I thought this week I would look at the most decisive plays in the game and examine closely what happened to make them work. To determine the "most decisive" plays, I used the generic NFL win probability model from the website Advanced NFL Stats. The basic idea is that for every possible game situation (down and distance, yard line, score, time left) in a game, there is a given probability that one team will win and the other will lose.* When a significant play alters that game situation, the probability changes. So, for example, Tanner Miller's interception shifted Iowa from having .49 WP in the NFL model (a 49% chance of winning) to .83 WP (an 83% chance). Big plays like that are the ones I looked at here.
* Usual caveat: the Advanced NFL Stats model is based on historical NFL play, so it doesn't totally apply to college, but does (in my opinion) give a good general idea about the importance of individual plays. I used the WP calculator page to generate WP for all the game situations below. For more explanation of the whole WP concept, see previous versions of this column or look around the Advanced NFL Stats page.
1) Northwestern ball, 1st and goal from the Iowa 7, 4:27 left in first quarter, Dan Persa throws interception returned 98 yards for a touchdown by Tanner Miller; Iowa WP before interception: .49; WP after interception and return for TD: .83; End result: +.34 WP for Iowa
This was indisputably the play of the game. Northwestern was in great position to score, with the ball on the Iowa seven and first and goal. This followed a very efficient nine play, 62-yard drive. Even assuming Persa had taken a sack at the Iowa 18 (as I did for my calculation here), Northwestern would still have been in decent position to win the game. A touchdown might have been an improbable (but not impossible) outcome from that point, but even a field goal would have left the game tied and essentially a toss-up in the WP model. Giving up an interception for a touchdown, however, was just a catastrophic mistake. It was a 10-14 point swing, it shifted the field position battle, and not surprisingly, that single play dropped Northwestern's chances of winning dramatically. So what happened?
Iowa experimented with quite a few defensive schemes to counteract Northwestern's spread offense, frequently throwing five and six defensive backs out on the field, spying Persa with a linebacker, and playing a lot more man-to-man press coverage on the line.* On this play, Iowa was playing a cover 1 with Miller in the deep safety role, Prater, Hyde, Bernstine locked up in man coverage on the receivers, and the linebackers in shallow zones (Donatell winds up running with the tight end into the end zone, and Kirksey and Nielsen just kind of mill around and watch Persa). This look is a little different from the usual cover 2 or quarters looks, but still a pretty standard zone. And it's not like the reason for the end result was anything all that complicated: basically, Persa screwed up. Northwestern's receivers couldn't get open, Broderick Binns had time to break free and get to Persa's ankles, and the guy threw a ball he shouldn't have. Iowa did well to cover the receivers and apply some pressure to the quarterback, but they needed Persa's cooperation to turn the play from a mere setback into a calamity.
* Nothing Iowa did on defense was all that successful at stopping Northwestern. The Wildcats frequently exploited man coverage by sending receiver Jeremy Ebert on choice routes that no one on Iowa (not even Micah Hyde) managed to cover. Spying worked on occasion (Iowa forced at least one third down stop this way in the first quarter), but also took a man out of coverage and left Northwestern with good man-to-man match-ups with receivers on linebackers. The basic story of the game on defense seemed to be that Iowa played the first half in a quasi-nickel, with 205-pound Tom Donatell as outside linebacker, and that approach did manage to slow Northwestern for a while. Then, in the second quarter, Northwestern realized what was happening and shifted to the run, rushing 10 times on their 13-play second quarter touchdown drive. Iowa adjusted in the third quarter by sending in freshman linebacker Quinton Alston for Donatell, then Northwestern adjusted back by returning to the pass.
Here's the video in real time:
2) Northwestern ball, 3rd and 8 from the Iowa 20, 14:07 left in second quarter, Iowa up 10-0, Mike Daniels sacks Dan Persa for -9 yards; Iowa WP before sack: .72, Iowa WP after sack: .77; +.05 WP for Iowa.
This sack by Daniels not only stopped Northwestern from converting on third down, it put them in a much worse position to make a field goal. Add in Northwestern's false start on the subsequent play, and this sack probably saved Iowa three points (Northwestern decided to punt).
3) Iowa ball, 1st and 10 from NW 47, 9:34 left in the second quarter, Iowa up 10-0, James Vandenberg completes pass to Keenan Davis for 47 yards and the touchdown. WP before: .86; WP after: .93; +.07 WP for Iowa.
