基于模糊近似支持向量回归的股价预测研究

股价预测是投资策略形成和风险管理模型发展的基础。为了降低股价变化趋势中的噪声信息和投资者关于两种股价预测误差的不同偏好对股价预测的影响,提出了基于信噪比的模糊近似支持向量回归(FPSVR)的股价预测模型。首先构建信噪比输入变量,然后引入模糊隶属度和双边权重测量方法对支持向量回归(SVR)模型进行改进,最后借助沪深300成份股2008至2019年的股票时间序列日数据,按照股市的波动情况将其分为三个阶段(牛市、熊市、震荡市),并建立三个基准模型进行对比分析。研究结果表明:与三个基准模型相比,所提出的股价预测模型的预测误差最低;与原有的SVR模型相比,FPSVR模型可以更好地对处于牛市和震荡市阶段的股票时间序列进行股价预测。

Research on Stock Price Prediction Based on Fuzzy Proximal Support Vector Regression

Stock price forecasting is the basis of the formation of investment strategy and the development of risk management model.In order to reduce the influence of noise information in the trend of stock price change and investors’different preferences for two kinds of stock price prediction errors on stock price prediction,we propose a stock price prediction model based on fuzzy approximate support vector regression(FPSVR)based on signal-to-noise ratio.Firstly,the input variables of signal-to-noise ratio are constructed,and then fuzzy membership degree and bilateral weight measurement methods are introduced to support vector regression(SVR)model.Finally,with the help of the 2008-2019 stock time series day data of the 300 component stocks in Shanghai and Shenzhen,according to the volatility of the stock market,it can be divided into three stages(bull market,bear market and shock market),and three benchmark models are established for comparative analysis.The results show that compared with the three benchmark models,the prediction error of the proposed model is the lowest.Compared with the original SVR model,the FPSVR model can better predict the stock price in the bull market and volatility stage.