Abstract
This paper examines the forecasting of crude oil price volatility using the Nonlinear Neural Network Autoregression (NNAR) model based on daily data from 2018 to 2026. Given the high volatility and complex dynamics of energy markets, traditional linear models often fail to capture structural shifts. The empirical results demonstrate that the estimated NNAR(31,1,16)[5] model effectively captures nonlinear patterns and delivers robust short-term point forecasts, offering valuable insights for energy market risk management.