This play followed up another play that was nearly as significant, Vandenberg's 31-yard completion to Davis on 2nd and 16 from the Iowa 20. That play was worth .05 WP in the generic model, which is pretty remarkable when you consider that it was worth almost as much as a 47-yard touchdown. Getting out from under that unfavorable down and distance/field position turned out to be a big deal. But getting back to the touchdown, what happened there?
(Video of this play is included in the YouTube video above at the 30 second mark). This was partially good play design and partially a big mistake by Northwestern's safety. On the Big Ten Network broadcast, they showed this play with a camera on the defensive backfield, and it turns out the play-action fake was not hugely successful in drawing the safety in. The big mistake was that the safety decided to cover the Iowa tight end (Derby) on an in route rather than sticking to his deep responsibility. That meant it was a foot-race between Davis and the cornerback, and Davis won that easily. You can see how the play is designed to force that kind of choice by the safety, though, and it could be that the play-action did just a little to confuse or delay his reaction. Plus, it does look like the cornerback was peeking in the backfield as the play started, which probably gave Davis a step.
As a whole, Iowa changed did seem to alter their offensive game plan to feature play-action pass quite a bit more (although I wasn't able to chart everything out), and a lot of that had to do with the presence of Brad Rogers. The no-huddle was (I think) completely gone, and the shotgun looks were greatly reduced. Iowa still used a fair amount of three receiver sets, but also a lot more two-running back, one-tight end sets than in past weeks. Part of that was that Iowa found itself able to run on Northwestern, part of it was that Iowa got up in the game and decided to run more, and part of it was that Iowa finally had their starting fullback back.
4) Iowa ball, 2nd and 2 from Iowa 42, 3:54 left in second quarter Iowa up 17-7, Vandenberg throws interception returned to NW 33. WP before: .86; WP after: .79; -.07 WP for Iowa.
This was Iowa going to the well one too many times. Another play-action pass, but Northwestern wasn't biting this time, and Vandenberg threw a pick right into zone coverage.
5) Northwestern ball, 4th and 1 from Iowa 30, 4:19 left in third quarter, Iowa up 17-14, Budzien makes 47-yard field goal. WP before: .41; WP after: .47; +.06 WP for NW.
I've harped a lot about Kirk Ferentz's conservative mindset with regards to fourth downs, field goals, punting, etc., but I do want to clarify one thing: if Ferentz is making mistakes in these situations, so are many, many other coaches. And in my opinion, Pat Fitzgerald made one here. Converting a first down at this point would have increased Northwestern's WP in the generic model by +.09 WP, from .41 to .50, while failing to convert would have dropped their WP -.08, from .41 to .33. Making a field goal would have improved Northwestern's WP +.06, from .41 to .47, but a make from that distance was no sure thing. A miss would have dropped Northwestern's WP -.08, from .41 to .33, just like a failed fourth-down conversion. But let's say Fitzgerald had reason to feel very confident in his kicker, and knew he could make a kick of that length 70% of the time (that's approximately the rate in the NFL). In that case, we can work out the break-even probability where going for it would be the better choice:
p*.09 + (1-p)*(-.08) = .70*.06 + .30*-.08
.17p = .098
p = .58
So if Northwestern felt they could convert on fourth and one better than 58% of the time, going for it would have been the better choice. It's not cut and dry, but neither was a 47-yard field goal. A fact that was demonstrated shortly thereafter ...
7) Norwestern ball, 4th and 11 from Iowa 29, 9:49 left in fourth quarter, Northwestern down 24-17, Budzien misses 47-yard field goal. NW WP before: .21; NW WP after: .14; -.07 WP for NW.
The play that set up this long field goal -- when Hyde tackled the pitch man on the option, leading to a seven-yard loss --wound up being a huge play. It forced a longer field goal, but just as crucially, it probably took any thought of a fourth-down conversion out of the minds of the Northwestern coaches. But should it have? That is, was attempting the field goal the "obvious" right call for Northwestern? Again, let's assume the same optimistic 70% success rate on the 47-yard field goal. The improvement in WP on a conversion on fourth and 11 in that situation would be quite large, .12 (from .21 to .33), while the cost of failure would be smaller, just -.07 (from .21 to .14). Making a field goal would have helped a little (+.04, from .21 to .25), but not as much as you would think (it would have left Northwestern still needing a touchdown to win the game, after all), and missing a field goal hurts just as much as a failed conversion (-.07). Without getting into the gory details (they're similar to the example above if you want to try it yourself), the break-even probability for Northwestern to go for it would be 40.5%. Could Northwestern have converted a fourth and 11 40.5% of the time? Maybe, maybe not, but the point is that even this play was a close call, despite the unfavorable down and distance. All of which goes to show that settling for field goals in this no man's land area (the opponent's 25-35 yard line) is a dicey proposition, even for NFL coaches with NFL kickers. For college coaches with college kickers, it's quite possible that the chances of making a field goal from 40-50 yards are worse than the chances of converting even quite long fourth down plays. Given that most college coaches opt for field goals (or punts) in this situation, any coach who decides to break the mold and leaves the kicker on the sideline stands to gain an advantage over his peers.
