A Study Of Corn Futures Prices Prediction Based On Chaotic Time Series

Corn has gain property, fodder crops property and energy crops attribute, it is becoming more and more in the agricultural production and the development of the national economy. Corn futures market is a product of the spot market which developing to a certain stage, it has the function of discovering price and avoiding risk, but also brings some potential risks to the investors. Price risk is core factors in corn futures market risk. The complexity of corn futures prices internal structure and the variability of external factors make traditional prediction methods not to meet the needs of corn futures price prediction. So, this paper which considering the nonlinear characteristics of corn futures market and combining chaos theory and support vector machine theory analysis and forecast corn futures prices time series through noise smoothing, phase space reconstruction and method predicting.1) This paper summarizes the research status of domestic and foreign futures price prediction, evaluates a variety of prediction methods and introduces the basic theory of chaotic time series prediction, from phase space reconstruction, chaotic nature identification and chaotic time series prediction method.2) This paper researches data quality issues from noise smoothing, linear trend elimination and standardization, focused on noise smoothing. Through comparative analysis, this paper uses the nonlinear threshold method in the wavelet analysis to an empirical test on corn futures price time series, The result is satisfying; According to the practical problem of this paper, it which eliminating monetary total linear trend makes follow-up study more realistic. At last, standardization is used to avoid the impact of data volatility.3) Considering the nonlinear characteristics of corn futures market, this paper puts aside the idea of the traditional univariate time series phase space reconstruction, and uses multivariate time series phase space reconstruction theory to process the data. First of all, use mutual information method and minimum error method to calculate the delay time and embedding dimension. Secondly, reconstructing phase space. Finally, use the improved small data volume method to calculate the largest Lyapunov exponent. The result is that the largest Lyapunov exponent of corn futures market is greater than zero. It means corn futures market has obvious characteristics and short-term predictability.4) This paper introduces Support Vector Machine regression principle and Least Squares Support Vector Machine regression algorithm. It proposes a prediction method of multivariate time series based on chaos theory and least squared support vector machine theory. The experiment on the prediction of corn futures opening prices time series is carried out. The results indicate that least squared support vector machine prediction model of multivariate time series outperforms least squared support vector machine prediction model of univariate time series. The results have very important theory and practice sense for nonlinear modeling and forecasting of futures prices.