Do Industry Portfolios Lead Market Portfolio

I investigated whether the industry portfolio returns could be utilized to forecast the overall market portfolio in China. I find that huge number of industries displays predictability to some extent, but the predictability is very sensitive to the time horizon containing the information, the time lag and the time horizon to predict. These finding suggests that industry portfolio information needs to accumulate to certain level before market reaction and that market requires time to diffuse/ process the information. Both these requirements are in line with the time horizon the information contained and to predict. Overall speaking, the results of conditional predictability from past industry portfolios are against the weak form market efficiency.