To complement the luck score (how many troops you are ahead/behind average based on your rolls) add a luck percentile, that estimates your odds of having done worse.

As discussed in http://www.wargear.net/forum/showthread/3055p1/How_unlucky I'd be willing to supply code in your language of choice for being told the dice involved in a given attack roll in your format of choice, and returning the mean and variance for the losses.  (By attack roll I mean one roll of attackers/defenders, and then total up the losses.)

If we sum the variance up over all combat rolls in the game, that gives you the total variance.  Its square root is the standard deviation.  Divide the luck score by the standard deviations to get the z-score.  Look that up in a standard table of normals and you get the percentile.  (There are libraries already built for the normal table lookup in most languages.)

I almost always prefer designer features to be at the top of Tom's to-do list, but this one give the luck stats relevance, and makes them more useful for both players and designers.

M57 wrote:

I almost always prefer designer features to be at the top of Tom's to-do list, but this one give the luck stats relevance, and makes them more useful for both players and designers.

With just a little more work, you could calculate the strength of the correlation between victory and the luck percentile.  Which would tell you how much of a given board is skill, and how much is good combat rolls.

As has been pointed out in other threads, there are so many factors that influence victory. In fact, there are cases where an inverse relationship could probably be made for victory vs. luck. Take for instance games with multiple players where the player who is strongest out of the blocks tends to get picked on.  On the other hand, there are boards where early luck is clearly desirable, and end-game clean-up luck effectively becomes nothing more than garbage stats.  While interesting, I question how much value such a chart might have.

Regardless, a straight luck percentile stat with a handy-dandy range slider would let players isolate scenarios, giving them the option of analyzing in whatever context they perceive appropriate.

+1 to this idea.

Big +1 from me too.