Filed under:

# Statistical In-Ferentz, Week 13: Pyrite

Every football coach has to be, in part, a clairvoyant. He has to have an intuitive feel for where the game likely will be three quarters from now when he acts in the first quarter, and adjust his decisions accordingly. To take an example, imagine a coach of a team with an average defense and offense playing a team with a high powered offense, like Oregon or Wisconsin. An erring coach might look at the first quarter scoreboard, see that he is only down 7-0, and think "this is a close game -- I must play conservatively to keep it close." A wise coach would look at the same score and think "I'm already behind and worse still, that score can't last -- this team is sure to score at least 30 more --  so I need to be playing with a sense of desperation right now." It turns out that a sense of the future of the game is critical to making decisions in the present.

What's that have to do with the Nebraska game? In a couple important respects, it seems that the Iowa coaches did not have a clear understanding of what kind of game it was likely to be. Most obviously, they did not play with any sense of desperation once the score got to be 10-0 late in the second quarter. They did not choose to try for a score in the last 30 seconds of the half, even though a win probability* analysis of the game (and simple intuition) would indicate that Iowa was already a heavy underdog at that point, with only a 20% chance of winning in the generic NFL model. There were a few other non-desperate decisions in the second half -- not going for it on 4th and 2 from their own 26 late in the third quarter and punting when they were down twenty -- but in truth the game was all but decided fairly early in the third: by the time Nebraska kicked a field goal to go up 13-0, the Cornhuskers were 91% favorites in the generic model.

* Win probability figures derived from the Advanced Football Stats generic model; caveats, college football is not pro football, yadda yadda yadda, etc. -- see earlier columns or explore the Advanced Football Stats website for more on the concept.

In fact, you could argue that Iowa needed to be playing with a sense of desperation from the very start of the game, for several reasons. first, they were on the road; second, Nebraska was the better team by most measures; and third, it was clear early that Nebraska was having more success on their drives than Iowa. This fact was masked partially by two critical penalties Nebraska incurred, both of which effectively ended early drives, but you could already see in the first quarter which team would be racking up long carries all day and which team would be struggling with incomplete passes. It's interesting to note, by the way, that some of the most positive plays for the Hawks all days were those early Nebraska penalties:

As it turned out, Iowa's best opportunities for playing desperately also happened early.The drive (or lack thereof) that really sealed the game was the golden opportunity Iowa had starting from the Nebraska 48 after a shanked punt by Brett Maher. If Iowa was going to score points, they would get no better opportunity than this. Two incomplete passes left Iowa with a 4th and 9 from the Nebrasksa 36, exactly the no-man's land where punting ought to be forbidden. The decision in terms of WP was pretty clear: Iowa stood to move from .50 WP to .63 WP with a conversion, while a failure would drop them to .46 and a punt to the Nebraska 9 (the punt they actually got) would leave their position unchanged at .50 WP. The break-even probability to go for it there turns out to be just 23.5%. A nine-yard play is by no means assured, but, really, getting that much just a quarter of a time is not unthinkable. Even on a bad day, James Vandenberg completed 46% of his passes.

There was a similar opportunity on a long fourth down on 4th and 13 in the second quarter. Iowa had the ball at the Nebraska 38 after an incompletion, a cringe-worthy Marvin McNutt receiver screen, and a Vandenberg scramble for no yards. Making those 13 yards wouldn't have been easy, but because Iowa was already down three points and because the eventual alternative -- a touchback on a punt -- was hardly better than a turnover on downs, this winds up being a very reasonable bet. Converting would have increased Iowa's WP in the generic model from .39 to .52, while a punt to the 20 would actually drop Iowa's WP from .39 to .38 and a turnover on downs at the 38 would drop them to .35. In this case, the break-even percentage winds up being the same as the previous one: 23.5%. It would not have been an easy call, and it might well have failed, but the point of this analysis is that the downside was really not all that much worse than the downside of punting. When you consider the effect of giving the ball back to Nebraska's punishing run offense, any chance to keep possession becomes even more attractive.

You could also argue that Iowa missed a golden chance to go for it with 4th and 1 from their own 34 late in the first quarter. It's not one of those clear cut, no-man's land calls -- the cost of giving Nebraska the ball that deep in one's own territory is pretty hefty -- but it's exactly the kind of thing a desperate team would do. The break-even probability works out to be 50%, which seems doable for a one-yard gain. I can understand why coaches would be reluctant in most situations to take that kind of gamble in their own territory, but there's one last reason that should have predisposed Iowa to early desperation: the nature of Nebraska's offense.

