CHARGE rolls until the number of dice being thrown drops by one.
If you are attacking with 5 armies, CHARGE attacks until the total troop count drops to 3. At that point, clicking charge again would continue attacking until the troop count drops to 1.
I believe the current Blitz algorithm is a major contributing factor to the raging over "unfair" dice. Granted, people are bad at statistics, but the odds are wonky if you roll blindly until you run out of troops. Given an even number of attacking and defending troops, the odds for the attacking army to win drops below 50% at 4 troops and tanks at 3:
So you start out with a 5v4 or 7v4, you get a bad roll or two and boom: you're entire army is toast. String a few of these together (once per game?) and it's easy to get the impression that the dice aren't rolling fairly.
I think you only need to compensate for this at the low end because larger numbers not only favor the attacker (leading to the "correct" outcome) but can also absorb variance in the dice outcomes.
Here are the raw stats from the original paper:
This is a great idea. Thanks for all the effort and research put in!
Awesome work dude!