摘要
股票价格波动规律的研究是预测的基础 ,多重分形过程是迄今为止最为符合价格波动特性的模型。本文验证了中国股票市场的多重分形游走 ,并根据多重分形过程的局部尺度特性和多尺度相关性建立了小波和神经网络相结合的股票价格预测模型。实证研究结果表明 ,本文模型预测精度较由其他模型得到的预测精度明显提高。
Forecasting is based on the rule of price fluctuation.Multifractal process is the best model describing the price fluctuation property up to now.This paper verifies the multifractal walk of our country's stock market.And according to the local scale porperty and multiscale correlation of multifractal process,we build a hybrid forecasting model integrating the wavelet and neural network.Case study shows thaf this forecasting model is better than other forecasting models based on tradicitional price fluctuation model.
出处
《中国管理科学》
CSSCI
2002年第3期11-17,共7页
Chinese Journal of Management Science
关键词
中国
股票市场
多重分形
小波
神经网络
预测
multifractal
wavelet packet
neural network
stock market
forecast
