I think we can all agree that entering a territory defended by 150 Gold of units with a single 3 Gold Trooper is a Dumb Idea. And yet...
And yet that strategy has been used effectively throughout history.
Net Result: For every 1 gold the enemy lost, I lost 4.67 Gold.
And you got/kept the country, so get more gold and research points the next turn (and as Elwyn mentioned, you are not comparing that to losses in RTS). If sending out lone troopers is an effective strategy that will win the game, then there is a problem with game balance. If not, a loosing strategy being a little less expensive than a winning one isn't a big deal.
Are you going to continue sending lone troopers into entrenched territories to soften them up? Other than complaints of auto-resolve results, nobody has said that that was their main strategy or that insane AI wasn't good enough because it is easily defeated by lone troopers.
I think (and that's my honest opinion) that there shouldn't be an RTS simulation in the background because it's completely pointless. You could roll some dice in the background, it would be the same. At least, rolling some dice would reflect the Risk-origins much better....
It is pointless to be consistent or logical? So the game should require 2 different strategies, one that works well if you autoresolve and one for the RTS?
Somehow a single bomber balloon or upgraded imp fighter being able to wipe out vast armies of ground troops without anti-air capabilities is much more realistic than a lone trouper managing to take out a couple hunters or armours?
How much would it slow down the strategy phase if you had to take all the unit types, upgrades and counters into account there, as well? Would that make the game more fun?