The Paradox of the Northwestern Offense
By Eric Rynston-Lobel
One of the main talking points after Northwestern’s 24-15 loss on Saturday at Wisconsin wasn’t how well the defense played or even how out of sync the offense looked for the third time in four games, but instead how Pat Fitzgerald went for two-point conversations twice—after the ‘Cats scored touchdowns to make it 24-9 and 24-15. Based on the final score, you probably already figured out that both were unsuccessful.
Another main talking point from the loss was Northwestern’s inability, or more precisely its reluctance to stretch the ball down the field, and the continuing theme of the ‘Cats being overly conservative on offense. Interestingly, though, going for two after scoring a touchdown is a strategy most teams don’t employ. So how do we make sense of an offense that operates conservatively and aggressively? We turn to psychology and statistical probabilities.
Risk Aversion in the Context of Football
In the late 1970s, psychologists Daniel Kahneman and Amos Tversky conducted research on how people’s decision making tends toward the side of risk aversion. What they found is that even if the outcome of an event was going to be the same, how that outcome was presented profoundly impacted the choice the subjects made. For example, imagine the following situation:
There is a disease that puts 900 people at risk and you can choose between two options:
- Guarantee that 300 people are saved
- 1/3 probability that all 900 will be saved and a 2/3 probability that no one will be saved
In this case, people are more likely to choose option 1.
Now, consider these scenarios:
- Guarantee that 600 people die
- 1/3 probability that no one will die and a 2/3 probability that all 900 will die
In this case, people are more likely to choose option 2.
But wait a minute, if we do the math, both sets of options are exactly the same. However, because people are generally risk averse, they tend to choose option 1 when the situation is framed as saving people and option 2 when the situation is framed as people dying.
You might be asking yourself how this applies to football. University of California, Berkeley Economist David Romer published a paper in 2006 titled “Do Firms Maximize? Evidence from Professional Football.” Rather than two-point conversions (which I’ll get to in a minute), Romer focuses on NFL teams going for it on fourth down instead of punting the football or attempting a field goal. After rigorous calculation through droves of data, he finds that in most situations, it increases a team’s chances of winning the game if they go for it on fourth down. He finds a 2.1% increase in a team’s win probability if they always go for it on fourth down, which would result in approximately one more win every three seasons. That might not seem significant, but according to Romer, it is if we consider the amount of money invested in players, coaches, etc.
Last season when Northwestern won the Big Ten West, they went for it on fourth down 32 times in 14 games and converted on 12 of those attempts for a success rate of 37.5%. Only Florida Atlantic kept the offense on the field on fourth down more frequently last season. So in that regard, Fitzgerald was not risk averse. For good measure, the ‘Cats have gone four of 10 so far this season which puts them slightly ahead of the pace they had last season and underscores the program maintaining its aggression in that department. Romer’s analysis backs up Northwestern’s frequent decisions to keep the offense on the field. He says that most “teams’ actual choices are dramatically more conservative than those recommended by the dynamic-programming analysis.” But NU is not one of those teams.
Last year in the NFL, teams went for it on fourth down on average 16.8 times. Over the course of a 16-game regular season, that’s just over one attempt per game, something Romer would argue is outrageous. But it nevertheless supports the notion that coaches are generally incredibly risk averse when deciding whether to go for it on fourth down.
Granted, Romer’s research focuses on the National Football League, not college, but even still, it’s not too far-fetched to assume his argument would translate. In fact, it might be even more useful to go for it on fourth down in college because teams’ kickers typically do not have the same range that NFL kickers have. This would come into play because in the NFL, a team might opt for a field goal at its opponents’ 35-yard line, but in college, that might be too far for a team’s kicker, so it would make more sense to keep the offense on the field.
However, most college kickers are capable of making the point after touchdown (PAT) which is the equivalent of a 20-yard field goal. More specifically for NU, junior kicker Charlie Kuhbander is 76 for 77 on PATs in his career which made the ‘Cats decisions to go for two this Saturday all the more interesting.
I’ll be the first to admit, when I saw the offense stay out after Drake Anderson’s touchdown to narrow the deficit to 24-9, I was very confused. Being down 24-11 instead of 24-10 still means you need at least two more possessions to tie the game.
An alternative rationale was perhaps that Fitzgerald didn’t want the game to go to overtime—the ‘Cats would’ve won 25-24 if they converted to make it 24-11 and scored two more seven-point touchdowns while holding the Badgers scoreless. But the way I viewed it was that I’d much rather lose the game 27-24 in overtime on the road in a game the ‘Cats were 23.5-point underdogs rather than 24-21 because my team failed on three two-point conversion attempts.
In the postgame, though, Fitz defended the decision making:
“When you’re down 15, you go for two because those are the analytics that we pulled to try to shorten a game, have less possessions to win the game.” Fitzgerald said. “Those are things that we talk about. It’s on our chart…We work with some analytics companies on two-point plays and that was a decision that you can’t ever know what’s going to happen next, so you have to assume we got to score as many points as we can right here right now.”
Based on a FiveThirtyEight diagram that displays the optimal situations to go for two, Fitz was correct to go for two according to this chart. The first time, the ‘Cats trailed by 15 and the second time, they were down by nine.
But here’s the problem I have. Northwestern clearly has a plan on how they want to handle going for two which is reassuring. What’s less so is the fact that the offense has struggled mightily this year, especially in short yardage situations. Just last week, the ‘Cats failed to punch the ball into the end zone on second, third and fourth and goal plays from the Michigan State goal line—they needed less than a yard on three plays, and they couldn’t execute. Perhaps an availability heuristic is skewing my evaluation, but Saturday’s 0 for 2 performance on two-point tries further highlights the offense’s ineptitude in those type of situations.
This brings me back to the paradox I allude to in the title: in general, especially this year, Northwestern’s offense operates incredibly conservatively—except when it comes to fourth down and going for two when they are risk takers. To put some numbers on that, the ‘Cats rank 125th out of 130 FBS teams in yards per play this year at 3.8. While that might not perfectly capture “conservative play calling,” it’s hard to quantify a qualitative attribute. The point is, in most aspects of the offense, the ‘Cats operate very conservatively compared to the rest of college football. But when it comes to going for it on fourth down and going for two point conversions, NU is the polar opposite of conservative.
I’d be the first to say I’d rather the opposite be true. But it’s not. To put a positive spin on it, if the offense can turn it around (which is very possible given how strong the defenses on Michigan State and Wisconsin have been this year) and continue to be aggressive on fourth down and two-point tries, that should only increase the ‘Cats chances of winning some games—maybe even one or two that they wouldn’t have if they were completely risk averse.