Quote from StockBagger:
I have backtested the dow 30 stocks and SPY from the beginning of the year and have done quite well.
Backtesting must cover at least one market cycle and preferably two. Since this data is what you used to tune your system, it must be validated with at least one market cycle outside of the two you used to develop you system.
"Since the beginning of the year" is woefully short for an adequate test. Even after you have simulated your system you must prototype it. You do this by paper trading it. Only then do you commit money to your system. Both backtesting and paper trading have differences from reality which have been covered elsewhere on this forum.