Statistical In-Ferentz: Trench Warfare

The Win Probability (WP) graph tells the story of a game that Michigan State could very well have won, and nearly did.*

* The graph ends before overtime because the WP numbers are derived from the Advanced NFL Stats WP Calculator, and overtime is obviously different in the NFL compared to the NCAA.

After James Vandenberg's interception and the resulting Spartan touchdown, Michigan State hovered around a WP of .7 (that is, a 70% chance of winning) for the next three quarters, and even .92 when they got the ball back with a seven point lead and 6:41 left in the fourth quarter. Things looked particularly bleak for Iowa after a personal foul left the Hawkeyes with a first and 24 from their own 18 and just over five minutes left in the game, but just eight plays later Mark Weisman was running into the endzone to tie the game and force overtime.*

*Interesting note: Iowa covered 82 yards in eight plays on that drive for an average of 10.3 yards per play; for the rest of the game, Iowa had just 175 yards on 45 plays, for an average of 3.9 yards per play. Yikes.

So what was the cause of Michigan State's collapse? Iowa's stout defense deserves the most credit, preventing the Spartans from pulling away time and time again after Iowa's offense stalled. But I would argue that Michigan State made one crucial strategic error, and one kinda-sorta error, that kept Iowa within striking distance.

Decision #1: Michigan State ball, 4th and 1 at the Iowa 7 yard-line, MSU elects to kick FG (9:13 left in 2nd quarter, MSU ahead 7-0)

This was a decision that we're used to seeing from Kirk Ferentz: taking a field goal within the 10 yard-line when the distance to convert is manageable. In this case, Michigan State elected to let Iowa off the hook with a kick, even though they only needed one yard to convert the fourth down and even though they had Le'Veon Bell in the backfield. In WP terms, a first down there would have bumped the Spartans up from .82 to .88, while a field goal actually dropped their WP to .77. On the other hand, if Michigan State had gone for the first down and failed, their WP would have dropped to just .73. Right away you can see that field goal is scarcely better than a failed conversion, and a little math shows confirms that going for it was probably the better decision here. The break-even probability to go for it turns out to be just 27% -- that is, if Sparty could have gotten that one yard just one out of four times, it would have been worth it for them to go for the first down.

Decision #2: Michigan State ball, 4th and goal at the Iowa 5 yard-line, MSU elects to kick FG (7:45 left in 4th quarter, MSU ahead 10-6)

This is a tougher call, but once again, Michigan State settled for a field goal from deep in Iowa territory when a touchdown would have sealed the game. The WP for a touchdown would have been .93, the WP for a failed conversion would have been .77, while the WP for the field goal was .86. After doing the math, we can see that Michigan State would have been better off going for it if they had felt they could make the touchdown 56% of the time. Given the struggles on both sides of the ball to do anything on offense, I can see how Michigan State might have been skeptical of scoring on that fourth down (and of Iowa ever responding with a touchdown), but this play represented a chance to go up two scores with just under eight minutes to play against a team that had struggled to score points all day.

Bonus Decision: Iowa ball, 4th and 5 at the Michigan State 10 yard-line, Iowa elects to kick FG (14:08 left in 4th quarter, MSU ahead 10-3)

Iowa was not without its own questionable decision-making. This was a tough one, but Iowa's decision to kick the field goal here could have proven very costly. By cutting the lead from seven to four, Iowa still left themselves in need of a touchdown to win the game. At that late stage in the game, being down four is almost as bad as being down seven: the WP if Iowa had gone for the conversion and failed is .2, while the WP for a made field goal was just .29. Converting the first down, on the other hand, would have raised Iowa's WP to .42. It turns out that Iowa's break-even probability to go for it is just 41% in this situation. You could argue that Iowa had very little chance of making those five yards, especially given their meager offensive output to that point. By the same token, though, by kicking the field goal, they ensured that they would most likely need to not only keep Michigan State out of the endzone, but get all the way back there themselves. Luckily, that's exactly how things worked out, but it was a close-run thing.

In a way, it seemed as though Mark Dantonio and Kirk Ferentz were involved in an extended staring contest, with neither wanting to be the first to make a risky play. From Michigan State's point of view, this was understandable; they were the heavy favorites after they built up a 10 point lead, and probably thought they could rely on their defense to win the game, provided they didn't make any terrible mistakes on offense. From Iowa's point of view, it made sense to play somewhat conservatively, precisely because Michigan State was playing so conservatively. Neither team wanted to take the risk of relying on their offense to overcome a tough opposing defense. It was similar in many ways to the 9-6 LSU-Alabama game of last fall, when both teams went into shells on offense, punting back and forth all game and waiting for the other team to blink. Or, to move out of the realm of football, it was a little like trench warfare. Neither side wanted to take a risk, for fear of heavy damages, so both sides settled in for a war of attrition. The lack of daring on either side kept the contest close, but it also made it more of a toss-up. Michigan State might be wishing they had tried a few more daring maneuvers over the wire.

WP Heroes/Losers

I would have guessed that Keenan Davis' 35-yard catch to keep the drive going in the fourth quarter would have rated as the most significant Iowa play, but the model actually says that it rated third at +.11 WP. Second was Mark Weisman's 37-yard run on 3rd and 6 later in the same drive (+.15 WP), and first was Weisman's touchdown on 3rd and goal to finish the drive (+.34 WP). I wouldn't put too much credence in the exact WP values of any one play, though, as the model jumps around a lot in the final minutes of a close game. As a whole, though, that final drive was naturally very important, bringing Iowa back from the dead (.15 WP) to even (.5 WP at the end of regulation).

And the worst WP play for Iowa was actually not Vandenberg's interception in the first quarter (-.10 WP). It was instead a 37-yard completion from Andrew Maxwell to Keith Mumphrey on 3rd and 13 early in the fourth quarter. That kept the drive alive and allowed Michigan State to move into position for a field goal that gave the Spartans a 7-point lead. If that drive had led to a touchdown, that one play would have represented Iowa's last good chance to pull back into the game.