This is, of course, a big part of what makes college football so fun. Between recruiting, spring football, a long offseason, and a full week between games during the season, more time is spent on evaluation, analysis, and predictions than on actually watching or participating in the games themselves.
That said, it’s usually better to start the year with realistic expectations than with the unreasonable optimism most fanbases (encouraged by many media members) bring into every year. Reasonable expectations leave room for the variance caused by things like luck and injuries, understanding that sometimes the better team doesn’t win. That’s why they play the game, after all.
For those of us who played sports ourselves, this involves a difficult transition in our thought process. Every team and player should go into every year believing they can and will win every game. If you don’t believe you can win, you’re pretty much guaranteed to be right. But the expectations of fans and media really don’t affect the performance of the team (as much as we might often feel otherwise), and realistic expectations often differ from realistic goals.
For example, it is reasonable—and should be expected—for Alabama’s coaches and players to set a goal for a third straight national championship this year. But it is unreasonable for anyone outside their team to expect them to meet that goal. Alabama has as good a shot as any team to win it all, but their odds of not winning the title are significantly higher than their odds of winning it. Thus when setting expectations for the year, a healthy understanding of probability and variance is especially helpful.
That is where most fans and media members go wrong when assessing an upcoming season. When looking forward to the season, nearly everyone produces a projection that looks something like this:
@ South Carolina: L
@ Georgia Tech: W
@ Virginia Tech: L
Boston College: W
@ N.C. State: W
@ Pittsburgh: W
Old Dominion: W
Final Record: 10-2 (7-1 ACC)
The problem is that this sort of projection does not adequately account for the statistical variance of college football. Important players—like Shakeel Rashad—get injured. Footballs take weird bounces. Sometimes the defense tips the ball into the air and intercepts it. Other times the tipped ball gets caught for a touchdown. Upsets happen. The result is that fans of good teams consistently tend to overestimate their team’s final record, setting themselves up for disappointment.
A better approach is to apply the concept of win shares to a season, establishing a reasonable probability for each game and then adding these probabilities together to establish a final expected win total. Each game is thereby treated as an independent trial with its own odds (like a coin flip or a single hand of poker) and the season as the sum total of those trials, thus accommodating for variance.
One standard way of setting odds for each game is to look at the point spreads. To some degree, that’s what those guessing wins and losses at the beginning of the year are doing also. But most don’t realize how often even sizable favorites lose: 14-point favorites only win approximately 80 percent of the time. This means that in terms of expectations, a team that is thought to be two scores better than its opponent on average actually loses 20 percent of the time.
The key from here is to understand compound probability. The odds of an equally-weighted coin coming up heads is 50 percent. But the odds of it coming up heads twice in a row is only 25 percent (50% x 50%). Compound probability suggests that over the course of a football season, even if a team is a 17-point favorite in every game—giving it a 90 percent chance to win each individual game—that team still has only about a 28 percent chance of going undefeated (90% probability ^12). That is why it is rare even for an outstanding team with an easy schedule (like Boise State the last few years) to go undefeated, and why unbeaten seasons are so rare in general.
Expectations for 2013
Given an understanding of compound probability, what might realistic expectations look like for this UNC season? The current betting lines produce an over/under mark around 9.5, as shown here. But I think we can get a better estimate than that, as gambling lines are often skewed somewhat due to a team’s popularity or the level of respect for a team’s conference.
Starting from a database of the recruiting rankings of each player still on the depth chart, past percentage results of games between teams with comparable talent gaps, and a preliminary assessment of how teams match up schematically, here is what I see as a realistic outlook for the Heels in 2013, with the column on the right representing the odds of winning that game and each number totaled up at the bottom:
Those numbers put the most likely overall result at 9-3, followed by 8-4, with a most likely conference record of 5-3 and a shot at 6-2. I’ve done the win shares for every other team in the ACC and SEC, and according to my numbers, North Carolina is (by a very small margin) the most likely winner of the ACC Coastal Division.
As you can see, my percentages differ somewhat from that of many other prognosticators, particularly for that first game against the Gamecocks, which I think Larry Fedora’s bunch has a very good chance of winning. (In fact, I’d likely favor UNC were the game played in Chapel Hill.) Based on raw numbers, last year’s games between teams with the relative talent gap between UNC and South Carolina put the odds of a Tar Heel win at about 40 percent, but I think the chances are slightly better than that due to matchups, experience, and scheme considerations. I will cover my reasons for that in my forthcoming South Carolina preview articles, but my next column will discuss what it means to have balance in a Larry Fedora offense.
Regardless, the win shares approach is an improvement over the usual manner of setting expectations, both alleviating possible disappointment and helping to appreciate overachievement. This does not mean this Carolina team will not win ten games; that is within the reach of this team. But it does mean that if the Tar Heels finish the regular season with ten wins, it should be understood as exceeding rather than meeting expectations.
Jason Staples was a walk-on wide receiver at Florida State in the early part of the last decade and has provided in-depth football analysis on the Scout network since 2007. A member of the Foootball Writers' Association of America, he is presently finishing a Ph.D. at UNC-Chapel Hill and will be providing scheme and stats analysis for InsideCarolina in 2013.