Prediction Of Shanghai-Shenzhen 300 Index Based On EEMD_LSTM Model

As an important channel for real enterprises to raise funds,China's stock market is an important channel for investors to allocate their own assets,which is an indispensable and important part of China's financial market.Prediction of stock index has always been of great research significance,and there are many classical theories.However,the stock data is highly noisy,dynamic,non-linear,non-parametric and chaotic in nature,which is contrary to some basic assumptions of classical theory and restricts the scope of application of classical theory.Therefore,it is of great practical and theoretical value to find a reasonable method for feature extraction of stock market data and to construct a non-linear dynamic system model that can describe the complexity of the stock market.It will further reveal the inherent operation law of the stock market,better play the proper functions of the stock market,and more timely expose financial risks.In this paper,the closing price of Shanghai and Shenzhen 300 index is decomposed by using Ensemble Empirical Mode Decomposition(EEMD)algorithm.Ten intrinsic modulus functions and one residual function are obtained successfully,and the time-frequency characteristics of the intrinsic modulus function are analyzed..Using standard R/S analysis,it is found that the longest memory period in logarithmic return of Shanghai and Shenzhen 300 is approximately consistent with the longest period of intrinsic mode functions,which reveals the meanings of the period of intrinsic modulus function.Finally,a prediction model of Shanghai and Shenzhen 300 index based on EEMD?LSTM model is established,and a variety of evaluation indicators are established to measure the performance of the model,and a reference model is established to compare the prediction effect of the model.The final results show that the EEMD-LSTM model has better prediction performance.EEMD method can greatly improve the prediction performance of the neural network model.After changing the time range and frequency of data sets,EEMD-LSTM model also has better prediction results than other models.Experiments also show that the choice of data sets has less impact on prediction results than EEMD decomposition.