I agree. One useful step is to separate signal differences from execution differences.


For every demo trade, log the signal time, spread, requested price, actual fill price, commission and any SL/TP modification. Then compare the same trades one by one with the backtest.


If the entry signals are different, the issue is probably data or candle construction. If the signals match but results are different, then spread, slippage and execution are more likely the cause. Testing with variable spread, especially around rollover and news, can also reveal a lot.