Originally Posted by mjball
You only casted 9 conflags in each dummy fight. That sample size is WAY too small - you need at *least* 1000 or 2000 conflags in order to get reasonable data.
|
If we were trying to use brute force statistics to determine previously undetermined behavior, this would be accurate. Since we have an accurate model of the base spell's behavior, it is not true in this case. In particular, because there is no random variabilty in conflagrate if you filter out the crits, we need only a few samples. You do have to query the logs to get the raw data.
If you query the first log with 2xT8 and 2xT9, all immolate ticks were 1327-1328. Non-crit conflagrates all hit for 5610.
The known formula for an immolate tick is
|
(157+ 0.2*spellpower)*(1+aftermath*0.03+emberstorm*0.03+improved_imo*0.1+glyph_immo*0.1+2T8 bonus*0.1+4T9bonus*0.1)
|
The known formula for conflag with 2Tx8 is
|
(157+ 0.2*spellpower)*4*(1+aftermath*0.03+improved_imo*0.1+glyph_immo*0.1)*(1+emberstorm*0. 03+2T8bonus*0.1)
|
Given a standard destro spec (without immolate glyph), the predicted ratio of conflag to immo tick with 2xT8 is
|
4*(1+2*0.03+3*0.1+0*0.1)*(1+5*0.03+1*0.1)/(1+2*0.03+5*0.03+3*0.1+0*0.1+1*0.1+0*0.1) = 4.223
|
The observed ratio of conflag to immo tick is 5610/1327.5 = 4.226
So far, so good.
Now, if the 4xT9 bonus affect conflag the same way 2xT8 does, we should expect a conflagto immo tick ratio of ~4.22. If it is bugged and doesn't affect conflag we should expect a ratio of
|
4*(1+2*0.03+3*0.1+0*0.1)*(1+5*0.03+0*0.1)/(1+2*0.03+5*0.03+3*0.1+0*0.1+0*0.1+1*0.1) = 3.886
|
In the second data set with 4xT9, we see immo ticks for 1334 and conflag hits for 5568. The ratio of conflag to immo tick is 5568/1334 = 4.174
So we're definitely not ignoring, the bonus, but something is different. If we take the intuitive leap of "maybe they stuck the coefficient in the other set of parentheses", we can hypothesize that the fomula might be
|
(157+ 0.2*spellpower)*4*(1+aftermath*0.03+improved_imo*0.1+glyph_immo*0.1+4T9bonus*0.1)*(1+ emberstorm*0.03+2T8bonus*0.1)
|
Doing so would give us a conflag to immo tick ratio of
|
4*(1+2*0.03+3*0.1+0*0.1+1*0.1)*(1+5*0.03+0*0.1)/(1+2*0.03+5*0.03+3*0.1+0*0.1+0*0.1+1*0.1) = 4.171
|
Which leads me to believe that the 4 piece T9 bonus works, but the proper formula for conflagrate before modifiers is:
|
conflagrate damage = (157+ 0.2*spellpower)*4*(1+aftermath*0.03+improved_imo*0.1+glyph_immo*0.1+4T9bonus*0.1)*(1+ emberstorm*0.03+2T8bonus*0.1)
|