Validating a division I-A college football season simulation system

NCAA division I-A college football remains the only major collegiate sport not to have a true playoff to determine its national champion. A controversial and flawed process termed the bowl championship series (BCS) has emerged with the goal of matching two teams in a so-called ´national championship´ game. The BCS has employed a varied set of computer-based ranking models to help determine the two participants. The effectiveness of these computer models has never been empirically investigated in a controlled setting. This paper presents the development of a controlled test-bed of simulated college football seasons, a necessary preliminary step. A simulation model is developed and validated against past college football season data. Results show promise. It is hoped that this simulation process will allow future systematic study of the various mathematical models used to rank football teams.