8) Northwestern ball, 2nd and 15 from the NW 25, Iowa up 31-17, 8:51 left in fourth quarter, Broderick Binns strips Dan Persa, Steve Bigach recovers fumble. WP before: .96; WP after: .99; +.03 WP for Iowa.
If this play had happened earlier in the game, it would have taken on more importance, but because Iowa was already in such a favorable position (up two scores with eight minutes left), it wasn't quite as decisive. Of course that interpretation is based on the generic NFL model, not the Northwestern-as-unkillable-Michael-Myersesque-horror-villain model, so it could be off a little. Recovering that fumble certainly set my mind at ease. And there was actually a very significant play after this one, when Marcus Coker picked up 15 yards on third and 27 from the Northwestern 37. Iowa had managed to squander great field position by dint of two consecutive penalties, and would probably have been forced to punt if Coker hadn't gained those yards. A 40-yard Mike Meyer field goal put Iowa up three scores and really increased the degree of difficulty of any potential Northwestern comeback.
8) Iowa ball, 4th and goal 1 from the NW 15, Iowa up 34-24, 3:56 left in fourth quarter, Iowa goes for it and gains 14 yards on a Vandenberg bootleg. WP before: .96; WP after: .99; +.03 WP.
This was a fascinating decision by the Iowa coaches. The choice was to either take a short field goal and go up by 13 (still a fairly safe lead), or go for the conversion and all the benefits that went with that (going up three scores with a possible touchdown and taking more time off the clock at the very least). I think the fact that Iowa went for it gives you some idea of how much respect Kirk Ferentz had for Northwestern's offense. The nightmare scenario must have been running through his head: Iowa goes up by 13 on a field goal, then Northwestern scores a touchdown, recovers an onside kick, scores another touchdown and wins by one point as 70,000 Iowa fans commit seppuku with improvised daggers fashioned from card-stunt cards. Iowa's coaches probably couldn't handle that kind of guilt, so they went for the conversion.
But was that wise? The general idea is to come up with the best strategies to win the game, but that could just as easily be a traditionally "safe" play as a "risky" play. So was the bootleg here the best strategy to win the game? It depends a little on what you assume the possible outcomes were. Do you assume that if Iowa converts, they just make it by the bare minimum, i.e. one yard? Or do you assume that on a high risk naked bootleg like this, it was basically all or nothing -- Iowa was either going to lose five yards and turn the ball over on downs or score a touchdown? I'm going to assume the latter for this analysis (I know, I know, Vandenberg didn't score officially, but he got so close that a touchdown was virtually guaranteed).
The WP of scoring a touchdown and going up 17 in the generic model in that situation is +.03, from .96 to .99, while the WP of losing five yards and turning the ball over is -.04, from .96 to .92. The WP of kicking a field goal, which I'm going to assume was a sure thing for simplicity's sake, was +.01, from .96 to .97. The break-even probability, then, for trying the naked bootleg works out to be .71. That's actually a pretty high number. For a coach who has decided not to go for it earlier in the year in situations where the break-even probability was something like .25 (I'm thinking of those punts in Penn State territory here), going for it here marked a real Walter White-like change in philosophy.
That whole analysis assumes, of course, that the percentages above accurately reflect Northwestern's chances of coming back and winning if Iowa had taken the safer route. It could be that going up 13 with four minutes left would not have made Iowa an iron-clad, 97% favorite. If that number was lower, then the whole analysis changes. What I take away from the decision are two things: a) as I said above, that Iowa's coaches must have really respected Northwestern's offense, and b) Iowa's coaches must have felt very shaky about their own defense, to the point that they were willing to take a sizable gamble just to end the game on one play. And given the quality of our defense so far this year, I can't say that was the wrong attitude to have.*
* But that naked bootleg... oof, that's a spicy meatball. It could have gone very, very wrong, like strip-sack-fumble-returned-for-a-touchdown wrong. Such a catastrophic outcome was not particularly likely, but running that play did put that possibility out there in a way that running a sweep or a dive or something would not have. Northwestern wasn't ready for it, but that may have been precisely because it was a little insane. That play seems like the kind of "Death Blossom"-style secret weapon that you use once then put away for a long, long time.