One Would Suggest

日期:2015/08/10

What do you think of Granger's article  "Forecasting Stock Market Prices: Lessons for Forecasters"  - A Reader

One would suggest that many of the studies use multiple classifications to find the best fit, none of them update the results past 1990, the earnings data they use suffers from the retrospections bias, the samples do not use as-is-files in most cases but data assuming the present members of the index are still there, and they don't take account that people are smart, and they follow the essence of the academic papers. As for low e/p being superior, it is no accident that growth has beaten value for each of the last 12 years or so. As for low volatility being good, naturally if you divide up stocks into 9 regimes, one of them would be best. Granger shot his wad with attempts to apply Fourier analysis with Morgenstern ineffectively, and now he retrospects. Sort of like the sage looking for stocks selling at below cash liquidating value. Great if you can ignore the transactions costs of 25%. However, I agree with Larry and Granger concerning the conclusion but for completely more refined and specialized analysis.