Welcome to the world of algorithmic trading! Just to be clear, when you say your results are "different", are you meaning 1) they are materially different than your backtest engine results that you (hopefully at least periodically) compare to your live trading performance, or 2) do you mean they are different than what you would expect in the future based on your back test results?
If you mean 1), then what backtesting engine are you using and what kind of orders are you using in the backtest? If 2), then what methodology / metrics are you using to define "different"?