Research On Stock Price Forecasting And Investment Strategy Based On Support Vector Machine

As a common investment method,stock investment also changes with each passing day.An increasing number of investment analysts are beginning to combine the securities investment approach with computer technology and take the advantages of analyzing data by computer to conduct stock trading.SVM(Support Vector Machine)as a kind of data mining technology,to solve problems such as nonlinear,high dimension and over fitting,shows the unique advantage in stock price prediction.At the same time,it has strong theoretical and practical significance to the government,enterprises,equity investment institutions and individuals for providing guidance when making decision or investment.Forecasting the stock price in our country based on support vector machine,this paper summarizes the domestic and foreign literature on stock investment methods and expounds the basic principle of support vector machine,and then compared with the securities investment analysis,mathematical statistical model and neural network,found that the SVM can solve the problems of nonlinear,high dimension and local minimum fitting.Secondly,a support vector machine prediction model is set up by selecting the best parameters of four different kernel functions.The way to build a model for forecasting stock price is choosing a kernel function and the parameters by minimizing error.The predicted results show that the fitting effect of training sample is very good,and for predicting samples,due to high stock price change frequency,prediction effect is a bit weak,but the Gaussian kernel function performances best.Next,using this model to predict the stock price and using the price information to build the mixed investment strategy(allocate funds and "vote" with signals from different strategies).The results of mixed investment strategy based on the predicted is better than simple strategy,and it provides a good way of stock investment.Finally,the paper summarizes the model and strategy of support vector machine,and points out the problems in the model and strategy and the direction of future improvement and research.