that Petty’s tendency to play relatively worse against better defensive teams is more than just variance. Inserting Michigan State’s defensive pass rating into that equation would predict Petty to average 6.64 yppp in this game, which is 0.50 yppp worse than the math projects. However, Michigan State’s defense has the same issue that Petty has, as the Spartans were relatively worse against better passing teams, as they allowed 9.7 yppp to Oregon and Ohio State. The Spartans also had some good games against good quarterbacks but the linear equation projecting their pass defense as a function of the opposing quarterback projects the Spartans to allow 7.62 yppp to a quarterback with Petty’s overall rating, which is 0.48 yppp higher than the math model prediction. So, Petty could be predicted to be anywhere from 6.6 yppp to 7.6 yppp in this game and ultimately I’ll stick with the math model prediction of 7.1 yppp – although there is obviously a lot of variance in that prediction. Overall Baylor is projected to gain 440 yards at 5.9 yppl against Michigan State’s defense.
Michigan State has the overall edge from a yards per play perspective, which isn’t surprising given that they’re offense rates slightly higher than Baylor’s offense (+1.3 yppl to +1.2 yppl) and the Spartans have a much better defense, and the math favors Michigan State by 3 ½ points (with a total of 62 ½ points). There is a lot of variance in that prediction, however, given how inconsistent these teams have been against better competition and my alternate model favors Baylor by 1 ½ points, which represents the biggest difference in the prediction of the two models of any bowl game this season. Even the model that favors Baylor still favors Michigan State to cover, however, and Baylor’s level of enthusiasm for this game is certainly in question. I like Michigan State here but if the line is less than +3 points then the money line would be a better option, especially given the higher than normal variance associated with these two teams (the higher the variance the more likely an upset will occur, or a blowout). I’ll consider Michigan State a Strong Opinion at either +3 or more or on the money line if the line you’re getting is less than +3.
The Under appears to be the better play here, as my model predicts far fewer plays than other models might. Most models would probably look at Michigan State’s total plays per game (134.7) and Baylor’s total plays per game (160.2 in the 10 games I’m using) and add those and subtract the league average of 140.2 total plays in regulation. That would give you 154.7 total points. However, Baylor faced a lot of other up-tempo teams that combine to average 6.5 more total plays from scrimmage than average (and Michigan State’s opponents combine to average 1.1 fewer total plays from scrimmage). A simple compensation based on those numbers would get the predicted plays down to 149.3 plays (154.7 – 6.5 + 1.1). My model predicts just 146.8 total plays, as my model takes into account Michigan State’s average of 35.24 minutes of time of possession. Baylor averages 29.64 minutes of TOP per game and they run their offense at a fast pace when they have the ball. However, the Bears are projected to have the ball just 24.36 minutes in this game while the clock eating slow paced Spartans have it for 35.64 minutes. Baylor is expected to have the ball for 5.28 fewer minutes than their average and the difference between their average plays per minute and Michigan State’s average plays per minute in those 5.28 minutes is pretty significant and is why my model projects fewer plays than most other models probably do. I am concerned about the high variance in the predicted passing numbers for each quarterback, which could lead to higher scoring, but if I assume Petty will average at the high end of the spectrum and that Baylor’s defense will continue to struggle against good quarterbacks as they did for the second half of the season (which would project Cook at 10.2 yppp in this game) I still only get 68 total points if each team plays at their normal pace on offense and Michigan State possesses the ball for around 35 minutes as they normally do. So, even in an extreme case where both quarterbacks play better than expected and the teams combine for 6.6 yards per play I still have the total going under 70 points.
The other reason the total is high is because a points based model would predict a game over 70 points but both of these teams had combined red zone efficiencies that were really high and contributed to each team’s higher than expected total points averages. Baylor’s offensive points per red zone opportunity was 5.2 points per RZ, which is what a team with their overall offensive rating should average. However, Baylor’s defense allows 5.2 points per RZ, which is 0.6 points higher than projected based on their overall defensive stats. Michigan State, meanwhile, is also projected to be at 5.2 points per RZ on offense but the Spartans are at 5.4 points per RZ and their defense, which is really good overall, has allowed teams to average 5.4 points per RZ opportunity, which is extremely high for a defense that is as good as their defense is overall (they should allow 4.4 points per RZ). The red zone variance of these two teams accounts for a total of 5.0 points per game, which has also created some value on the under. I also get a total of 71 ½ points if the teams combine for 154 plays and continue to have extremely high red zone scoring averages but I don’t see that many plays being run and the red zone scoring averages should regress towards what is expected. I can still envision both quarterbacks having more success than my model predicts, so I won’t make this as big a play as the math would suggest. I’ll go UNDER 69 points or higher in a 1-Star Best Bet.
Missouri (-5) 24 Minnesota 20
Thu Jan-01-2015 at 10:00 AM Pacific Rotation: 259 Over/Under 47.5
There isn’t much exciting or interesting to say about this game so I’ll keep it short. Missouri somehow got to the SEC Championship game for a 2nd straight year with a mediocre offense that averaged only 5.4 yards per play. That attack struggled in the middle of the season when injuries hit their very thin corps of wide receivers. And, when I say thin I really mean thin. Missouri has 3 wide receivers that see the field, as Sasser, Hunt, and White combined for 229 targets while the rest of the wide receivers combined for just 35 passes thrown to them all season, and most of those were when either Hunt or White were out with injury. Hunt and White both missed the week 5 South Carolina game and quarterback Maty Mauk averaged only 2.9 yards per pass play in that game against a weak South Carolina pass defense. White missed the next game against Georgia in which Mauk averaged only 3.0 yppp and he missed week 10 against Kentucky (4.2 yppp for Mauk). Normally, having one receiver out wouldn’t matter much, especially given that White averaged a mediocre 8.0 yards per target, but the backups combine to average a pathetic 3.1 yards per target. The Missouri game, in which Hunt was also out was predictably bad for Mauk, as Hunt leads the team at 10.4 yards per target and those ill-equipped backups were filling the void of two starters. The other receivers don’t play unless one of the top 3 are out, and Mauk’s yards per pass play rating would go up 0.3 yppp if Hunt and White played every game. Even with that adjustment Missouri’s mediocre attack is still projected to gain just 357 yards at 5.5 yppl in this game against a solid Minnesota defense that has allowed 5.4 yppl to teams that would combine to average 5.7 yppl against an average team.
Minnesota’s offense is 0.1 yppl better than average with starting quarterback Mitch Leidner in the game (he missed week 4 against San Jose State), averaging 5.5 yppl against teams that would allow 5.4 yppl to an average team. That attack isn’t good enough to do much damage against a very good Missouri defense that has allowed just 4.8 yppl this season despite facing teams that would average 6.1 yppl against an average defensive team. The Gophers are expected to gain just 304 yards at 4.6 yppl and their great special teams doesn’t give them the big advantage that is does against most teams, as Missouri also has very good special teams (although Minnesota is better in that regard). Overall the math favors Missouri by just 4 points with a total of 43 points and I’ll lean Under 47 points and I have no opinion on the side
