中文摘要：

细分股票市场中投资者类型，运用收益方差分解的方法（适用于连续报价市场和集合竞价市场），得到两类知情者的信息性交易概率及噪声因素所占的比例，并在此基础上进行特征分析。采用2003年10月8日至2003年11月30日的分笔高频交易数据，选取上证50中45支股票为研究样本，将交易者区分为三种类型：提前知道公告信息的第一类知情者（内幕交易者）、分析判断股票信息的第二类知情者、噪声交易者。结果表明，内幕交易者的信息性交易概率大约为3．49％，远远小于第二类知情者的24．47％。噪声比例平均为58．14％，公告前的噪声比公告后的噪声大3．29个百分点。随着市值逐渐增大，信息性交易概率逐渐下降，噪声比例逐渐上升，两者都存在明显的日内效应，而日际效应却不显著。

英文摘要：

The paper subdivides the traders and proposes a method to decompose the variance of returns into noise and information components, then obtains the different informed traders＇ probability of informed trading and noise percentage. Adopting the high-frequency trading data from 2003. 10. 8 to 2003. 11. 30 and selecting 45 stocks in Shanghai Stock Exchange 50 indices, the paper divides the traders into three types： the traders who acquire t information in advance; the traders who analyze information privately and the noise traders. The empirical result show. the first type informed traders＇ probability is about 3. 49% which is less than the second type informed traders＇. The average noise percentage is 58. 14% and the noise before announcement is 3. 29% larger than after that. With the increase of market value, the probability of informed trading declines, but the noise percentage becomse large. Furthermore, both the probability of informed trading and noise percentage exist intraday pattern clearly and insignificant interday pattern. Besides, the research indicates the duration is larger around the announcement.