Analysis And Forecast Of Volatility Of China Containerized Freight Index Based On EMD

In October 2013,when General Secretary Xi Jinping visited the ASEAN countries,he proposed to build the "21st Century Maritime Silk Road." With the advancement of the "21st Century Maritime Silk Road" project,the study of waterway transportation,especially the freight rate of container transportation,is even more important.In order to grasp the fluctuation trend of container transportation freight rate,an effective method is to use the a ppropriate model to analyze and predict the volatility of the existing authoritative container freight index.China Containerized Freight Index(CCFI)refers to a shipping price index that reflects the changing trend of China's export container shipping ma rket price.It was released by the Shanghai Shipping Exchange in April 1998.A number of shipping freight rates have been released globally,the most authoritative of which is the world's three major shipping freight index,including CCFI,Baltic Dry Index(BDI)and Baltimore Tanker Freight Index.CCFI objectively reflects the changes in the freight rate of China's container shipping market.It helps the government to make macro-control and policy making,as well as business decisions.Empirical Mode Decomposition(EMD)is a new time-frequency analysis method proposed by Norden E.Huang et al.(1998),especially for nonlinear and non-stationary data processing.EMD is a direct,local,adaptive,a posteriori data processing method,which can be adaptively decomposed into a series of limited ones according to the local feature scale of the original data without prior setting.An Intrinsic Mode Function(IMF)with a different time scale feature and a residual term,this performance preserves the information contained in the data to the greatest extent,reducing the influence of human operations on the content of the data information,which is beneficial to Exploit the economic,physical and other practical meanings of the data.CCFI is a complex time series that is fluctuated by many internal and external factors.The single statistical method used in the past to predict or analyze it is not effective.Based on the empirical modal decomposition algorithm and based on the idea of decomposition and reconstruction,t his paper proposes a CCFI volatility feature analysis model,which measures the CCFI volatility correlation and predicts the volatility trend.The volatility analysis model can be divided into two major steps: sequence decomposition and feature analysis.Firstly,the CCFI and BDI sequences are adaptively decomposed into a finite number of IMFs and one residual term based on the EMD method,in order to further highlight the characteristics of the freight rate fluctuation period and enhance the conclusion.The practical application effect is to reconstruct the IMF component and the residual term of CCFI and BDI into the high frequency part,the intermediate frequency part,the low frequency part and the trend item according to the average period and the variance ratio,and then calculate the correlation coefficient and feature analysis.When predicting the fluctuation trend,the IMF component and the residual term obtained by the decomposition are modeled and predicted by BP neural network.Finally,the predict ed values of each part are synthesized and reconstructed to obtain the CCFI sequence prediction value,and the ARIMA model,BP model and EMD-The ARIMA model performs a comparison of prediction accuracy.Based on the analysis method of wave characteristics based on EMD model,this paper analyzes the monthly data of CCFI and BDI from March 2003 to June 2018,and explores the linkage relationship between container transportation market and dry bulk transportation market.The results show that: There is a moderate positive correlation between the CCFI and BDI time series as a whole;the reconstructed components based on EMD decomposition show that the overall trend of CCFI and BDI is highly positively correlated,consistent with the world economic trend,showing a downward trend;the average period is 92 months.The low frequency part is strongly negatively correlated,reflecting cyclical fluctuations,which differ by 1/2 cycle;the intermediate frequency is moderately positively correlated,with an average peri od of 14.431 4 months,about 1.20 years,reflecting seasonal fluctuations,both CCFI and BDI are present.Seasonal fluctuations,but the fluctuation pattern is different;the average period is 4.134 August,the high frequency part has no correlation,and its fluctuation is small and the frequency is fast,which is an important factor affecting the synchronization of the two;using the modal characteristics and fluctuation characteristics of each scale The classification management and analysis of the fluctuation correlation caused by different influencing factors can be well realized.For the CCFI week data from January 7,2000 to June 22,2018,four prediction models were used for single-step prediction.By calculating the five error evaluation indicators,the prediction accuracy of CCFI prediction using EMD-BP model is far.Higher than EMD-ARIMA model,ARIMA model and BP model,the rationality and superiority of the EMD-BP model established in this paper are verified.At the same time,for nonlinear unstead y time series such as CCFI,the prediction accuracy and effect of BP model are better than ARIMA model when fitting and predicting it.In addition,whether it is a BP model or an ARIMA model,the combination of EMD can improve the performance of the fitting prediction.Introducing the EMD method into the study of the fluctuation characteristics of CCFI provides a good idea and method for further study of the fluctuation law of CCFI.The experimental results show that based on the EMD method,the variation law and correlation of the freight index at various fluctuating frequencies can be obtained.The combined model of EMD and BP neural network significantly improves the prediction quality and confirms the new method in predicting the freight index.Effectiveness can help CCFI volatility trend forecasting.