作者: BARR ROSENBERG , ANDREW RUDD
DOI: 10.1111/J.1540-6261.1982.TB03575.X
关键词:
摘要: One of the most basic tests market efficiency is test for serial correlation returns. If return on a typical stock in period t correlated with its 1, then best (unbiased minimum mean square error) prediction equals prior multiplied by coefficient. An investor concerned only about and variance portfolio would use to improve mean/variance trade-off shading toward stocks predicted have high returns away from low Hence, using knowledge previous month's returns, produces superior (mean/variance) performance. This violates weak form market-efficiency (Fama [1]), sense that expected each stock, based upon return, different current price. The same argument may be applied any component as distinct total return. For example, linear multiple-factor-plus-specific-return model security excess decomposed into factor-related specific empirical results this paper find positive dependence negative which nearly offset one another, resulting zero Formally,