Research On Forecasting Method Of Corn Futures Price In China Based On Machine Learning

The futures market is an important part of the market economy.Among them,as one of the organization structure of futures market,agricultural product futures play a very important role in the whole market.Corn,as the first batch of futures products listed in China's agricultural product futures market,has an important influence on its downstream products and other futures markets.Therefore,it is of great significance for corn growers,investors and managers to analyze the factors affecting the price of corn futures and to predict their prices.Because of the leverage of futures investment,investors are faced with greater risks when futures investment is made.Therefore,it is the key to make accurate determination of future trend of corn futures price,which is the key to make corn futures investment With the development of computer technology and statistical learning,the methods of stock market analysis have been greatly enriched,in which machine learning has a good approximation ability to nonlinear systems,and its application in the field of securities analysis has been paid more and more attention.In order to consider the problem of futures market in an all-round way,and based on the analysis of the fundamental factors that affect the price of corn futures,the paper combines the technical index and the machine learning method effectively to model and apply the model and the space of the fluctuation of the price of corn futures.In the study of price trend prediction,210 data fragments with upward trend were intercepted in the closing trend of corn futures,and 168 groups of training data were collected from trend fragments by 50% cross-validation method,another 42 groups were tested from January 2006 to December 2015.Then the original price data are processed by the method of data feature extraction,and then reduced dimension is processed by independent component analysis(ICA)and principal component analysis(PCA),respectively.Finally,support vector machine(SVM)in machine learning method is used to predict the trend of futures price.The results show that it is feasible to use SVM classification method to predict the futures price fluctuation after extracting the feature of futures price data and reducing the dimension.The prediction effect of SVM on dimensionality reduction is significantly higher than that of only feature extraction.In addition,the prediction effect of ICA dimensionality reduction in this study is better than that of PCA,and the prediction accuracy of the trend fragment is improved by 11%.Finally,ICA and SVM models are used to forecast the closing price of 240 trading days corn futures c05 contract in 2017.The accuracy of forecast is 65.83,which indicates that the method has some reference value.In the prediction of price volatility space,fuzzy information granulation(FIG)and least square support vector machine(LSSVM)are combined to predict the price volatility space of corn futures.In this study,the corn futures opening price data from January 2006 to December 2015 were used as samples for simulation training,and the parameters of SVM were optimized by cross-validation,and 5 days were used as an information granulation window.The regression prediction of the three parameters of triangular-type fuzzy particle Low-Rover up is carried out.Finally,the change range of the opening price of corn futures in the next 5 trading days(January 4,2016 to 8)is in line with the actual results.The prediction results based on FIG and LSSVM are compared with those of traditional and other two SVM models,and the mean square error is the smallest.It is known that the prediction results based on FIG and LSSVM are the closest to the real values.The results show that the forecasting method based on FIG and LSSVM is feasible and effective in corn futures price forecasting.Finally,the model is used to predict the fluctuation space of the opening price of 48 grain space corn futures c05 contracts in 2017,it shows that this research method has certain reference value,and it has practical guiding significance for futures investors to study the price fluctuation.