Howdy!
I've often wondered about that also, and I'm sure TheHutts or KezzaMachine will have some insight also...
excellent question, thanks for asking
For background, I have been a chess composer, and thoroughly enjoy many kinds of game also hehehe

It amazes me, that folks come up with games that are so enjoyable for people.
learning what was involved, and the 'story behind the story' would be awesome for SWMinis.

though I digress,
from what I've garnered about the making of SWMinis over the years,
(most of which gleaned from Hothie, _NickName_ and a couple of others)
40-odd people were involved in playtesting stats, most being the former mods etc from the WotC site;
the rest was done via some simulation methods (see "The Electronic Journal of Combinatorics" for ideas about testing gamestates of a game,
specifically, test for fibonacci-wieferich bias, or for thue-morse-turing bias...)
Not many tests were performed, at least, from a 'statistical' standpoint,
fewer than 10-000 games were played with the first 2 sets, which were developed at the same time.
something like 6 or 7 000 game reports were made; brief things, only a couple of pages if that, usually a paragraph about a game with some corroborating sketches of the game-state.
we have nothing near what chess does - many compositions are already known since 1883 and are written in an almanac, as that game is NP-complete... similar to Backgammon... this is inclusive of images of the gamestate on the board.
certainly a few thousand tests ought to be expected for a 'rigorous and compelling" analysis or gamestate evaluation,
to see what meta-combinations might arise etc...
rather than do that, WotC used the 'rapid prototyping' model, and let us have all the fun ,
and just 'used the force' to get a 'feel' of where balance oughta be.

Subsequent sets had less playtesting (less 'practical testing", and more of that simulated EJC kind).
in that context, the game works exceedingly well.

VSets, comparatively, have a lot more playtesting relative to what WotC put in,
though, are calibrated more to the later sets IMHO.
I am unaware of exactly how it is done - I suspect its a couple of paragraphs about 'weird' game situations,
and maybe illustrations or sketches of the board/pictures etc...

I have subjected a couple of grad students and a few fellow freethinkers to similar questions:
asking "at what point does SWMinis diverge from chess, and,
is SWMinis at any time t NP-incomplete by any factor other than that pieces are continually being innovated?"
I suspect, at some distant point in the future,
SWMinis will become NP complete, when:
both, the maps become fixed,
and, pieces cease to be made.
When the maps are fixed and the pieces are fixed, the combinations would become large yet fixed.
already, using nPr, it is a huge number of possible squads,
and a huge number of pieces/gamestates... we're talking in the 10^9 range. That's a lot.
the bell curve on it is also interesting to reflect upon...

Certainly though, SWMinis has many orders more combinations than Checkers or Chess

I've waffled on for perhaps too long;
I'm keen to see what other perspectives and insights folks have on this question.
Cheers.

thanks for the extra insight, TheHutts;
and, in recent times, the VSets team have shared a lot more of the behind the scenes;
so, would you say that the testing is as extensive as 'chess compositions"?
ie, move, turn, phase etc, as well as the grid-square it was mapped to?
that'd be awesome, as those would become part of a SWMinis Almanac hehe

I don't envy that task, as it would be similar to l'biblioteca d'babel...
given all the maps and pieces out there hehehe

Indeed, WotC did less playtesting the more the game went on, and the last 4 sets... it seems it was exceedingly theoretical in the testing.

trying to model how the pieces move and the interactions,
was an extreme part of the appeal of SWMPlay's star wars game concept.
|I hope they're still chipping away at it - I'd help them if my coding were better hehehe|

being able to see where all the pieces could go within the different maps would be a great asset,
not to mention that unenviable of tasks - finding all the min-maxer optimal squad combos
as well as composing other kinds of game, etc.
(Against All Odds format: you have one squad which has no more than 50pts vs 250pts. there are a couple of victory conditions: gambit, wiping out of squad, or earning double the squad's size in unique kills etc)

one variable I have been testing over the years is: starting position - does altering the starting position on maps have a significant impact on the outcome of a game of SWMinis?