It’s possible this is just my impression, but announcements in recent days from Notre Dame, Oklahoma, and South Carolina, among others, seem to have the arrow pointing more towards college football happening as planned than its pointed since Rudy Gobert’s coronavirus test came back positive. Schools are planning to have students on campus, and not just a few students, but a quorum of the student body. Schools appear confident in their resources, and confident enough in their understanding of the virus to be, in the cases of Notre Dame and South Carolina, weighing rather specific risks—Fall Breaks, when masses of students traveling could presumably cause significant spread; and the possibility of the virus surging alongside influenza and common colds as winter begins. Obviously, there’s a lot that’s yet uncertain; and we shouldn’t forget how fast things can change, having learned that lesson so forcefully in March; but it appears fair to say that the likelihood of college football happening in a rather conventional, nationwide fashion this fall is, indeed, the highest it’s been in two months.
This provokes some questions. Of course, there have been questions all along, but with the “will college sports even happen during the fall semester?” question leaning a certain way, the other questions are coming into clearer focus. Personally, this is the first time I’ve felt it worthwhile to formally spell out those questions for myself as I figure out how much time I should be spending on The Barking Crow’s college football model and similar projects. Since I’m spelling them out for myself, I might as well spell them out for our readers too.
1. How Much Postseason Will There Be?
There are concerns about a “second wave” of the virus in the winter (as an aside, it’s unclear whether this would really be a second wave, or simply an increase from what might be currently turning into a plateau—I’m no epidemiologist, but I’m hesitant to use the phrase “second wave” without that addendum). Travel is an issue. There are vague liability incentives for schools to have some time with campuses empty between the fall and spring semesters.
It’s not hard to imagine bowl games being canceled. Perhaps not all, but perhaps all, or perhaps all but the College Football Playoff. It’s also not hard to imagine scenarios in which the country goes full steam ahead with bowls, or begins with that plan and slowly, or quickly, adjusts as the dates approach.
Here at The Barking Crow, we need to decide what to do in each of these scenarios. Our models’ biggest strength, to date, has been predicting the actions of selection committees. If there is no College Football Playoff, we could pivot to trying to predict the AP Poll’s national champion, but frankly, that’s something FiveThirtyEight would be better at, given their heavy analyses over recent years of the horse race that is college football polling. If bowls are in jeopardy, we’d have to pivot away from our goal of giving team-specific bowl likelihoods for every bowl and every team in an easy-to-navigate format that includes conditional probabilities based on the results of each of a team’s games (this is the goal—we aren’t sure yet whether we could get there this year even without a pandemic inserting so much uncertainty into the equations). In short, we don’t really know at this point what exactly we’re modeling.
2. How Fluid Will Scheduling Be?
It appears highly possible that there’ll be coronavirus breakouts on some college campuses this fall. Putting even this aside, states and localities may “lock back down” in the face of rising case numbers at a point. Some game-cancellation uncertainty already exists each year thanks to thunderstorms and hurricanes, but it’ll be heightened this season, and while those weather-related threats are most prominent in the season’s first weeks, the coronavirus-related threats seem, on the surface, to grow more significant as the season goes on and the leverage of individual games grows.
For our model, we’ll have to decide how to incorporate such uncertainty, and whether to incorporate it at all. It would be palatable, and a lot simpler, to leave it out, but the more accurate thing to do would be to attempt to incorporate it somehow. We’ll have to see whether we can figure out an empirical way to do that.
3. What Happens if Scheduling Becomes Regionally Restricted?
There’s been some talk in recent days regarding Alabama and USC’s scheduled season-opener at Jerry World and whether California’s restrictions imperil the game’s feasibility. The question is larger than that specific game, of course—if California says no to college football, it’s likely Oregon and Washington will as well, and that Hawaii already will have already done so, cutting the Pac-12 and Mountain West ranks by eight and four, respectively. If more western states make restrictions, those numbers keep dropping.
California is an interesting case specifically for a few reasons. Its governor spoke positively in recent days of the possibility of professional sports resuming soon, but the state’s been the tip of the spear in the Name-Image-Likeness battle with the NCAA, suggesting it is very cognizant of differences between professional and amateur athletics. It’s been aggressive in its containment measures of the coronavirus, and that aggression seems to have paid off, but it’s still home to four of the twenty largest metro areas in the country, and while none are as dense as New York, nowhere in the United States is as dense as New York. Will it continue with the aggressive approach? Will it have a choice?
This binary situation in which some schools cancel their seasons and others play would be easy for us to model—once we knew which teams weren’t playing and knew the schedules for those who would, we could proceed rather normally, and presuming the Pac-12 and Mountain West cut a deal to associate for a year, we could make some educated guesses as to how the College Football Playoff committee would view teams from the hybrid league, getting more concrete answers with the first CFP Rankings of the year.
But what if travel, and not simply playing the games, is what’s limited? What if nonconference games are eliminated in a substantial number of cases? What if athletic departments confine themselves to playing schools within their own region?
College football’s normal nonconference slate makes it possible to create something called a Bayesian network where each team’s results are connected through common opponents. Sometimes these are second, or third, or fourth-degree common opponents, and more, but still, they’re connected one way or another, and through multiple channels. In a suddenly-regional sport, these nationwide Bayesian networks would break down, meaning some rating systems and metrics would have to be reconfigured, including some likely used either directly or indirectly by the selection committee.
Our model this past year actually wasn’t very Bayesian itself—there were a few elements that were, but those were few. This wasn’t the result of any rigorous decision, though. It was done due to the constraints of time and other resources. It’s not essential to our model to have a full Bayesian network, but it might be helpful, which means we’ll likely have to set up a few contingency plans for varying degrees of regionality until we get specific plans from schools.
4. How Will Other Sports Be Impacted?
The presumption has sometimes been that colleges won’t just bring back football and leave the other sports unplayed, but it’s worth questioning whether this would really be the case. Yes, it would be bad optically in the eyes of some, but how much energy would the media really expend on highlighting the matter, and would consumers care? Yes, certain approaches in this manner might be illegal under Title IX, but how willing would certain schools be to risk a lawsuit? It’s not a big secret that college football makes money and many other collegiate sports don’t. Some schools may actually need football to remain solvent financially.
The biggest question for The Barking Crow is men’s college basketball, given our NIT-blog roots, but we’ve tentatively planned to roll out models for ten more collegiate sports this upcoming academic year, including men’s and women’s soccer, women’s volleyball, women’s basketball, and men’s and women’s ice hockey, all of which see their seasons begin in the fall semester. At some point, uncertainty around these seasons happening or the format in which they happen might make us decide to scrap some or all of the new models. For the time being, we’re watching all the news come in, and hoping safe decisions are made one way or another.
