Abstract: | ![]() McInish and Puglisi reccntly claimed that weak form efficiency should be tested by runs tests and not autocorrelation tests, when returns are not Normal. It is the opinion of this author however, that autocorrelation tests are not invalidated by a non-Normal distribution. Furthermore, it can be shown that the runs test has lower power than autocorrelation tests and ihat the runs test is not preferable. |