The probabilities below are based on research of 1,794 college football games between 2011 and 2013 using information that I have found particularly predictive of winning and, to a lesser extent, covering the spread (or not). In general, I believe many models attempt to take too many factors into consideration. Distinguishing between the signal and the noise is important.
While the data for the larger spreads is much clearer, there is a larger sample size for the smaller spreads which makes me more confident, in general, in those numbers (though I am positive Florida State beats Duke).
An important distinction here – I’m not predicting what will happen in these games. I have no idea if the Seminoles will turn the ball over 5 times or Jameis Winston gets hurt in the first series. What I’m saying is given the data that I have teams with similar attirbutes and facing a similar opponent as Florida State have won 96.2% of the time since 2011.
To me this is a logical way to look at things. It’s easy to say Baylor should beat Texas. But what effect will the freezing rain forecast for Waco Saturday have on these teams? No one knows. But I do know what’s happened in the past.
So think of these less as predictions and more as a look at history of similar games.