Looking back at the preseason polls once a season has ended is like looking at an old yearbook before your high school reunion. I really thought that hair style looked good? And what’s with that shirt? What was I thinking?
I really thought Tennessee was a top-10 team? How could I have believed that Ole Miss and Michigan State were going to be fine after what they lost following 2015? And really? Notre Dame?
College football is inherently and beautifully unpredictable, and we wouldn’t want it any other way. Still, we should be using every tool at our disposal to keep the humiliating recollections at a minimum. That includes computers. Now, to be sure, computers wouldn’t have saved you from Notre Dame, and they would have worked really hard to convince you that 2016 really was Ole Miss’ time to shine. But they would have told you not to waste your time with TCU or Iowa, and they would have told you that you should really think twice about Michigan State. They would have also told you to trade Stanford for Washington. That would have been a start, at least.
I’ve been maintaining the S&P+ ratings at FootballOutsiders.com since 2008. They represent an attempt to dig deep, deriving efficiency and explosiveness from the play-by-play level. They combine aspects of football that affect college football’s Five Factors — efficiency, explosiveness, field position, finishing drives and turnovers — into a single opponent-adjusted number. This system of ratings attempts to derive value from every play and drive of a given game (filtering out garbage time), and it watches every game, so to speak.
Each year, S&P+ produces results that make you doubt its creator’s sanity. But if you know what you are looking for, a system like this can be very useful. Among other things, it can drop pretty strong hints regarding who’s about to have a really good, or really poor, season.
What might seem like strange rankings could be reminders that the college football season is a small-sample exercise. Distill a college basketball team’s schedule down to four non-conference games and eight conference games, and you can easily produce mediocre records for great teams and great records for mediocre teams.
Last year, in this space, we discussed some potential statistical outliers for the 2016 season and listed six teams that might end up doing quite a bit better than their 2015 results would suggest.
Washington was one of those teams. “If you only pay attention to wins and losses, you may be a little bit confused by the Washington hype,” we said. “Sure, they beat USC and looked good in their bowl win, but really? Pac-12 North favorites? Beneath the surface, though, the Huskies were far better than your typical 7–6 team, and they were incredibly young as well.” Indeed, those close losses became close wins, and the easy wins got even easier.
Washington improved by five wins in 2016, and two other teams we listed — Wake Forest (plus-4) and Nebraska (plus-3) — drastically improved in the win column as well. Granted, two others (LSU and Tennessee) basically held steady, and one (UCLA) fell apart.
Still, that’s like hitting two home runs, a double and a sac fly in a baseball game while striking out once on three pitches. You’ll take it.
On the flip side, S&P+ projected AP top-20 teams Houston (53rd), Iowa (38th), TCU (31st) and Michigan State (22nd) far lower than the humans did. Combined, these teams won 22 fewer games in 2016 than in ’15.
What are the computers — and the S&P+ rankings in particular — saying about 2017? To answer that, let’s begin by once again looking at teams that may have either looked out of place in the 2016 rankings or might be due to surprise in 2017.