In this paper:Black F, Jensen M C, Scholes M. The capital asset pricing model: Some empirical tests[J]. Studies in the theory of capital markets, 1972, 81(3): 79-121. It states that
One procedure for solving this problem which makes appropriate allowance for the effects of the non-independence of the residuals on the standard error of estimate of the average coefficient, alpha, is to run the tests on grouped data.
Why would a grouped alpha solve that probelm? I can't understand it.