The story of the Minnesota game is one that has been rehashed over and over again: a string of insane, hideous, frustrating missed opportunities. A good way to appreciate this is to look at the number of "Expected Points" Iowa left on the field. The concept of Expected Points allows us to estimate the number of points a team can be expected to score on average on a drive, given a certain down and distance.* Here's how Iowa rated on a few choice drives Saturday:
- 13:30 left in 1st quarter, 1st and 10 at the Minnesota 39. Expected Points: 2.72; Actual Points: 0 [turnover on downs]
- 8:15 left in 1st quarter, 1st and 10 at Iowa 49. Expected Points: 1.98; Actual Points: 0 [punt]
- 3:00 left in 1st quarter, 1st and 10 at Minnesota 14. Expected Points: 4.44; Actual Points: 0 [missed FG]
- 12:00 left in 2nd quarter, 1st and 10 at Minnesota 11. Expected Points: 4.65; Actual Points: 0 [missed FG]
- 3:45 left in 3rd quarter, 1st and 10 at Minnesota 20. Expected Points: 4.08; Actual Points: 0 [sack/fumble]
* All EP figures come from the Advanced NFL Stats calculator
, and thus the usual caveats apply about the model being based on historical NFL stats, not college stats. But the general idea that most teams score either a touchdown or a field goal when they have a first and 10 from their opponent's 10-20 yard line probably carries over.
If you add all those missed points up, you get 17.87. And that total should probably be a little higher, because the strip sack/fumble recovery by Minnesota so drastically changed field position that it was in effect worth an extra point or so in expected points for Minnesota. Of course if that first drive had ended differently the rest of the game would have changed and those subsequent drives wouldn't have happened (yes, I've seen "The Butterfly Effect"), but looking at the numbers does give you a sense of just how many scoring opportunities Iowa let slip away. So how did this happen? Part of it was the dumb bad luck of missing field goals, but part of it was a clever defensive game plan by the Gophers that frustrated Iowa all day. I saw three main components to their plan:
Step 1: No Deep Stuff
Jerry Kill and his staff seemed to have taken the lesson that Iowa's offense was at its most dangerous in the long passing game, because they responded to Iowa with heavy use of deep cover 4 coverage, with the corners playing 8 yards off the line of scrimmage. Here's an example from the third quarter that also shows the unusual set-up of the defensive ends.
If Iowa was going to beat the Gophers, they would have to do it inch by inch, not in large chunks. Which you can see by the yards James Vandenberg passed for by area of the field:
VDB threw deep only four times, and not for a particularly great average (15 yds/pass).
Step 2: Crazy wide defensive linemen
It certainly seemed like Minnesota was copying the recently popular NFL concept of setting their ends up in the so-called "wide nine" stance, as in outside the tight ends. And unlike NFL teams, who prefer this look solely on passing downs, it certainly seemed like Minnesota went with this look all the time. Why go this route? Here's a good article
on the concept, but the basic idea seems to be to pressure the quarterback. And Vandenberg did seem to face pressure all day, even when Minnesota didn't have an explicit blitz call on. It's hard to say if that was due to the alignment of the ends and how much was due to a mediocre day by the Iowa offensive line, but whatever it was, it worked: Vandenberg never seemed to have time or a lot of space to really set up and uncork his throws. The set-up of the linemen probably also frustrated Iowa's ability to run its standard outside stretch play and the play-action off that stretch play, as the ends were positioned perfectly to meet the running back going outside or the quarterback bootlegging it outside the other way.
There are obviously some weaknesses with this set-up, first being that it leaves an enormous gap between the defensive end and the defensive tackle. And as the Smart Football article above notes, this allows the offensive line to get out into the second level with ease. If you look at where Marcus Coker
generated his yardage on Saturday, it wasn't outside, it was all straight up the middle. Minnesota had their linebackers lined up pretty narrowly, within the tackle box, so it was apparently their job to take care of the inside stuff when it inevitably got by the defensive line. That didn't work so well, as Coker went off for 250 yards in three quarters of work, and probably could have gone for more if given the opportunity. The deep corners and safeties actually wound up helping the Gophers here, because even when Coker broke through the linebackers, there was still a layer of defense to stop any long runs.
