Friday, August 23, 2013

No NCAA FBS Predictions Here

Freakonomics has a nice podcast about the folly of predictions.  Given that lately I have spent some time doing research on NCAA FBS teams, I get the occasional question about how "team X" will do in the upcoming season.  My answers is:  "I don't know"; which is about as honest an answer as I can give.  In other words, I don't make predictions (except for bowl games) about how NCAA FBS teams will fare because the model that I use is not designed to make predictions, and according to Freakonomics - it is folly anyway.  (Glad they agree with me).

While this may not be very satisfying let me give two examples using two sets of teams (Mississippi State and Mississippi) and (Oregon and Oregon State) where both sets of teams are from the same state.  I mention this as the teams have a similar population to draw from for recruiting and each set of teams plays in the same FBS conference and none of these teams changed coaches during the time period I am looking at.  I state this because I starting reading a book on college football and most of the examples used to make a point are not what I would call a typical case.  I will write more on this later.

First let's look at Mississippi State and Ole Miss (i.e. Mississippi).  Using my NCAA FBS Production Model, Mississippi State finished as the #43 highest ranked team in FBS for the 2011 season and followed that up with a #44 ranking for the 2012 season.  Both were above average teams and notice that there was not much movement from one season to another.  On the other hand, Ole Miss finished the 2011 season as the #119 most productive team (or the #2 least productive team) and then dramatically increased to #45 in 2012, just one place below Mississippi State.

Second let's look at Oregon and Oregon State.  Again using the NCAA FBS Production Model, Oregon was the #8 most productive team in 2011 and the #3 most productive team in 2012, while Oregon State went from the #96 ranked team in 2011 to the #32 ranked team in 2012.

So what makes one team remain similar and another team change dramatically?  Using the NCAA FBS Production model - I don't know.  In other words the model is not designed to make predictions, but rather to take on-field actions by the teams offense and defense and turn them into measures of overall production.  Once that I done, I sort from highest to lowest (for the team production) and the team that has the highest overall production number is the #1 ranked team and the team with the lowest production number is the #124 ranked team in 2012.  I can tell you that Oregon and Oregon State both improved on both sides of the ball, which helped increase their productivity relative to other teams in the FBS "league".  But I can't tell you before the season starts or for the next game what will happen.  I do make "predictions" for the bowl games and am about 75% 60% correct on those.  Hey, upset's do happen - Jack!  (Channeling my inner Si Roberson from Duck Dynasty).

UPDATE:  As mentioned below - I was 60% correct on the bowl games.  Thanks for your close reading of the blog!