The Power of Improved Shadow Bolt
A while back, I switched from a classic affliction build (raiding 42/0/19) to DS/destro (0/21/40). I had 202 spell hit, over 1100 shadow damage, and 21% chance to crit with Shadow Bolt. My DPS dropped like a stone against multiple T5 bosses. 
Warlock's Den posters sternly advised at least 25% spell crit for Shadow Bolts after talents (see Crit needed for Destro build?, for example). I collected some side-grades, traded shadow damage for spell crit, and respecced again after I was up to the approved 25%. Warlock's Den was right; this time my DPS went up quite a bit, even though it was now hard to move while casting.
The thing that bugged me about the whole incident was that Leulier seemed to be wrong! This seemed almost impossible to believe, as I was convinced (and still am) that his DPS spreadsheet is the greatest thing since sliced bread for warlocks. However, it had indicated that I would be doing better damage as destro after 20% SB crit.
So, I went to build my own set of models in a spreadsheet and checked to see if Ruin and Improved Shadow Bolt, separately or together, changed SB damage dramatically at any point. I know that Leulier's spreadsheet already models these variables, but not in such a way that I can see or understand them, so I had to do this by hand.
I assumed 100% spell hit, and varied spell crit. My ISB model is a very simple one which assumes I am the only shadow caster involved, I never let my ISB buff expire due to time, and that the fight has been going on for at least four Shadow Bolts already.
Sure enough, there is a sharp bend in the ISB probability curve at 25%, but is it enough to affect SB damage altogether? I continued working on the model to calculate how average SB damage varies with spell crit.
The average Shadow Bolt rank 11 does 572 points of damage, and has a spell coefficient of 85.71%. I assumed that the warlock had +1000 shadow damage from gear.
Damage per average Shadow Bolt basically goes up linearly with spell crit if all other factors are held constant. What surprised me about this model is that ISB adds as much damage as Ruin, but it's only a first-tier talent! I'd always thought of it as filler, silly me. No wonder Blizzard wants to nerf it; but I understand that Shadow Priests would then be largely out of business, so they can't.
One limit of the model is since it just calculates average damage, it doesn't feel much like the actual experience of playing a destrolock, where the moment-to-moment difference between crit and non-crit Shadow Bolts is huge and there are no average shadowbolts. In-game damage for a destrolock depends not so much on damage from individual crits, but on chained crits because of ISB, so it's more about luck than about averages. However, averages are easier to model!
In a raiding situation, you would also want to account for shadow priests and afflocks who consume the ISB charges, but shadow priests spend a lot of their cast time doing DoT damage, and raiding afflocks will also have ISB among their talents, so I don't think that they reduce ISB uptime that significantly.
Ruin really smooths the average-SB damage curve out even with ISB factored in. And ISB helps an afflock too. So there is no clear reason so far that the breakpoint of about 25% spell crit makes such a big difference in warlock raid DPS when switching from affliction to destruction. The model grows more complex next entry, and I may have an explanation (or two)!
Warlock's Den posters sternly advised at least 25% spell crit for Shadow Bolts after talents (see Crit needed for Destro build?, for example). I collected some side-grades, traded shadow damage for spell crit, and respecced again after I was up to the approved 25%. Warlock's Den was right; this time my DPS went up quite a bit, even though it was now hard to move while casting.
The thing that bugged me about the whole incident was that Leulier seemed to be wrong! This seemed almost impossible to believe, as I was convinced (and still am) that his DPS spreadsheet is the greatest thing since sliced bread for warlocks. However, it had indicated that I would be doing better damage as destro after 20% SB crit.
So, I went to build my own set of models in a spreadsheet and checked to see if Ruin and Improved Shadow Bolt, separately or together, changed SB damage dramatically at any point. I know that Leulier's spreadsheet already models these variables, but not in such a way that I can see or understand them, so I had to do this by hand.
I assumed 100% spell hit, and varied spell crit. My ISB model is a very simple one which assumes I am the only shadow caster involved, I never let my ISB buff expire due to time, and that the fight has been going on for at least four Shadow Bolts already.
Sure enough, there is a sharp bend in the ISB probability curve at 25%, but is it enough to affect SB damage altogether? I continued working on the model to calculate how average SB damage varies with spell crit.
The average Shadow Bolt rank 11 does 572 points of damage, and has a spell coefficient of 85.71%. I assumed that the warlock had +1000 shadow damage from gear.
Quote:
Without talents:
Average SB damage = (572 + 857.1) * (1 + spell crit chance * 0.5)
With Ruin, not ISB or Shadow and Flame:
Average SB damage = (572 + 857.1) * (1 + spell crit chance)
ISB chance = 1 - (chance of no SB crit)^4
damage with ISB = Appropriate avg. non-ISB SB damage from above * (1 + ISB chance * 0.2)
Average SB damage = (572 + 857.1) * (1 + spell crit chance * 0.5)
With Ruin, not ISB or Shadow and Flame:
Average SB damage = (572 + 857.1) * (1 + spell crit chance)
ISB chance = 1 - (chance of no SB crit)^4
damage with ISB = Appropriate avg. non-ISB SB damage from above * (1 + ISB chance * 0.2)
Damage per average Shadow Bolt basically goes up linearly with spell crit if all other factors are held constant. What surprised me about this model is that ISB adds as much damage as Ruin, but it's only a first-tier talent! I'd always thought of it as filler, silly me. No wonder Blizzard wants to nerf it; but I understand that Shadow Priests would then be largely out of business, so they can't.
One limit of the model is since it just calculates average damage, it doesn't feel much like the actual experience of playing a destrolock, where the moment-to-moment difference between crit and non-crit Shadow Bolts is huge and there are no average shadowbolts. In-game damage for a destrolock depends not so much on damage from individual crits, but on chained crits because of ISB, so it's more about luck than about averages. However, averages are easier to model!
In a raiding situation, you would also want to account for shadow priests and afflocks who consume the ISB charges, but shadow priests spend a lot of their cast time doing DoT damage, and raiding afflocks will also have ISB among their talents, so I don't think that they reduce ISB uptime that significantly.
Ruin really smooths the average-SB damage curve out even with ISB factored in. And ISB helps an afflock too. So there is no clear reason so far that the breakpoint of about 25% spell crit makes such a big difference in warlock raid DPS when switching from affliction to destruction. The model grows more complex next entry, and I may have an explanation (or two)!
Total Comments 0
