Hooray. Bucks 42, Ducks 20
The College Football Playoff Selection Committee took a lot of heat when they seeded OSU in the tourney.
Well, last nite’s game vindicated the committee.
Also vindicated was Rev. Thomas Bayes.
You know, the guy who developed Bayes’ Theorem.
Here’s what I mean …
In my Strategic Business Analytics course we cover Bayesian Inference … applications of Bayes’ Theorem … how to statistically adjust your so-called “prior beliefs” when you get new evidence.
Even I admit that sometimes the subject can get a bit dry …
So let’s bring it to life…
Prior to yesterday’s Ohio State – Oregon game, Nate Silver’s fivethirtyeight.com put the game into a Bayesian context, centered around Cardale Jones – OSU’s 3rd-string wunderQB.The fivethirtyeight: analysis ….
Jones, a sophomore, was only regarded as the 41st-best QB prospect in his freshman class.
He began the 2014 season third on Ohio State’s QB depth chart, behind Heisman candidate Braxton Miller and touted redshirt freshman backup J.T. Barrett.
Then Miller was injured and lost for the season in an August practice, upon which Barrett took the reins and led the Buckeyes to a 11-1 record (on the strength of one of the nation’s best passing performances) before going down with a season-ending injury in late November.
Jones took the helm … having thrown just 19 career passes before starting in the Big Ten title game against Wisconsin.
But Jones and the Buckeyes were spectacular in that game, winning 59-0, and then upset favored Alabama in the College Football Playoff semifinals.
It effectively created an interesting Bayesian problem:
How do we balance a sample of two great games against the prior assumption that Jones is a lightly regarded third-stringer?
Thanks to Bayes Theorem – adjust your “priors” when you get new evidence — the betting line has narrowed to Oregon winning by a touchdown.
The Bayesian assessment of Jones — adjusting “priors” based on new evidence –turned out to be on the money last nite..
Way to go CFP, Ohio State, Cardale … and, oh yeah, Rev. Bayes.