Don't get me wrong. But the progress in a wargame is often exponential. Thus the imbalace grows more over time as well. Lucky rolls should be able to overcome this for a while.
What? The opposite is true, at least if you are going by rule of thumb (I wouldn't dare qualify such a statement absolutely).
If you are playing a strategic game with an element of chance at some base level, then you still have the non-random state information of the larger parts of the problem to work with. In other words, you use dice or what not to randomize the individual battles somewhat, but otherwise the units are where they are at, they have their mechanics fixed even if those mechanics are chained to dice-generated variances, the board areas, whether they are geographical abstractions or not, are generally fixed, and typically you know what the end goal is, etc..
So the only major luck involved is on the dice, really, assuming your opponent doesn't make too many random mistakes. A skilled player will be able to optimize his battles and this reduces the effect of luck, to the point where good luck helps him but he would win anyhow, and bad luck is mitigated through careful play.
A bad player, on the other hand, is ruined by 'bad' luck, which they have more of because they cannot optimize their battles, and may or may not capitalize on good luck, either squandering the margin, or just getting dumb lucky again. In general, it doesn't work out for them as well.
In either case, luck is dominated by skill because skill gets you stabilization of the higher level game states. It is skill which eventually avalanches, so that no amount of luck will help a losing player, at which point they simply concede the game and leave you hanging with a hollow victory.
So that is the key issue of balance for most strategy games. If you make it mostly pure skill (which hard core people like), then you need to play at conventions or something if you are very good because eventually your friends just get tired of losing all the time. If you put too much randomization into it, the game turns into a crap-shoot and your player base is turned off because most people play war simulation board games so they can play at solving puzzles and test their skill against the opponent(s) at the same time.
Right now I am playing around with some different mechanics to get a subjectively optimal feel that would make me happy. I want a game that requires a fair amount of skill and optimization, but which has some new twists on randomization at various heierarchies of the state information, so that is harder to dominate the puzzle through careful play. The randomization shouldn't make your skill arbitrary, however, and it should be fun no matter what. I am also employing varying degrees of hidden information, so that nobody can be entirely sure if another player is winning or losing simply by looking at the visible state information, but will have a fairly good idea of how they are doing.
In the end, I want a wargame where a highly skilled player can still be knocked out of play when the odds would appear to be with him, and where it is rare to just quit because you can simply see that you can never win. It shouldnt be too often, but enough that players know they still might have a chance. It should keep a skilled player on his toes, and more than anything, it should be fun to play.
Im getting closer. Ive been dreaming of this game for years.