Research On Petrochemicals Price Fluctuation And Forecasting

In 2014,the international crude oil prices decline sharply.And for the past few years,there is little chances to return to high level as before.However,the prices of petrochemical products have fluctuated more frequently and sharply than ever due to various arbitrage behaviors and speculation of floating capital.This seriously affects the healthy development of the petrochemical industry.Therefore,how to obtain corresponding profits while avoiding risks has become one of the most concerned issues for petrochemical enterprises.As a representative product in the petrochemical products,the research on the price of benzene is great significance for exploring the regularity characteristics of price fluctuations and price prediction methods after the comprehensive marketization of petrochemical products in ChinaIn recent years,with the rise of big data,various emerging big data algorithms have emerged that provide more possibility to achieve the previous expectations.In this paper,by econometric model and machine learning methods,the characteristics of benzene price fluctuations are analyzed and summarized,and the long-term trend prediction and short-term price prediction methods of benzene prices are studied.The specific research content and results are as follows:(1)Research on the price fluctuation law of benzene.First of all,in a qualitative point of view,analyzes its price fluctuation law from the price volatility series chart of benzene.And then,the statistical magnitude characteristics and distribution histogram of the benzene price time series are obtained by statistical software.After that,the fluctuation law of benzene is quantitatively analyzed.At last,establish a model to fit the volatility of the benzene price in order to get the regular fluctuation characteristics,such as random,volatile clustering,asymmetry and long-term memory.(2)Study on the long-term trend prediction of the benzene price.First of all,from the perspective of theoretical and practical work experience,selecting Brent crude oil spot FOB price,benzene’s FOB South Korea price,benzene’s FOB U.S.Gulf price,benzene’s FOB Rotterdam price,and styrene’s FOB South Korea price as topics.Then,from the correlation among the variables,the long-term cointegration relationship was verified staying among the variables by the E-G two-step method that means there is price guidance relationship among them.Finally,through the Granger causality test to prove the result that predicting the FOB Korea price changed trend of benzene in 1 month by the current changed trend of the Brent crude oil price and the styrene’s FOB Korea price.(3)Research on short-term prediction of benzene prices.Firstly,the price of benzene is predicted by the ARIMA model.And then,the ARIMA-SVM combination model is established to predict the short-term price.At last,compare the prediction value of the ARIMA model with the forecast result from the ARIMA-SVM combined model,and then,the result is that the effectiveness of the combined model is better than the single ARIMA model on the short-term prediction price of benzene.