I tried to test Steve Rowes weekly system on the S&P500 Stocks.
It works fine with his parameters, but I was going to determine, how robust they are.
To my suprise, the optimal parameters for the periode and the trailfactor of the ATR-Trailing-Stopp from 2000-2007 are very poor for 2008-2012. Even the walk forward analysys looks quite chaotic.
An ATR-Trailing-Stopp is nothing specific to Stewe Rowes system, so I'm a litte confused. How would you optimise a system for robustness.