Step 3: Confusion on passing downs
Sticking with zone so heavily put a lot of pressure on James Vandenberg
to make quick reads and find receivers, and he actually did a pretty good job of it: 16-24 for 177 yards, no interceptions, 7.7 yards per throw. Vandenberg generally did a good job of finding an open receiver on a short out or check down, and got the ball out accurately (his receivers didn't help him by dropping four or five of those passes). But the effective part of the Gopher scheme came not in making it impossible for Vandenberg to pass successfully, but in cajoling him into throwing it short rather than long. They did this three ways: first, by dropping very deep, second, by blitzing on obvious passing situations, most successfully on the crucial third down in Minnesota territory where Vandenberg was stripped of the ball:
As you can see, Kevonte Martin-Manley
is open at the top of the screen, but the play is a sprint out away from that side of the field, and Vandenberg never sees the blitz coming. Minnesota has eight men coming against six blockers, and one defender slips totally free. If the first read (probably McNutt) had been open, Vandenberg would have been able to get the ball out quickly, but he hesitates a second, and it's all over. It was really a brilliantly timed blitz from Minnesota, one that came just late enough that the quarterback never saw it coming. The offensive play call also happened to mesh perfectly with the defensive play call, as the direction of the play led Vandenberg's eyes away from the blitz. Still, given Minnesota's tendencies, Vandenberg needed to be aware that pressure was likely coming, and that he would have to get the ball out of there post-haste. By the way, this was clearly the WP play of the game, taking Minnesota's WP from .13 to .31 in one fell swoop.
The third way Minnesota got into Vandenberg's head was by swarming a lot of men at the line of scrimmage as if they were going to blitz, then backing off into zone. And sure enough, whenever they presented this look, Vandenberg would check off to a short throw. On one occasion, this tactic tricked Vandenberg into throwing a four-yard throw on third and seven, and Iowa punted on the next play.
All in all, Minnesota's defensive strategy wasn't all that successful -- Iowa did go for over 400 yards and 21 first downs -- but it did enough in crucial situations, mainly third downs, to stall several promising Iowa drives (Iowa was 4-11 on third down). And relative to Minnesota's recent defensive efforts (giving up 41, 45, 58 and 37 points), it was a masterpiece. The biggest accomplishment of Minnesota's defense was to take away Iowa's strength -- the deep passing game -- and hope Iowa didn't punish their weakness -- the rushing game -- too horribly. The Gophers are, after all, not quite so terrible when it comes to defending the pass (68th in yards/completion), but are truly horrific when it comes to rushing defense (111th in yards/carry).
That Onside Kick
There's been some debate about whether Kirk Ferentz could reasonably have expected an onside kick in the fourth quarter. There was certainly a high pay-off for Minnesota to recover an onside kick: in Win Probability terms, a generic NFL team would improve their chances from .27 to .34 by recovering the onside kick at their opponent's 41. A failed onside kick, on the other hand, would reduce the generic team's chances of winning from .27 to .22. So it turns out the break-even probability to attempt the onside kick there was .42, or 42%. Given that surprise onside kicks are successful in the NFL about 60% of the time
, Coach Kill had good reason to go for the kick there, assuming it would
be a surprise. But should Ferentz have expected it? One way to look at the problem is to look at how other coaches have behaved in similar situations. In the article I just cited, Advanced NFL Stats guru Brian Burke produces a chart that shows onside kick success plotted against WP. When WP is low -- that is, when the kicking team has a low chance of winning and desperately needs to recover the kick in order to come back -- onside kicks are not very successful. Specifically, when the WP of the kicking team is below .2, the chances of recovering the kick are in the 20%-30% range, but when the WP goes above .2, the chances of success jump up to around 60%. Minnesota had a WP of .27 when they kicked off with eight minutes left, so they were just in the zone where most coaches are "surprised" by onside kicks. So in the generic NFL case, most coaches would have been surprised by an onside kick in that situation.
