backtesting is usually meaningless. People try alternative histories, until they find one that works. They rarely are reliable into the future
Best thing is to backtest the amount of time that you get the best results from, then trade the program with all your money. Thats what most people do.
This is not enough information to give an answer. If you cannot even apply high school stats... Like Standard Deviation and Confidence Intervals to your data... Then you should not be trading... Unless to NEED to give away your money to rich brokers and traders.