In the short amount of time that the CFB QB stat filter app has been up I’ve done a little bit of mining of the data to explore for predictive relationships. To offer some context, I should say that I think predicting QB success is an extremely difficult endeavor and I don’t know that things are going to get a lot better any time soon. The primary problems that you encounter when trying to predict QB success is that teammate effects at both the college and pro level are going to obscure the picture. Also, we have small samples to test against when it comes to QBs. Whereas NFL teams will often have 3 or 4 WRs on the field, there’s just 1 QB on the field per team at any time.