This study aims to forecast the wholesale price of green grams in the Chinese market using a nonlinear auto-regressive neural network. The dataset covers a ten-year period from January 1, 2010, to January 3, 2020, and is significant for economic analysis. The neural network model, which combines various basic nonlinear functions, is evaluated for its prediction performance across different configurations of data splitting ratios, hidden neuron counts, and delay counts. The model demonstrates good stability and accuracy, with relative root mean square errors (RRMSE) of 4.34%, 4.71%, and 3.98% for training, validation, and testing, respectively. Benchmarking against other machine learning models and classic time-series econometric methods shows that the neural network model outperforms them statistically. The findings suggest that neural networks are a valuable tool for predicting green gram prices, which can be used for independent technical forecasts or combined with other forecasts for policy research and trend analysis. The study highlights the potential of machine learning models in agricultural commodity price forecasting, particularly in capturing nonlinear patterns in time series data.This study aims to forecast the wholesale price of green grams in the Chinese market using a nonlinear auto-regressive neural network. The dataset covers a ten-year period from January 1, 2010, to January 3, 2020, and is significant for economic analysis. The neural network model, which combines various basic nonlinear functions, is evaluated for its prediction performance across different configurations of data splitting ratios, hidden neuron counts, and delay counts. The model demonstrates good stability and accuracy, with relative root mean square errors (RRMSE) of 4.34%, 4.71%, and 3.98% for training, validation, and testing, respectively. Benchmarking against other machine learning models and classic time-series econometric methods shows that the neural network model outperforms them statistically. The findings suggest that neural networks are a valuable tool for predicting green gram prices, which can be used for independent technical forecasts or combined with other forecasts for policy research and trend analysis. The study highlights the potential of machine learning models in agricultural commodity price forecasting, particularly in capturing nonlinear patterns in time series data.