I swapped the label to Special Sv. I'll try the re-rolls somewhere this week (I'm frantically trying to finish painting another squad of models

)
Battle Focus requires quite some re-work. My current mathematical model allows for virtually any kind of "on X" kind of mechanic but I never applied it as such.
I wrote a whole code-block for "1 attack -> 1 wound" and glued on a variable amount of attacks and damage before and after this. For Battle Focus, I'd have to use my new mathematical model and re-write the whole thing, including any special rule and re-rolls implemented thus far.
This is one of the main motivators to rewrite the calculators from the ground up, and because I can take it even up a notch where people can invent their own rules and try them out. That would be helpful for designers, beta-rules and can serve as a backup in case I'm slacking on keeping the calculator up-to-date. It's that flexible as a model.
To explain (if you're up for a short read):
My earlier statistical models relied on a statistical distribution. I think I posed the theory of using a binomial distribution for combat calculations somewhere in the earlier 2000's (2004?). For Warhammer Fantasy Battles, 6th edition, it was sufficient for -most- combat calculations. And it certainly was quite novel at the time.
But a binomial distribution has 3 restrictions: (1) a fixed number of tests / attacks, (2) a fixed outcome per test (fixed number of wounds, not d3/d6, etc), (3) a fixed chance for each test (so 1 type of attack only).
The entire coding of a combat calculator had to focus on the chance of 1 attack causing 1 wound. With p (chance to wound), n (number of attacks), d (damage per wound) you have all the parameters for the distribution.
I included this model
http://tools.druchii.net/9th-Combat-Cal ... .25:10:1:0 in the "generic" calculator.
I wanted to overcome these restrictions so that I could model the impact of a chariot (D6 Impact Hits at the time) and multiple wounds. I solved this by calculating the outcomes for each number of attacks, and then adding up the individual outcomes. For example, the chance to do 1 wound was the chance to do 1 wound on 1 impact hit, plus 1 on 2 impact hits, plus 1 on 3 impact hits etc.
The same logic was applied for random damage.
But this still required the same code for "1 attack -> 1 wound".
When I built these calculators, I did what I had always done and made that code-block. I later figured a simple, generic model that supports random attacks, random damage, and any "on X -> event" kind of rules. It uses a plain discrete mathematics and polynomials. I did go back to re-write the random attacks and damage using this method, but did not re-write the entire "chance to wound" code-block because I had no need.
And that is why this will require more effort... Although I have the solution lying right in front of me, it requires me to rewrite a sizeable portion of the calculator that is still stuck in mid-2000 logic.

I will get there. That's a promise. I just don't know when.