Research On Stock Price Prediction Based On LSSVM-Markov Combination Model

The prediction problem has always been the focus of scholars' research.With the rapid development of artificial intelligence,more and more scholars have applied machine learning to prediction research and tried to improve optimization algorithms to improve prediction accuracy.Among them,SVM(Support Vector Machine),as a representative of machine learning,can better solve problems such as nonlinearity and high dimensionality,and has great advantages in prediction.LSSVM(Least Square Support Vector Machine)is an improved algorithm based on SVM,which transforms the quadratic programming problem into a linear solution,which greatly simplifies the SVM solution process and speeds up the solution speed.Markov(Markov)model is a simple and effective model,which shows that the future state is only related to the current state,and has nothing to do with the past state.It has a good prediction ability for random dynamic nonlinear data.Both the LSSVM model and the Markov model are widely used in economic and financial forecasting,but the single model has the problem of low prediction accuracy.The combination and optimization of suitable different prediction models is an effective method to improve the prediction accuracy of the models.Therefore,in order to improve the prediction accuracy of the single model—LSSVM model and Markov model,this paper proposes the LSSVM-Markov combination model based on the prediction of the price of Shanghai Stock Index.First,the LSSVM model is optimized.Second,in order to reduce the prediction error of the model,the Markov model is used to modify the prediction error value of the LSSVM model.This paper has the following research:(1)Research on the theoretical knowledge of LSSVM and Markov model.(2)Study the LSSVM model from the aspects of kernel function selection and parameter optimization,and use grid search algorithm and particle swarm algorithm to optimize this model.(3)Construct the LSSVM-Markov combination model,conduct an empirical comparison analysis on the Shanghai Stock Exchange Index,and use this model to predict the Shanghai and Shenzhen 300 Index and the Hong Kong Hang Seng Index.The final empirical results show that the particle swarm optimization algorithm has better optimization effect than the grid search algorithm.The LSSVM model optimized by the particle swarm algorithm has better generalization ability;the prediction accuracy of the LSSVM-Markov combined model is better than that of the single optimization.Compared with the ARIMA model and the SVM-Markov combination model,the LSSVM model also has better prediction capabilities,and the combination model is equally effective and feasible for the prediction of other stock index prices.