Research On Futures Price Prediction Based On Multi-model Analysis

With the improvement of China's market economy status and the increasing demand for wealth management,more and more investors are willing to participate in the futures market,and futures trading has become an important trading variety in financial transactions and financial derivatives trading.The healthy and stable development of the futures market has also become a hot spot for fund managers and investors.Whether it is investing in the futures market or conducting speculative operations,controlling risks is particularly important.It is of great significance to the healthy development of the futures market by predicting and analyzing the corresponding trading principles and controlling the risks according to the trading principles.Sugar futures officially listed on the Zhengzhou Commodity Exchange in China in early 2006.This measure has promoted China's more favorable position in the formulation of sugar prices worldwide,and greatly enhanced the influence and competitiveness of China's white sugar in the international market.Because its price is the result of multiple objective and subjective factors,it is difficult to quantify and measure many impact factors.According to the above reasons,this study will select the sugar futures closing price to establish a time series forecasting model,and forecast the price as the focus of this paper.At present,with the in-depth study of scholars at home and abroad and the advancement of modern science and technology,there are more and more prediction methods for futures.Taking statistical methods as an example,there are time series prediction models,gray prediction models,and neural network prediction models.In this paper,three single prediction models,ARIMA model,GM(1,1)gray prediction model and BP neural network model,are used to predict the price of sugar futures.Based on the single prediction model,a linear combination prediction model and a nonlinear combinatorial prediction model based on ARIMA-BP neural network are proposed.Through comparative analysis,the combination prediction model can effectively reduce the error of a single forecasting model,with better prediction effect and higher prediction accuracy.In order to test the applicability of the model,this paper selects three kinds of market conditions: ascending,oscillating and descending,and predictive modeling and analysis based on 9 different situations under different time spans in the long,medium and short term,so as to test the combined forecasting model.This paper has drawn the following conclusions: the empirical results show that the BP neural network is better than the other two single prediction models in any long-term or short-term span,and the two combined prediction models have better comparison with the single prediction model.In the two combined prediction models,the nonlinear combinatorial prediction model based on ARIMA-BP neural network has higher prediction accuracy and universal applicability in most cases.This conclusion has good theoretical and practical significance for the future research of futures price forecasting.Based on the above comparison and analysis of different forecasting methods,this paper takes this as the basis to strengthen and control the continuous market risk,pay attention to nonlinear combination prediction,fully consider the data time span,strengthen futures knowledge and skills training and education,and strengthen the futures market trade monitoring,etc.This paper puts forward some policy suggestions to help investors make decision-making and develop China's sugar futures market.