Scholarlink

This paper introduces a new open-source ranking model for college football. Historically, teams have been ranked by voters and this has led to numerous controversies and allegations of human bias. Additionally, many polls were designed last century and do not consider the new nuances of the College Football Playoff (CFP) era. Our model is trained only on playoff seasons and is purely driven by quantitative metrics. We explain our research framework and model architecture then share the results. Before the playoffs began our model successfully predicted the champion Indiana Hoosiers and had the runner-up Miami Hurricanes rated higher than the CFP Selection Committee.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Design, Deployment, and Evaluation of an Open-Source College Football Championship Prediction Model

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Charles Jeffrey Danoff,
  • Cecil Isaih Battiste

摘要

This paper introduces a new open-source ranking model for college football. Historically, teams have been ranked by voters and this has led to numerous controversies and allegations of human bias. Additionally, many polls were designed last century and do not consider the new nuances of the College Football Playoff (CFP) era. Our model is trained only on playoff seasons and is purely driven by quantitative metrics. We explain our research framework and model architecture then share the results. Before the playoffs began our model successfully predicted the champion Indiana Hoosiers and had the runner-up Miami Hurricanes rated higher than the CFP Selection Committee.