Power grid material purchase price prediction method associated with futures contract

The invention provides a power grid material purchase price prediction method associated with a futures contract. The method comprises the following steps: carrying out trend fitting on a power grid material price; determining an associated futures variety; constructing a quantile regression model by using a long short-term memory (LSTM) network, and training the model to obtain a future contract price estimated value; predicting a power grid material price according to the futures contract price estimation value; according to the method, the influence mechanism of financial products on material prices is optimized from logic conduction to learning conduction, the linkage relation between related materials and various parameters is automatically identified through an intelligent algorithm, and the workload of defining the physical composition of each material is reduced; by means of strong feature extraction and learning capabilities of the LSTM model, interval prediction with reliable sharp fluctuation secti