Now that the 2017 season is in the books, I guess we’re expected to jump immediately to previewing 2018. The way-too-early rankings were right on time – moments after the final gun on the final game in Atlanta – and while I generally look at all of these I know the ingredients going into the stew aren’t really in season. We can approximate the talent returning, maybe shade a little recruiting in there and the schedule’s the schedule (though as these rankings prove we’re always going to miss on how good some teams are or aren’t no matter what it looks like on paper).
So how about if we just remove all of that and focus on which teams over- and underperformed in 2017? I have the Pythagorean win totals calculated going back to 2007, and looking at the year-end numbers is also one of the first things I do once the season is over. While the predictive value of this method is continuously debated by people much smarter than I, at the extreme ends of the spectrum there does seem to at least be some correlation between over/underperforming one year and being better/worse (in terms of wins) the following year.
Between 2007 and 2016 (1,224 individual seasons), teams that had a Pythagorean expectation that was 1.6 wins or more greater than the actual win total improved by at least one win 76.3 percent of the time. On average, those teams improved by 1.96 wins.
The correlation isn’t quite as strong on the other end, but teams that were 1.6 expected wins or more below the actual total had a lower win total the following year 68.6 percent of the time and averaged 2.2 fewer wins.
That’s what we’re looking at. Is it elegant? Not when I’m doing it. I’m not smart enough in these matters to be elegant, but it is interesting. So here we go.
Pythagorean expectation had a good year on the good end in 2017. Of the eight teams in 2016 that were at least 1.6 expected wins in the black, seven of them improved. It was a star-studded list, too, featuring some major players and teams that became trendy as the season went along: Notre Dame, Auburn, Iowa State, Michigan, Michigan State, Missouri, UCF and LSU. (Miami just missed the cut.)
This year’s group isn’t quite as exciting:
Five of the eight teams that were likely better than their records showed last year had four wins or fewer. No disrespect, but if Tulsa gets a little better does anyone outside of Tulsa notice? Probably not, but if you happen to find yourself at a bar in Ypsilanti over the next eight months and want to make friends by saying, “You know, Eastern Michigan was really unlucky to go 5-7 last year,” go ahead and do it. It’s a niche audience, but the numbers back you up.
Of the Power 5 teams, Baylor had four one-score losses. Strangely, two were to Liberty and UTSA and two were to two of the best teams the Bears played, Oklahoma and West Virginia. North Carolina is one that occupies a sort of middle ground. The Tar Heels could get back to bowl eligibility relatively easily and maybe upset some team they shouldn’t along the way. That would constitute progress without necessarily exciting anyone.
But the two big names here are clearly Texas and Virginia Tech. The formula says the Longhorns should’ve won nearly nine games this season. If you, like some vocal Texas fans, were underwhelmed by Tom Herman’s first season, just wait. Virginia Tech is also poised to get even trendier after winning 19 games over Justin Fuente’s first two seasons. I have no idea what the Hokies return overall – remember we’re trying to keep that out for right now – but I do know that quarterback Josh Jackson had a pretty great year for a freshman. He should be better in 2018, and as the quarterback play goes so goes the team’s ceiling. (Most of the time.)
On the other end, these were the teams that were at least 1.6 expected wins in the red, and that group includes two opponents on Nebraska’s schedule in 2018.
Akron, the Huskers’ opponent in the opener, was nearly two full wins better than projected. The Zips had two one-point wins (sorry Western Michigan and Buffalo) and a three-point win over Ohio, which put Akron in the MAC title game, where it was smoked by Toledo. The Zips then got rolled by Florida Atlantic 50-3 in the Cherry Drink Bowl. As people get into looking at Nebraska’s schedule, remember this when they mention Akron as a bowl team that played for a conference title.
Michigan State is the other one, which isn’t a surprise. The Spartans really ride this particular rollercoaster. The 2015 team that went to the College Football Playoff won 12 games with an expectation of 9.5, so a downturn may have been expected the following season. Nobody would’ve projected three wins, but that’s what the Spartans got in 2016. Those three wins looked more like five, however, so Michigan State ended up on the good list in 2017 and jumped back up to 10 wins. Can’t wait to see if it maintains this on-off rotation. Nebraska hosts Michigan State on Nov. 17.
I’m guessing you noticed UCF on this list. That happens to teams that are perfect. The only way to end up with a Pythagorean expectation of zero losses (and thus 13 wins in UCF’s case) is to give up no points at all. Since that’s almost impossible, an undefeated team can only “underperform” based on this method but the degree to which the Knights come in under their win total is a reflection of four one-score wins (SMU, USF, Memphis, Auburn). UCF didn’t “underperform” by any means, but it did have four close to coin-flip games (though some coaches seem to have a knack for winning those). If the Knights do take a downturn, and anything less than 13 wins would technically be that, it’ll probably be more about losing Scott Frost than about randomness catching up to UCF.
What about Nebraska? The Huskers are essentially as good, from this perspective at least, as the record showed in 2017. The team won four games, the numbers said 3.7. That’s a little to the negative side, but not significantly so. And even if it was worse, you’d probably have to throw it out anyway with the coaching change.
That’s why those things happen.