I agree with the -202 Det adjustment, which is the raw Det of a naked lvl 50 BLM using the lvl 1 weapon.
I also think that Int has a similar adjustment, in the sense that the Int we see, obviously and not obviously is (Job Base + Race Base + Allotment + Gear). So the lowest baseline to use for Int would be just the Job Base, for which we need to know the Race Base, and we can assume +30 Allotment.
This gives us a few options for these 2 parts of the equation:
A*Int , A*(Int-IntBase) , A*(Int/IntBase) , A*[(Int-IntBase)/IntBase] , or A*[1+(Int/IntBase)]. The same structures go for Det.
Weapon Damage most likely has its own linear coefficient, and is THEN acted on by Det... Although it IS possible that WD is acted on by BOTH Int and Det, but it is unlikely from a coding standpoint... But I need more data to confirm that.
As to variance in crit damage/healing. The modifier for crit is 1.5x, there is no variance in the modifier itself. However this does not help much since the variance in the "uncritted" damage is still there. I could just take the total crit damage and divide it by 1.5, then throw it back in... but meh, I'll try that later.
Back in RIFT, before 2.0, critical hits were an excellent source of data because they had a very simple equation: Your max damage * 1.5
So for a given set of stats and buffs, a given ability would have variance in its non-crit hits, but every single crit hit for the exact same amount. You could crit for 3287 with an attack, and later under the same conditions, that attack would crit for 3287. So we could do a lot of optimization based off crits, stating that the crit was 1.5x your 100% damage, and your real damage would be ~ 95.3% of your 100% damage, given a +-5% swing from your average.
Sadly there is no way to get a 100% consistent output for any ability in this game, which makes the math a little harder.
Edit: BLM Job Int seems to be 224.
