VN: On Long Drives

by Bill Connelly

We've all written about it at this point -- Brian Fremeau did just Wednesday, in fact -- and we've all marveled at it in the same way. But that doesn't make it less remarkable: With 9:31 remaining in the Big Ten title game, down four points with a trip to the College Football Playoff on the line, Michigan State got the ball via punt at its 18-yard line. Twenty-two plays, 82 yards, and 9:04 later, freshman L.J. Scott extended the ball just past the goal line before his knee hit the ground, and Michigan State took a 16-13 lead.

Because of the magnitude of the situation and the sheer unlikeliness of the drive, this will forever be remembered in East Lansing. At the very least, it won them a conference title in a year in which defending national champion Ohio State was the overwhelming favorite and Michigan, led by Jim Harbaugh, made major noise for a while. But if Michigan State goes on to win the national title, this becomes one of the most famous, significant drives in college football history. (Granted, State probably isn't going to win the national title, but we'll worry about that on December 31.)

That it was so incredibly unlikely made me a little bit curious. I wanted to follow up on Brian's Wednesday piece by looking at exactly what it means when a possession lasts a certain number of plays.

I omitted end-of-half drives (since there's quite a bit of context there) and looked at drives that resulted in either a safety, a turnover, a punt, a turnover on downs, a field goal attempt (made or missed), or a touchdown. As your play count increases within a given drive, what becomes more and less likely? This isn't an attempt to draw any logical conclusions we couldn't come up with on our own -- it's more a simple look at data from an angle I hadn't looked at before. I assume that if I find that interesting, others might, too. (I've been wrong before!)

First, here's a link to the total number of drives from 2015 with a given number of plays. There have been 20 drives of 18 plays, 10 of 19, five of 20, and one each of 21 and 22.

That gives you a sense of where sample sizes begin to affect matters. Now, here's the outcome chart.

Some observations (none of which wouldn't have been guessed beforehand):

• If your drive lasted one or two plays, you did either something great or something terrible.
• 80 percent of three-play drives (and 24 percent of all drives) are of the three-and-out punt varieties.
• The threat of a turnover on downs doesn't really fluctuate an incredible amount between four and 18 plays. I thought it might go up a bit (since, in theory, you're advancing into opposing territory and are therefore more likely to be willing to go for it on fourth down), but other than a small-sample spike at 19 plays, there's not really evidence of that.
• Your odds of a touchdown don't really increase significantly after 10 plays either. I think that's basically because of what Brian said in his piece this week regarding drives of 20-plus plays: "[D]rives of at least 20 plays are pretty good indicators of an offense that is struggling to get any big chunk plays against a defense, and sputtering in the red zone after that kind of methodical play is more likely than not."
• While we always remember the 99-yard pick-six, it appears your odds of turning the ball over do decrease a bit as the field shrinks.

Sort of, at least. If you break drives out by where the drive ended, about 10 to 20 percent of them ended via turnover everywhere on the field except between about the opponent's 25 and 40.

Former ULM coach Todd Berry once told me that once his offense crosses the opponent's 40, he wants to take at least one shot at the end zone. That, combined with the fact that most college teams don't have kickers who can make 55-yard field goals with regularity, creates a situation where you are taking more risks with the ball in that area.

The most impressive part of Michigan State's drive, by the way? The Spartans were only down to begin with because they weren't capable of finishing drives in the first 50 minutes. In five previous scoring opportunities, MSU had settled for five field goal attempts, missing two. But with the game on the line and no capability of making a big play, the Spartans gained 13 yards on third-and-4, 4 yards on third-and-3, 16 yards on third-and-8, 2 yards on third-and-1, 2 yards on fourth-and-2, and 1 yard on third-and-goal from the 1. Nothing like removing all margin for error and winning anyway.

One more note, by the way: similar to the team stat profiles I began posting this year, on Wednesday I posted data-dump previews of each of the 40 bowl games.