I hope you know what out of sample, curve fitting, transaction fees, slippage and the like are. Also, make sure that your stops are hit b4 your limits in the backtest. When you test with time bars and not tick data you dont know what happened first within the bar.
50k to 1M in less than 10 days...