I have never heard of baseline co. But that sounds very interesting. Do you know whether they have fundamental data from before 1986? Now that would be excellent for more backtesting.
I wonder if you could outline the trading rules you used to generate your performance data, eg, stop loss level, exit strategy.
Stop losses> No stop losses were used to generate those results. Considering the performance of the historical tests, disaster stops losses may be used. Exit strategy> Very simple. Liquidate all positions on the last friday of the month at the closing. The goal was to build a low maintainence trading model which is diversified and capable of generating profits consistently by adhering to a specific fundamental+technical trading/investing philosophy. (aka market based value/growth combination+simple momentum criteria).
Exit every month ? Then , would you consider taking position a la pair trading > long top rank & short benchmark ?
This is what I have been thinking of doing: Long say the top 5 stocks and short the bottom 5 stocks to get a market neutral environment. I will run some tests and post the results here. Its less of a pair trading strategy than more of a market neutral long/short strategy. Naturally, I would rather build a model which will identify candidates to short. But shorting is quite different in equities (asymmetric unlike FX).
I would run a corr matrix and hedge with futures in lieu of share-pairs. Blended tax would be lower with a higher correlation. I don't think it's wise to run the quant on the long and short.
Yes I just generated some results. Although annualized the bottom ranked stocks tend to do poorly compared to the top, during strong growth cycles many of the lower ranked stocks were the dot coms. The bottom ranked stocks outperform the top ranked (which tend to be more value oriented than the bottom).