Why advanced stats will undervalue Oklahoma
Geeks won't think the Sooners are that sweet. (Image courtesy: kjrh.com)
A few years ago, SBNation’s Bill Connelly, the guru of college football advanced stats, coined the term "West Virginia Effect," a CFB-ized derivative of what is referred to in behavioral finance as “recency bias.” In Connelly's words:
Every year, we witness the West Virginia Effect, in which a team that looks amazing in its bowl game gets increasingly ridiculous hype as the long, arduous offseason plods along. You forget about the shortcomings the given team had during the regular season, and you just remember the perfect final act.
Who does that sound like in 2014?
With preseason preview magazines starting to hit newsstands, the punditry is doing its part to build up the Sooners’ Sugar Bowl win over Alabama and make OU this season’s It Team. The Sporting News already has pegged OU No. 1 in the nation. Athlon has OU checking in at No. 4, which would earn a spot in the new postseason playoff.
The nerds tend to view such offseason hype with appropriate skepticism, noting that their finely tuned metrics don’t share the same level of enthusiasm for these overheated stocks. The 2011 West Virginia team, for example, ranked 19th in F/+, a measure of efficiency maintained by Connelly and number-crunching partner Brian Fremeau for Football Outsiders. The Mountaineers kicked Clemson’s ass all over the field in the Orange Bowl at the end of the season, though, making them a popular dark horse pick to win the national championship in the summer of 2012. Connelly wasn't exactly buying it in the preseason, and, well, yeah.
Bill hasn’t written his Oklahoma preview for 2014 yet, but I suspect the West Virginia Effect angle will be prominent. Despite going 11-2 on the season, the Sooners finished 20th overall in F/+ in 2013. Connelly’s preliminary S&P+ projections put OU 17th in 2014. Fremeau hasn’t published his 2014 numbers, but his model says the Sooners most likely finish 10-2.
In light of the number crunchers’ ambivalence, is OU destined to go the way of the ‘Eers? If it’s optimism you seek, try this:
2013 Oklahoma offense, opponent differential on yards per play
*Data courtesy of cfbstats.com
The number in the left column represents the average yards per play that opponents allowed last season excluding their games with Oklahoma. The middle number is OU’s yards per offensive play against that team. The column to the right of that signifies the difference between the two. Lastly, the column on the far right is the difference as a percentage of the average yards per play allowed against all teams besides OU.
As an example, OU averaged 5.2 yards per offensive play against Louisiana-Monroe. ULM gave up 5.9 yards per play against all other opponents on its schedule, so the Sooners generated 0.7 fewer yards per play versus the Warhawks than their average opponent. The difference represented a shortfall of roughly 12 percent relative to the rest of ULM’s opponents.
Using yards per play as a very crude measure of offensive efficiency, the difference gives an indication of how well OU’s offense performed relative to the other opponents on a team’s schedule. Better offenses would be expected to roll up more yards than average per play on an opponent. In 2012, for instance, OU finished with a positive differential in 11 of 13 games. The Sooners ranked 6th overall in Offensive F/+ that year.
Looking at the table above, Blake Bell got the majority of the snaps, if not all, at quarterback in the stretch from Tulsa to Baylor. OU was all over the map with Bell under center. Things clicked on occasion. At other times under Bell, the Sooner O badly missed the mark. That explains OU’s fall to 21st in F/+ last year.
The Sooners might have finished even lower (and put more Ls on the ledger) if not for a strong close in the season’s final four games. That stretch included four straight wins to close the campaign. It also coincided with the return of Trevor Knight at QB.
To be fair, OU was far from a juggernaut in Knight's first two starts. Still, you could make a strong argument that OU just put a different (read: better) team on the field with Knight running the show. (Note that in the one contest in that four-game stretch in which OU didn't outperform, Knight only played the first half.) Phil Steele's game grades for Oklahoma in '13 also support that assessment.
Having seen so many bowl-game revelations flame out the next season, I can't blame anyone who's hesitating to embrace OU as one of the best teams in the country. I'm a believer in Knight, but even I don't think these Sooners can win a national championship in the new postseason format.
However, I also don't think you can ignore the evidence that Knight's presence in the lineup took OU to another level last season, even before the Sugar Bowl. It may be a cliche, but in that sense, the numbers aren't telling the whole story when it comes to projecting the Sooners in 2014.