And therein lies the problem: there is no boss in current content which hits that hard. None. (I haven't finished parsing out NiM EC yet, but preliminary numbers still look well shy of NiM KP) In reality, the steady-state damage is about half that, with rare spikes that are much higher. If damage numbers were as high as you assume, then I fully agree that the self-heal is a poor substitute for more mitigation. Damage numbers are simply not that high (remember: wet noodle).

Your assumed balance of KE to IE damage is also quite far from the mark. Only 9.41% of damage is IE, which really isn't that much.

Finally, I'm not really following the expressions you give at various stat budgets. Could you explain your numbers a bit more? Also, your shield/absorb/defense ratios look quite off across the various stat budgets (my main sits at 1529 divided as 552/510/467 and is ideally balanced to within 0.0044%).

Are you saying that there are no bosses that have more than 10% internal/elemental damage? In that case you would have sometyhing like:

1600) 0.3636*[0.9*(1-0.4087)+0.1*(1-0.23)]=0.2215

1700) 0.3522*[0.9*(1-0.4087)+0.1*(1-0.23)]=0.2146

now to explain the numbers:

247/6 is the heals from the combat stance

(.02*HP*3)/15 is the heals from 3 stacks FL/TT self heals

515/25 is the heals from the relic

now we compute for 2000 dps

1600) 2000*0.2215-(247/6+(.02*27000*3)/15+515/25)=273 dps taken

1700) 2000*0.2146-(247/6+(.06*26000)/15)= 284 dps taken

In this case the healing relic is better than the extra mitigation. However, if incoming damage reaches N such that

.2146*N-247/6-(1/15)*(0.6e-1*26000) = .2215*N-247/6-3*(0.2e-1*27000)*(1/15)-515*(1/25)

then 1700 buget is better... this N is 3565.

The stat weights I use are based on 1/2 90% accuracy and 1/2 100% accuracy, even though this number is not true around the board, I think it helps create a balanced build.

To determine optimal stat weights I used the following formula:

> d90(dr):=1/(100)*(10 +6+15+ 30 * ( 1 - ( 1 - ( 0.01 / 0.3 ) )^( ( (dr+46) / 50 ) / 0.55 ) )):

> d100(dr):=1/(100)*(10 +6+5+ 30 * ( 1 - ( 1 - ( 0.01 / 0.3 ) )^( ( (dr+46) / 50 ) / 0.55 ) )):

> sh(shr):=1/(100)*(5+15+20+ 50 * ( 1 - ( 1 - ( 0.01 / 0.5 ) )^( ( shr / 50 ) / 0.32 ) )):

> ab(abr):=1/(100)*(20+4+50 * ( 1 - ( 1 - ( 0.01 / 0.5 ) )^( ( abr / 50 ) / 0.18 ) )):

where d90 is the defence chance from 90% accuracy attacks

d100 is defense from 100% accuracy attacks

the +46 defense rating is there for rakata stim

the equation for squishiness (without armor and resists) is; ac90*(1-d90(dr))*(1-sh(shr)*ab(abr))+ac100*(1-d100(dr))*(1-sh(shr)*ab(abr))

where ac90 and ac100 is the fraction of attacks that are 90% and 100% accuracy.

I use the method of lagrange multipliers to find the 3 points for defensse absorbtion and shield to minimize the squishiness function.

In maple:

with(Student[MultivariateCalculus])

LagrangeMultipliers(ac90*(1-d90(dr))*(1-sh(shr)*ab(abr))+ac100*(1-d100(dr))*(1-sh(shr)*ab(abr)), [abr+dr+shr-N], [dr, shr, abr])

where N is the total stat budget you hvae for shield+defense+absorb (in rating).