Characteristics Of Gold Price Volatility On Multi-scale Analysis And Prediction Based On EEMD Model

Gold has money, finance, commodities and other attributes,which is widely used in various fields. Because of its complex properties, the gold price not only have a huge impact on the economic environment, and will be effected by geopolitics, money market, stock market, oil market and other factors. Therefore, how to effectively predict the price of gold has become an important issue.Firstly, we analysis the historical changes of gold, the world’s major gold market and macroeconomic factors that affect the price of gold systematically.Then the forecasting model used in previous studies were reviewed, the advantages and disadvantages of each Evaluation prediction method. As the gold price series are nonlinear and non-stationary, the traditional economic model built on the assumption that the data is linear, it is difficult to capture the gold price in the sequence of non-linear model, and thus can not accurately predict the price of gold. Accordingly, this paper presents an analysis method based on EEMD (Ensemble Empirical Mode Decomposition) of the price of gold, using a EEMD method for the international gold price history into several different frequency components of the price, market price fluctuations were extracted significant events prices and price trends. ICSS method combines three events with Chow price point inspection for structural change and compared through the use of an external event and point test results, in order to analyze the impact of external events on the gold price volatility, market volatility term cycle which is 4.3 months can be considered the gold price in the short term there is a self-adjustment function, capable of general market events, such as "Dubai events " and " Cyprus incident" occurred adjust its duration of about three months or so. Cycle Events items was 40.8 months, can be considered some of the unanticipated and unexpected events such as war, economic crisis can have a greater impact on the price of gold, and will continue for a long period of time; term represents the long term trend of price changes trend. Finally, using support vector machine model to decompose and extract the gold price market price volatility, major events and trends in the price regression to predict prices, the use of particle swarm optimization (PSO) were four different kernel functions optimized support vector machine, with the gold Price comparison of regression prediction program, found the use of particle swarm optimization RBF kernel support vector machines for the gold market price forecasts achieved good prediction.When using move window 5-step method to get ahead of the predicted value compared with the value one step ahead to examine the predictive ability of the sample, five steps ahead of forecast accuracy is one step ahead of forecast accuracy is worse than that.