I think you are saying the same thing (in general). Because I certainly agree with your last paragraph in particular."Winning" and "losing" are very simple to define--no need to complicate a simple concept with things like opportunity cost. Return on Capital and risk/reward are for sure important metrics, but they are most certainly separate.
Rolling an underwater position is explicitly realizing the loss of that position. It just happens to be also simultaneously opening up a new position. While some of us prefer that new position offset the loss of the previous one (so, roll for net even or credit) not treating them as separate positions from a tracking perspective is, at best, inefficient for the trader. In the context of tracking one's trades, abstracting the losing trade into some multi-cycle position is really more akin to cooking the books, which of course doesn't make sense if you're the one that wants 'the books' in the first place.
Trade trackers are not simply an accounting exercise, because everyone can get that just by looking at their broker's basic performance tools. A tracker's primary function is to capture a trader's logic/criteria used to enter/exit the position, explicitly for the purpose of generating a statistical, quantifiable data set from which the trader can evolve their overall strategy. The accounting rack up is simply the metric by which the trader's strategy is assessed.
Trade trackers are the only reliable method by which a trader can objectively assess whether or not a losing trade was a bad trade or simply a statistical eventuality.
“Cooking the books” is a pejorative phrase… I’d argue that “rolling” is more often intended to be “using all parts of the Buffalo” but on accident.
How else can people convince themselves how an end of year “tax refund” is awesome?
edit—- I also want to clarify I do regard rolling as being wrong