I have been returning to the idea of creating a sub-universe of stocks (eg within SP500) or maybe ranking all the stocks in the SP500 based on their past results with a specific trading system. I have noticed that some stocks frequently mean revert intra-day while some almost never do. When I run the backtest on individual stocks some have a very low win rate (eg 30%) while others are into the 60%. I am thinking that the ranking system or sub-universe could be re-optimized (not sure if that is best word) every 3-6 months.
For a breakout or momentum system, using the results of the system’s backtest to create a sub-set of stocks or ranking system would instead select stocks that did not show a tendency to mean revert in the short term
Curve fitting ?