The Prediction Problem
November 18, 2013 by Corrie Trouw • Print Story •
Last week, Ohio State wide receiver Evan Spencer raised the eyebrows of college football writers and the blood pressure of his head coach by responding a little too enthusiastically to a hypothetical question about his team's national profile. Asked how the Buckeyes would perform against the BCS' top two teams, Spencer said OSU would "wipe the floor" with Alabama and Florida State.*
*On a side note: What was Spencer supposed to say about this question? Sure, we might question his humility or the upside of poking a pack of wild beasts, but isn't this what he is supposed to think? And doesn't any rational person already know this? So was the point of even asking this question just to exploit a less-than-savvy student athlete for media fodder?
Once undergarments were unbunched and unruffled, the story was eventually reduced to the meat of Spencer's analysis. How would Ohio State or any of the other potential BCS titleist compete against each other?
What a waste of time.
So often, we are fooled into believing single games reveal the true nature of a team. Team A beats Team B, and it's hard to imagine a world where Team B won. It's an easy mistake to make given small samples and little time for nuanced discussion.
Consider Saturday's Georgia/Auburn thriller. On fourth down and with less than a minute left, Auburn quarterback Nick Marshall heaved a desperation pass easily defended by two Georgia defensive backs. But in their zeal to end the game, the backs knocked the ball up and into the path of Ricardo Louis, the Auburn receiver who looked like he had the third-best chance of receiving Marshall's pass. It was the most influential of flukes as the path of that tipped ball essentially decided the outcome of the game.
Many games are close, and Auburn has no reason to apologize or discredit their win over Georgia. Whatever you want to award to the winner of a game, the Tigers should have full hold of it. And yet, many will graft bland narratives onto the result, uncomfortable to consider that we experience but one set of outcomes.
In a perfect world, teams would play long series where flukes play a minimized role in determining an outcome. But even just acknowledging that winner-take-all outcomes disproportionately award credit would be a major improvement.
As we watch the BCS abdicate to next year's playoff, it's critical to remember the fatal flaw that limited the current system's success. By focusing so much on the polls, the BCS forced voters into mental experiments of "who would beat whom?" Even entertaining this framework is a mistake.
The truth is, we can never really know how a game between two teams would turnout. We can identify skill sets and tactics and determine one side would need more fortune to find a path to victory, but sports, and especially football, are a collection of randomized experiments linked in series. Fortune is always woven into the final score.
All of which leads to one simple plea: Consider teams based on their resumes, nothing else.
The Eye Test. Momentum. Who Would Win.
Those are the subjective ways lazy voters evaluate teams. But like the work of psychics, palm readers, and other hucksters, they are smoke-and-mirror tools.
When it comes to picking teams to play in college football's biggest games, I don't care about recruiting. I don't care about the NFL draft. I don't care about last year. All that should matter is what they did on 12 or 13 Saturdays in the fall (We'll use "Saturday" loosely here to also represent certain Thursdays, Fridays, and other days found to be deserving of college football).
Unfortunately, the sport is not setup to reward this approach. During SEC media days this year, Steve Spurrier was one of many coaches to suggest he was interested in scheduling no more than 10 competitive (read: BCS-conference level) games per year. For most top programs, at least three of those BCS-conference games will be complete mismatches. We've now pared down sports' shortest schedule to approach a half dozen opportunities to prove who is any good.
The reasons why are obvious. With a buffer of just months, most fans and boosters won't remember against whom coaches padded their win totals. Fans come to the home stadium to see a win, even if it comes over a tin can. And undefeated is undefeated, even if the vanquished were filled with straw.
But we need to be smarter as fans and we need to demand smarter of our year-end selection process. The polls are broken. The baggage of years of SEC vs. the world has piled up. Everyone seems to have some kind of agenda and some kind of inane way of envisioning future performance. The sport is screaming for objectivity and standard evaluation.
The NCAA basketball tournament, we're told, does this kind of analysis. "Who did you beat?" and "Where did you beat them?" are worthy criteria. Let's see it on the gridiron as well.
We can forgive Evan Spencer for not bringing this rationale to an off-hand comment about his own team in the face of a BCS snub. But without this soon-to-be retired system to kick around anymore, what will be our new excuse?