Ok, so now you "just" need to factor in long-term maintenance.

Besides, how often do you write code twice just to see which way is better? What you have is a comparative measure that is completely useless for measuring new code, i.e. code which implements a new feature.