Enhancing NFL Betting via Residual Neural Network Calibration

This presentation explores the calibration of NFL moneyline probabilities using Residual Neural Networks. It aims to address the inefficiencies observed in extreme cases within NFL betting markets and evaluates whether actionable signals can be derived. Using data spanning 2010–2022, the project incorporates feature engineering techniques, including implied probabilities and contextual factors such as weather and injuries, to improve prediction accuracy. The initiative combines AI-driven...