It goes without saying: Utah can't afford an Arizona sort of performance this week and have a prayer of defeating the Oregon Ducks. With one lapse in judgment or execution after another, the Utes last week let the Wildcats hang around... and hang around.. and hang around. It wasn't until the end of a grueling eight minute drive in the fourth quarter that the game truly felt out of reach for the Utes' opponent.
Yet going into that game, all of the advanced analytical signs pointed to total dominance—win probabilities north of 98%, three touchdown point spreads, you name it, the romp was on until it wasn't. So what happened?
The game wasn't, statistically, any closer than you would have expected. Arizona converted 41% of their third downs to Utah's 50%. They were held to about five yards per play compared to the Utes' 6.3. They gave up three sacks and seven tackles for a loss. So what the heck happened?
What happened is Arizona succeeded outside the boundaries of the models people like me use to predict results. Crazy stuff happens in football—stuff like fourth down conversions, crazy long field goals, and blocked punts for touchdowns. But it happens so rarely, and so close to randomly that a statistical prediction is MORE accurate when you pretend those 'wacky plays' don't exist.
Inside the boundaries of my spreadsheets and code, football is a different game. A team always punts on fourth down. No one ever tries a 57 yard field goal. Punts always flip the field, more or less. No one returns kicks or picks for touchdowns. It's all very organized and polite and nothing like reality for the players and coaches on the field.
We do it that way because experience shows that these kinds of wacky plays happen about evenly to either team, and that trying to predict them is a Sisyphean task where even perfect success would result in just a fractional change in average scoring margin. Take Utah's punt performance for example—it's been terrible, but almost all of their 32 punts have flipped the field as designed. Average the failures out across all of those tries, and the difference between Utah's punting and an average team's is a few decimals of a point.
The whole point of being a much better football team than your opponent is that you can afford to lose a bunch of wacky plays and still walk away with a clean victory. That's what happened to Utah last week—a blocked punt for a touchdown, a mammoth field goal, and 43 yards on a 4th down conversion added up to 16 'wacky' points, closing the expected 25 point margin to nine, still a couple scores from a Wildcats win.
That's not going to work against Oregon.
To win against the Oregon Ducks, Utah almost certainly can't afford to be on the losing side of wacky plays. 49% of the simulations I ran were a one possession game. Even if Utah plays well enough to win by five or six, as is projected, one blocked punt or missed assignment on a fourth down run play will be enough to tilt this result in the underdog's favor.
In my neat, imaginary football world, those plays don't happen and the Utes win a bit more often than not. The teams will each have to rely on their running games as their comparatively weaker passing offenses (based of the advanced analytics) face off against high-performing secondary units, but Utah's defense is more disruptive and will look to stymie Oregon's offense on a handful of critical drives.
Of course, our old friend reality will be ready to surprise us with every bounce of the football, whether one team likes it or not.
Utah 31, Oregon 26