On the other hand... this wasn't a generic case. Given Iowa's success at running the ball and Minnesota's general difficulties passing the ball, kicking off to the Hawkeyes represented a bad bet for the Gophers. A very probable outcome was that Iowa would run 2-4 minutes off the clock and score 3-7 points, leaving Minnesota with only 5 minutes to drive the length of the field and score eight points at least. One rule for underdogs is that they need to do whatever they can to make sure the expected result doesn't happen, and Minnesota made a very smart gamble to upset the normal course of events here. With a little psychological insight, Iowa's coaches could have seen that an onside kick was at the very least a possibility coming from a coach as desperate for a win (and with as bad a defense) as Jerry Kill, and put in the appropriate personnel. It might have been impossible for Kirk Ferentz to know for certain that an onside kick was coming, but it's not like the downside to being ready for the possibility was that high. He could have looked at it as a cheap insurance policy against disaster, one that would have only cost him (at most) a few yards on the kickoff.
One other interesting aspect of the onside kick: when you combine it with Iowa's non-drive kneel down to end the first half, Minnesota had 11 drives in the game to Iowa's 9. You can only score when you have the ball, so those two extra drives gave Minnesota a small edge in the game. One more reason that abandoning possessions at the end of the half is a bad idea.
Even with all that going wrong for Iowa, however, when they got the ball back with 2:48 left to go, they were still in very good shape to complete a comeback. In the generic NFL model, teams in that situation win 51% of the time, which probably is due somewhat to the quality of NFL kickers, but also to the fact that 2:48 is a lot of time. Minnesota was still doing the same stuff it had earlier -- extra wide defensive ends, lots of deep zone -- so the time and opportunity was there to take advantage of that defense with inside running plays. Going for passes on four straight downs meant going away from what had been a strength against Minnesota's defense all day.
- Credit where credit is due: the Hawks took a smart gamble going for it on 4th and 6 from the Minnesota 35. By my calculations, going for it made sense (as compared to landing a punt at the Minnesota 15) if Iowa felt they could make it more than 20% of the time. The break-even percentage for a field goal is roughly the same -- 21% -- so that would have been a decent option, too. It didn't work out, but it was still a decent bet. The problem is that none of the options are particularly good from the opponent's 35.
- Missed field goals from 24 and 43 yards: more proof that field goals in college are for suckers. The one that came on 4th and 2 from the 6 was an especially dubious choice.
- A lot of people have been talking about the decision by Iowa to sit on the ball going into the half, and I'm of two minds on the issue. Obviously, if the Hawks thought they had a better chance of scoring than getting scored on as a result of that drive, they should have tried to get some points. But given their struggles in obvious passing situations, I sort of understand why they are so skittish about the two minute drill. That's not really an excuse, though. The two minute drill is such a critical skill for any team to master that admitting you just can't do it is sort of like a pilot admitting he can fly the plane, just not at night.
- Iowa's defense again had trouble with a running quarterback, and you can see how it's starting to affect Iowa's pass defense. On the 61-yard pass play that set up Minnesota's first touchdown, you can see on the replay that the Hawks are in quarters coverage, but that Shaun Prater's eyes are focused on MarQueis Gray as he rolls to his right. A second later and the pass is over Prater's head and complete for a big gain that got the Gophers back into the game. It's not totally clear if Tanner Miller was expected to help out over the top on that play (he certainly wasn't covering anyone else), but it looks like Miller is focused on covering the middle of the field, not the outside. There was also a play near the goal-line where Prater had his eyes on Gray and ignored a tight end breaking open late for a touchdown pass. There have been a few plays this year where Prater's eagerness to get to the ball and make a tackle (I'm thinking of the Sunseri run/pass option play against Pitt) has taken him out of position in coverage. It's a tough dilemma to be in, especially when the other option is letting the quarterback run free for 15 yards, but there needs to be some understanding between Miller and Prater on those plays.