Factor Based Stock Screening for Up and Down Markets

Charles Kirkpatrick has been publishing his market newsletter for decades. His simple factor based stock-screener has been beating the pants off the market for a long time. Here is a paper he published in 2001 “Stock Selection – A Test of Relative Stock Values Reported Over 17-1/2 Years“. He has continued to refine and simplify his model which includes relative strength and price/sales as key factors. But hey, what’s wrong with 20%-30% compounded returns? Here is his homepage.

When listening to Lo’s speech I was struck by the comment he made on a factor that was predictive in up markets (EPS Growth), but not down markets. Kirkpatrick decided it was not a good search criteria, but he could not explain the difference in behavior related to the different market environments. Lo offered an explanation based upon his AMH model. In bull markets, investors are making decisions using the part of their brains (neocortex) associated with analytical skills and seeking to maximize profits. In down markets, the ‘fight or flight’ part of the brain influences the investment decision making process and fear dominates the markets.

I have done quite a bit of work with factor based stock models, but never divided the periods based on trend.

My Q: Is there any research out there that takes a look at the most predictive factors for up markets (say, above the 200 day moving
average) vs. down markets (below)? Any academic literature? Thoughts? Any quant shops that do this?

He has also recently published the definitive textbook on technical analysis, Technical Analysis: The Complete Resource for Financial Market Technicians.