The Cross Section of Stock Returns, from Victor Niederhoffer
I read an interesting paper [Hou, Xue, Zhang: Digesting Anomalies] with a new model of cross sectional returns with factors based on return on equity and change in assets divided by previous assets. Apparently it uses retrospective compustat files and with all the data splits would not seem to be any better than the 20 year outdated fama french model it is supposed to replace. But I have not fleshed out all the lacunae and possible ideas of merit yet.