:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/37436998/20120915_lbm_ab2_488.0.jpg)
Over the weekend, I read an interesting piece from Dan Wolken of USA TODAY Sports on how three guys have become a major source of analytics in college football. And though "analytics" might be a little strong for some of the statistics compiled by SportSource Analytics, and many of their statistics are proprietary, some of what was hinted at in the piece is interesting.
This, for example:
Prather uses an example from the 2012 season when Florida and Kansas State both finished the regular season with 11-1 records. Though the Gators averaged just 26.8 points and Kansas State averaged 40.7, the teams' offensive output was "almost identical" when adjusting for how many points their respective opponents were allowing on average.
"Kansas State was scoring more points but playing teams that were giving up more points," Prather said. "Twenty-eight points isn't equal depending on what teams you play."
Florida's 2012 offense wasn't great at scoring points, ranking 76th nationally at 26.5 points per game. (Florida's average dropped slightly after the Sugar Bowl.) But it was much better in context: The Gators played five of the nation's top 20 teams in scoring defense, and put up the most points anyone scored on both South Carolina and Florida State (tied with Clemson) in 2012.
Taking the sort of no-duh theory that doesn't really have a ton of statistical backing — Florida's scoring defense, despite ranking 15th nationally in 2013, was probably a bit underrated given its competition, right? — and backing it up with sound reasoning and evidence is good work, if obvious work. That explaining is how people can take analytics from niche markets to mainstream thought, though, and it doesn't do anyone any harm to get smarter about college football.
And analysis and analytics that challenge our preconceived notions are also good things, because they can force us to consider things like "Florida's 2012 offense was actually good" in different lights than we previously did.
Here's hoping we get more of these analytics — and maybe develop a few of our own — in the future.