Iowa was playing a somewhat risky game on defense early it seemed, keeping their linebackers close to the line and sending Jordan Bernstine down into the run game frequently (he wound up with something like 13 tackles). This proved somewhat effective early at slowing down Nebraska's run game, but set Iowa up for some deadly play-action passes. You can see this by looking at the most negative plays for Iowa during the game:

If I had to guess Iowa's general defensive strategy, it would be something like this: sell out against the run early, give the ball to the offense, let them create a few long drives that end in touchdowns and give Iowa enough of a lead that Nebraska is forced to pass and we can run a more balanced defense. Given the problems Iowa's had stopping the run in the past, I can understand that philosophy. But the key part of that defensive strategy was actually in the offense's hands: long drives that end in touchdowns. Without those, Iowa's defense was doomed in two ways: to the inevitable play-action pass, and to fatigue. And ultimately, the team was doomed in another way by the offense's inability to stay on the field, because Nebraska's run-first strategy ate up time and left Iowa with less time to mount a comeback. Given that Nebraska is known to be a run-first team and that Iowa's defense is especially weak against the run, it was not hard to predict that Nebraska would attack Iowa with the run unless Iowa forced them in some way to abandon it, namely by getting ahead in the game. And since getting points early was such a critical part of keeping the defense rested and forcing Nebraska away from its strength, Iowa should have been jumping at any and all early opportunities to put points on the board.

So yes, there were times when it seemed like Iowa was not playing with enough of a sense of desperation. When Iowa punted down 20 in the fourth quarter, for instance, I checked out.*

* There is actually a pretty amusing WP analysis for this situation. Basically Iowa had no chance of winning at that point, fourth down or no, so I don't really care all that much what KF and co. did here. They had a 1% chance of winning in the generic model going into that fourth down, and that may just be because the model doesn't go lower than 1%. If they improbably had converted, their chances would have gone up to 2%, but if they had failed, their chances would have stayed at 1%. And if they punted, their chances would have ... stayed at 1%. So basically Iowa had nothing to lose by going for it at that point. I'm not all that comfortable using the generic model in cases that test the extremes of the model like this, but it makes intuitive sense: if you are down a gazillion points and have slim to no chance of winning, there is practically no gamble that is not worth taking, going for it on 4th and 21 included. Unless, that is, your goal is not maximizing your chance of winning, but rather minimizing the average final margin of a loss. If the latter was the goal, then Iowa did the right thing by punting.

But at the same time, I can't say that lack of strategy lost Iowa the game. What lost Iowa the game was the fact that Nebraska was the team that was getting 4, 5, 6 8 and 14 yards on first down, while Iowa was the team that was throwing incomplete passes and gaining one or two yards on first down. There's a stat that a website called Football Outsiders uses, success rate, that is helpful here. Success rate rates plays as "successes" if they gain 50% of the necessary yardage on first down, 70% on second down, and 100% on third down (and fourth down). You can go through all of a team's plays and see how often they were successful at getting the necessary yardage to stay in a position where converting the next first down was probable. Success rate by itself doesn't really tell you how potent an offense is -- a team with two failures and a 90-yard success might wind up looking pretty lousy by the stat -- but it does give you an indication of how steady a team is at gaining necessary yardage. If we just look at the game up to the point when Nebraska went up 20-0, which is when it was effectively over and before Nebraska started running up the gut on every play and Iowa scored a late charity touchdown, here's how the two teams rate:

• Nebraska: 78 plays, 38 successes, 48.7% success rate
• Iowa: 36 plays, 12 successes, 33% success rate
For one, this shows you the sheer advantage Nebraska had in time of possession, but it also shows you just how mightily Iowa was struggling at getting the necessary yardage to get in positions where first-down conversions were probable. The simplest explanation is that, on that day at least, Nebraska had the stronger offense and the stronger defense, and that, if you played the game 100 times, Nebraska would probably win most of them.

I've looked a lot at strategic situations this year, and one of the things I think I've learned is that it's very tempting to think that strategy is everything in football -- and it can matter tremendously! -- but that ultimately, strategy is a secondary concern. To see this more clearly, imagine the following: you are assigned to be the coach of a team of high school sophomores, and your first game is against the Dallas Cowboys. What strategic decisions would you make? Assuming that you couldn't just yell "run away" and get those poor kids out of their for safety's sake, there would be strategic decisions you could make that would, by the book, increase your chances of winning. But they would be decisions made from absolutely terrible starting conditions, e.g. "what's the best thing to do when you're down 28-0 in the first quarter, your team is facing 4th and 67 from its own one yard-line, and DeMarcus Ware has just maimed your starting quarterback?" I suppose there's an answer to that question, but it doesn't really matter.

In an odd way, strategy only starts to matter when your team starts to approach its opponent's in general quality: if you are much worse, strategy won't matter much, and if you're much better, it won't matter either. Only in the mid-range does it become crucial, suggesting that strategy is a contingent force in football, whereas skill and speed and strength are the fundamental factors that must pre-exist to allow strategy to play a role. For a team that is in the realm of competent football but on the mediocre-to-bad end of things, though, strategy can represent a special kind of fool's gold, whispering that "if only we had made the right decisions, we would have been good." This is precisely backwards. The right thing to say is: "if only we had been good, we wouldn't have had to make so many tough decisions." Sure, I wish Kirk Ferentz would be more daring in certain situations (although he did make several good fourth down calls this year), but what I really wish is that Iowa would be up 28-7 in more games and we would all be too drunk with happiness and, presumably, alcohol, to care much about strategy.