Stock Prediction And Optimization Based On Wavelet Neural Network And Support Vector Machine

Stock market data is usually highly volatile,and forecasting stock market data has always been an important issue in the financial sector.Stock forecasting predicts the running trend of the stock price index,which is also a research hotspot in the field of statistical finance at home and abroad.The traditional stock forecasting method is mainly linear forecasting method,and the more common one is to establish an autoregressive moving average model.In the field of financial research,the autoregressive moving model(ARIMA)is a major predictive method.It is a linear forecasting method that predicts some stationary data better,but often has strong volatility in stock data.The effect is not good.Because the ARIMA model's nonlinear pre-capacity is poor,it can't produce satisfactory prediction results.Researchers need to find more methods suitable for stock forecasting.Machine learning has a unique advantage in the processing of financial industry data.It can accurately analyze the changes of a large number of stock data or financial data at the same time,and quickly draw corresponding conclusions,so that the efficiency of the financial market is significantly improved.In the trend forecast of the stock market,it can use the relevant characteristics of the stock price index to predict the stock market data.In the aspect of financial data management,the machine learning algorithm can effectively analyze the financial data of the company's balance sheet and cash flow statement.We need to find some prediction methods that can adapt to nonlinear data.Therefore,we intend to use machine learning related algorithms.This paper uses wavelet neural network(WNN)and support vector machine(SVM).Wavelet neural network is a theoretical method that combines wavelet with neural network.This method combines the advantages of both theories and is an effective stock forecasting method.The support vector machine is supported by solid mathematical theory.The kernel function method can effectively solve some complex calculations,and the application of structural risk minimization principle makes this method widely respected in the field of financial forecasting.In order to improve the performance of the model and the prediction accuracy,particle swarm optimization(PSO)is used to optimize WNN and WNN parameters to establish the stock prediction model PSO-WNN.In order to follow the logic of the experiment,the stock forecasting model PSO-SVM is also established.The simulation experiment is carried out by using MATLAB,and the feasibility of wavelet neural network and support vector machine in stock forecasting is proved by analyzing the experimental results.Set relevant statistical indicators to measure the prediction effect,evaluate the optimization effect,and analyze the prediction performance of the comparison model after optimization,and finally comprehensively analyze the overall prediction performance.