Hoho... good to see somebody is talk about Kent's article.
DHS is one of the best written behavioral paper in 1990's. The one posted above is about time-series property of returns, - long run reversal and short-term momentum; DHS have another one (JF, 2001???) discussing cross-sectional properties - B/M and Size. DHS (2005) roll out a very recent WP along their early line but challenging FF 3-factor type of regressions. I personally think, DHS (2005) and DT (2005, WP) is a big come back on Fama (1998).
Of course, hardcore rationalist would say Berk, Green and Naik or its variants explain things... but they do not: do you really think good news means higher risk?!