Feng: Defense is X-factor in predicting MSU win total

Ed Feng
Special to The Detroit News
Oregon could be ripe for an upset as the Ducks will be breaking in a new quarterback.

Mark Dantonio has elevated Michigan State football to new heights the past two seasons. To see this, check out this visual that encapsulates the past 30 years of Spartan football:

MSU has won 11 or more games four times under Mark Dantonio.

The top and bottom panel show wins and a rating from my computer calculations at The Power Rank, respectively. This rating gives an expected margin of victory against an average FBS team.

Dantonio has made a bowl game in each of his eight seasons as Michigan State coach and won at least 11 games four times. However, the computers think the last two seasons were his best as Connor Cook took over at QB.

Dantonio's team in 2013 and 2014 had a higher rating than any Michigan State team over the past 30 years. This includes the years in which Nick Saban coached (1995-99).

Can Michigan State build on this success and get even better in 2015? The numbers say no, but there are reasons to doubt those predictions. Let me explain.

Michigan State's turnover margin in 2015

My preseason college football rankings come from a regression model that considers a team's performance over the past four years, turnovers and returning starters. Despite its simplicity, these rankings have predicted the game winner in more than 70 percent of games since the 2005 season.

The model ranks Michigan State 16th this preseason, lower than most of the polls. However, the model underestimates this Spartan team due to the turnover variable.

Michigan State had a plus-19 turnover margin last season. This means the Spartans took the ball away 19 more times than they gave it away.

Randomness plays a big role in turnovers, and a team with such a favorable margin usually gets rated too highly in my calculations. The preseason model sees this high turnover total and adjusts a team downward.

However, Michigan State's plus-19 in turnovers didn't help them much last season. They had a plus-11 turnover margin in blowout wins against Jacksonville State, Eastern Michigan and Wyoming. In contrast, they had a zero margin against Oregon and Ohio State.

My preseason model rates Michigan State a bit low, and they should be closer to a top-10 team. Keep that in mind as we break down the schedule.

Breaking down the schedule

The preseason rankings imply a win probability for each of Michigan State's games, which lets us break down the schedule into four tiers.

Very likely wins: Air Force (92 percent), Central Michigan (95 percent), Purdue (90 percent), Indiana (92 percent), Maryland (92 percent).

Likely wins: at Western Michigan (82 percent), at Rutgers (80 percent), Penn State (77 percent).

Toss-up: at Michigan (55 percent), at Nebraska (53 percent).

Unlikely wins: Oregon (35 percent), at Ohio State (20 percent).

The sum of the win probabilities in each game gives a win total of 8.6 for Michigan State. The flaws of my model make this total low, and it is reasonable to expect nine or 10 wins in 2015.

Two games deserve further attention.

Michigan: Putting Michigan in the toss-up category seems strange after Michigan State clobbered the Wolverines last year. However, Michigan had a terrible minus-16 turnover margin last season.

The preseason model recognizes that Michigan was most likely better than their 57th rank last season due to the randomness of turnovers. It adjusts them up to 34th this preseason.

In addition, Michigan State travels to Ann Arbor this season. The three points for home field drags Michigan State's win probability down to 55 percent.

Oregon: Oregon travels to East Lansing in Week 2 for one of college football's best non-conference games. Despite their extraordinary success the past five years, the Ducks have big questions at quarterback and cornerback.

In addition, their defense took a step back last season as Don Pellum took over for long time coordinator Nick Aliotti. This year, they won't have a Heisman caliber QB in Marcus Mariota to hide any weakness on defense.

If Oregon is ever going to step down from college football's elite, this is the year. Keep an eye on the score of Oregon's opening game against a good FCS school in Eastern Washington.

Demetrious Cox


In the best-case scenario, Michigan State takes care of business against Michigan and Nebraska and then plays its their best game of the season at Ohio State on Nov. 21. An upset win puts the Spartans in the middle of the playoff conversation.

In the worst-case scenario, the defense struggles from poor cornerback play. Michigan State had two cornerbacks (Darqueze Dennard, Trae Waynes) drafted in the first round of the past two NFL drafts. Any defense would regress from the departure of that talent.

Darian Hicks struggled at cornerback last season and lost his starting job. Then he suffered another set back this August as he came down with mono. Demetrious Cox, a converted safety, will most likely start at the other cornerback spot.

Even with an elite defensive line, Michigan State will find it difficult to play their aggressive style of defense without dependable cornerbacks. If the defense takes a step back, seven or eight wins is possible.

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