The Research On Construct Forecasting Model And Forecast For SHFE Copper Price

With the improvement of China's market economy status and people's financial awareness,futures trading has become an important trading product in financial transactions and financial derivatives trading.The healthy and stable development of futures market has also become the focus of managers and investors.Whether the futures market is to used to invest or speculation,good risk control is particularly important.The premise of the risk control is to predict the futures price,and to establish the corresponding transaction principle after the forecast analysis,and to manage the risk according to the principle of transaction.At present,with the continuous progress and development of modern technology,the forecasting method is becoming more and more,taking the method of statistics as an example,there are time series forecasting model,gray prediction model,neural network prediction model and so on.In this paper,we select three kinds of single forecasting model of the ARIMA model,GARCH model and BP neural network model for prediction of SHFE copper price.Select the Shanghai copper master contract from January 5,2015 to September 25,2015 a total of 180 trading days of the closing price data as the object of study,which January5,2015 to August 28,2015 Shanghai copper master contract price data are used to fit the estimated model and the remaining data are used for forecasting.The empirical results show that the cumulative error value and MAPE value of the BP neural network model are smaller than those of the ARIMA model and the GARCH model,which shows that the BP neural network model has the highest prediction accuracy in these three models.This is mainly due to the BP neural network model has a strong self-learning ability,it can be learned through training to find the rules and characteristics between the parameters to grasp the interdependence between data.Secondly,based on the prediction effect of the previous three single prediction models,two combined forecasting models,namely the optimal weight linear combination forecasting model and combination forecasting model based on BP neural network,are optimized.These two combination forecasting models are also used to do the same empirical analysis.The conclusion shows that these two kinds of combined forecasting models have improved the prediction accuracy to a certain extent compared with the single prediction model.In these two combined forecasting models,the combined forecasting model based on BP neural network has higher prediction accuracy and has wider applicability.Finally,taking into account the different models have their corresponding applicability for different data,this paper also selected the October 10,2016 to November 18,2016,a total of 30 trading days of the Shanghai copper main closing price for short-term time span forecast study,compared with 180 trading days of the long-term time span of the forecast results,toexplore the applicable conditions and scope of these models in the SHFE copper futures price forecast.The result shows that the combined forecasting model based on BP neural network has a good prediction effect both in short time span and long time span forecast.