关键词: 技术分析 , 收益率非线性 , 数据窥查 , 事后选择偏差

Abstract: Whether technical analysis can bring investors excess returns is one of the financial problems widely concerned by theorists. Early studies mainly adopted traditional ttest method，leading to the conclusion that technical analysis was invalid；while along with the widespread computer application，many scholars，such as Brock used the bootstrap method based on the measurement error of ttest，drew opposite conclusions. Subsequently，some researchers paid attention to the nonlinear correlation of stock yields，by using the feed forward artificial neural network model，got the idea that technical analysis was effective. However，the reduced extra yields by elimination of data snooping and application of genetic programming model made the effective conclusion be questioned largely. So far，there is no consistent conclusion in terms of this problem that if technical analysis is valid，leaving a large researching space.

Key words: technical analysis , non-linear correlation of stock yields , data snooping , ex-post selection