if i have a mean reverting strategy with 2 parameters to optimize, can make sense optimize separately long and short condition? or it sure give an overfitted result? is there some rule of thumb to know in this situation?