关键词: 时间序列, 谱分布, 拟合优度, Wild Bootstrap, 市场有效

Abstract: This paper investigates the spectral testing of the martingale hypothesis for a given time series. Wild bootstrapping is used to approximate the asymptotic distribution of the goodness-of-fit statistics under the null hypothesis. Numerical experiments show that the new method can not only remedy the size-distortion problem of the original procedures but also improve their powers. The application of these methods to market efficiency testing is illustrated by two real data from the stock markets.

Key words: Time series, Spectral distribution, Goodness-of-fit, Wild bootstrap, Market efficiency