Feng: Can MSU upset 'Bama? Here are 5 insights

Ed Feng
Special to The Detroit News
Shilique Calhoun, left, and Lawrence Thomas take time out from Cotton Bowl practice for a little dancing.

Can Michigan State pull off the upset against Alabama in the College Football Playoff? The markets don't think so, as they have made the Crimson Tide a 9.5-point favorite.

This large spread might make the game seem uninteresting. In addition, my numbers at The Power Rank also predict an 8.4-point win for Alabama, which corresponds to a 27 percent win probability for Michigan State.

However, analytics suggest many intriguing matchups and stories in this game. Let's look at the top five insights from numbers for Michigan State against Alabama.

1. Michigan State's defense

The Spartan defense struggled at times this season. They gave up 6.0 yards per play to Rutgers in a tight game, and the triple option attack of Air Force racked up 7.1 yards per play.

Then Michigan State's defense surged toward the end of the season. MSU was magnificent against Ohio State, holding an elite offense to 2.9 yards per play.

For the season, Michigan State's rush defense ranks 10th in my numbers that take yards per carry and adjusts for strength of schedule. Despite allowing some big runs early this season, an elite defensive line led by Shilique Calhoun and Malik McDowell has mostly limited opposing ground games.

Through the air, the Spartans ranked 26th in yards per attempt adjusted for strength of schedule (numbers include sacks). The secondary has been suspect, but the pass rush helps by sacking the QB on 7.8 percent of pass attempts this season.

Michigan State's strength against the run will help in their matchup against Alabama.

2. Alabama's shift from pass to run offense

In last season's playoff game against Ohio State, Alabama insisted on throwing the ball against a defense better against the pass than the run. It's a habit that OC Lane Kiffin carried over from his days as head coach of USC.

Alabama continued this tendency early this season. Even with new starter Jake Coker, Alabama threw the ball on 53.3 percent of plays the first four games of the season.

Sometimes, the game situation called for more passes, as Alabama trailed Ole Miss most of the game in an eventual loss. However, the Tide still threw more than they ran in easy wins over Middle Tennessee State and Louisiana Monroe.

This tendency changed with the Georgia game. Over the last nine games of the season, Alabama ran on 62 percent of plays, giving opponents a steady dose of Derrick Henry.

Alabama hasn't been particularly efficient running the ball this season. Their 5.3 yards per carry ranks 44th in the nation (numbers do not include sacks), although schedule adjustments improve this rank. However, this insistence on running the ball won Henry the Heisman Trophy.

Alabama will mostly likely continue to run the ball against Michigan State, a good matchup for a Spartan defense better against the run than pass.

3. Winning the turnover battle

Michigan State has 16 more takeaways than giveaways this season. This plus-16 turnover margin helped the Spartans to a 12-1 record and a spot in the College Football Playoff.

However, these lofty turnover statistics have no bearing on the future.

Randomness plays a large role in turnovers, and a plus-16 turnover margin over 13 games has no ability to predict the turnover battle against Alabama.

I know this sounds stupid. Michigan State caused that fumble by punching the ball away from the Iowa receiver for a late-game turnover that sealed the Big Ten championship game. It seems like simple Newtonian physics.

The question is whether this play implies that MSU can cause Derrick Henry or Calvin Ridley to fumble on Thursday. The numbers say no. It takes the right set of circumstances for a defender to even get his hand on a ball, much less tear it away from the ball carrier.

Another play from the Big Ten championship game also shows the randomness of turnovers. In the first half, Demetrious Cox made an interception in the end zone after the ball bounced off the back of another Michigan State defender. The pick killed a potential touchdown for Iowa in a game that Michigan State won by 3.

The Spartans increase their chances of beating Alabama by winning the turnover battle. However, they won't win this battle simply because of their plus-16 turnover margin this season.

Macgarrett Kings Jr. (85) and Aaron Burbridge work through drills at the Cotton Bowl.

4. The Cook and Burbridge show

Michigan State can't run the ball. The Spartans average 4.3 yards per carry, an unimpressive 111th of 128 FBS teams this season (numbers do not include sacks).

Sure, LJ Scott seems like a stud for stretching the ball out over the goal line to put Michigan State ahead of Iowa. However, even Scott only runs for 4.9 yards per carry.

It will be even more difficult to run against Alabama. The 'Bama front seven has allowed 3.5 yards per carry this season, second best in the nation.

Leonard Fournette of LSU rushed for 31 yards on 1.6 yards per carry against Alabama. He ran for 6.8 yards per carry against all other opponents.

Michigan State should rely on its passing game against Alabama. Both QB Connor Cook and receiver Aaron Burbridge have NFL talent, and the Spartans must take their shots downfield with this duo.

Mark Dantonio will insist on running the ball, and it's a good idea to have some balance in run and pass plays. But Michigan State will not win this game unless Cook connects often with Burbridge. This could be particularly effective early in the game on first down when Alabama will expect a run.

5. Michigan State against AP top 10 teams

Over the next few days, you'll almost certainly hear stats about Michigan State's record against top-ranked teams. Since 2013, Mark Dantonio has gone 6-1 against teams ranked in the top 10 of the AP poll. This set of games includes a surprising win over Ohio State without Connor Cook this year and a Rose Bowl win over Stanford two seasons ago.

However, Michigan State's record against AP top 10 teams has all kind of small sample size warnings. Remember when the Tigers started the 2015 season 11-2? This was a small sample size fluke for a team that ended the season below .500.

In addition, humans tend to make up stories about these streaks of success. Fans assign words like "clutch" to teams like Michigan State that win a few big games. Brad Ausmus thought closer Bruce Rondon looked "confident" after one strong outing.

Unfortunately, these stories usually seem silly in the long run.

Michigan State can beat Alabama; my numbers give them a 27 percent chance to win. However, a small sample of wins over AP top 10 teams in no way improves this probability.

Ed Feng has a Ph.D. in chemical engineering from Stanford and runs the sports analytics site The Power Rank. Email Ed Feng here.