Regional steel price index forecasts with neural networks: evidence from east, south, north, central south, northeast, southwest, and northwest China

For policy makers and a diverse spectrum of market participants, understandings of commodity price forecasts are considered as an important matter. In this work, we focus on the metal business sector and examine forecast problems for seven daily regional steel prices in the Chinese market over an eleven-year period spanning 2010–2021. Specifically, our analysis covers steel prices of east, south, north, central south, northeast, southwest, and northwest China through the use of non-linear auto-regressive neural networks. For each price series, we test forecast accuracy based upon one hundred and twenty settings over training algorithms for model estimations, numbers of hidden neurons, numbers of delays, and ratios used for segmenting the data into training, validation, and testing phases. Our analysis leads to the construction of a rather simple model that generates high forecast accuracy and stabilities for each of the seven price series. Correspondingly, forecast accuracy measured by the relative root mean square error is lower than 0.60% for all considered series based upon the overall sample. Forecast results here could be utilized on a standalone basis or combined with forecasts from other models by different forecast users as part of evaluating price trends and performing policy analysis.