【摘要】 预测股指时间序列突变点是在股票市场上进行投资的一个关键问题,而检测突变点是预测的基础.在检测深沪两市股指时间序列月度收益率突变点位置和个数时采用了非参数方法,该方法基于小波数据依赖门限技术.研究显示了运用Lipschitz指数解释的突变点的数学特征.使用的模型证明了小波变换模的极大值能够检测出突变点的位置,实证结果也显示出突变点的位置和个数是精确的.更多还原

【Abstract】 Forecasting the change point of stock exchanges index time series is a key problem to the investment on stock market and then the detection of change point is a basis to the forecast.A non-parametric method,based on a wavelet data-dependent threshold technique for change point detection,is applied to detect the location and the number of change point of the monthly return rate of Shanghai stock exchanges index time series.It shows the explanation of the mathematical characterization of change point with Lipschitz exponents.The model mentioned in this paper proves that the local maxima of the wavelet transform modulus detect the locations of change points.The real example result shows that the location and number of change point is precisely.更多还原