i am just a novice but here is the solution i am thinking of :
All trades get closed at the last bar, so if you have a large amount of trades opened when the simulation reache the last bar, it could possibility alter your backtest result, the solution is simple get more historical data so that the portion of trades closes be cause of the end of the backtest become insignificant compared to the trades closed by your exit signal.

in my own backtestes the trades closed be cause the simulation ended are not more than 3% of the total number of trades.