MatthewNYC

Hey guys I have designed an application to allow me to test certain variables in a stock trading system, i.e. to determine which variables cause the system to become more profitable. Obviously the output is the profit.

What I would like to know is, should I test each input variable separately (without including the other variables) to determine if they increase the profitability of the system alone or should I test them as a group? For example, lets say there are 8 input variables, would I include the first 7 variables at some random constant value while varying the last one as a test or should I exclude the other 7 from the trading system completely and just test 1 at a time?

Secondly, most of the variables show a slight correlation I took care to remove any variables that were too closely correlated.

Thanks,
Matthewnyc