传统评价基金选股能力方法通常假定残差服从正态分布,但现实数据往往不满足此假设条件,非参数Block-Bootstrap仿真技术可以对此修正。本文以2004年1月至2008年4月1我国153支开放式偏股型基金月度数据为样本,对基金选股能力进行研究。结果表明利用Block-Bootstrap方法剔除掉“伪”选股能力基金后,确实存在具备显著选股能力的基金。
Traditional methods assume that residuals obey normal distribution, which is, however, not the ease in reality. Fortunately, the Block-Bootstrap simulation method can amend it. This paper uses the sample of 153 mutual funds from Jan. 2004 to Apr. 2008 to examine stock selecting ability. The results manifest that there exists funds with strong stock selecting ability after eliminating the fake ones by means of Block-Bootstrap.