Reply to thread

Message: [QUOTE="Cusefan95, post: 5460366, member: 173"] I think at times we get so enamored with data analysis that we don’t assess the quality of the data. With all the turnover in college football, the data a guy like Connelly puts into his model is mostly low quality guesses. It sorta kinda works out because teams usually stay at the level they are at - so if he uses multiple years of data it’s going to end up close on most teams. Then he gets to tout how accurate he was at the end of the year. But we really don’t need a model to tell us Texas and Georgia are going to be good, and Maryland and Purdue will be terrible this year. What’s far more interesting is a model which can identify teams that will defy expectations (either good or bad) - I don’t think that can come from pure data for college football. [/QUOTE]