Unfortunately, I don't model "reversing" orders - I have signals that generate targets and exits...once I'm out, I'm out until the next signal.
However, I'd say that it's unrealistic to assume that if you reverse (add a tick of slippage on the stop), you need to do the same on the reversing...