Stock Price Forecasting Based on Improved Support Vector Regression

The prediction of stock price is of great significance to social and economic development, market supervision, enterprise management and investors’ investment behavior. Through obtaining and analyzing the stock historical data of many domestic and foreign companies, this paper finds that the prediction deviation of Support Vector Regression (SVR) for individual stocks is very large. Therefore, in order to solve the problem that the deviation on specific stocks is too large, an improved SVR method based on dividing data segments is proposed. The experimental results on Amazon (AMZN) stock data show that our improved algorithm outperforms the original SVR in R2 and RMSE, which proves its effectiveness.