As far as I can imagine, the odds on a long string of attacks are the same as long as the last few territories are the same. That is, permutations of the first few territories don't make a difference, so attacks like 20v1,1,1,6,1,1,1 should be the same as 20v6,1,1,1,1,1,1. That'll mean that no matter where the big stack is, I should be able to place it first in line on defense, then place my smaller attackers before the large force and have it all work out the same.
This solution you've got is perfect.
In this case
Attack 4,6 vs 3,2 = 72%
Attack 6,4 vs 3,2 = 75%
The odds in the first one include a win 4v3, so that the next attack will be 3v2, which is poor, hence worse odds. I don't think anyone is silly enough to make the mistake of auto-attacking with the small force, so this should be fine. If the force is small enough, the chance of victory is negligible and you can place it first. If it isn't, hmmm.... when the numbers are small I can't intuitively tell what the best attacks are. Let me get back to this.