Roywyn

Since there is a lot of uncertainty about whether spells are on a 1-roll ystem like melee attacks or on a 2-roll system where crits can miss, I tried to find a proper test for it.

If anyone has done some testing or found some proper information about this before, by all mean, post or point me to it.


With, say, 30% crit and 15% miss, it would take a huge amount of data to make sure that it's a 1-roll or 2-roll system.
What I thought about was testing with the rogue ability 'Cloak of Shadows' (http://www.wowhead.com/?spell=31224) to use a test environment with extremely high miss chances.
Cloak of Shadows effect is written as "Apply Aura: Reduces Attacker Chance to Hit with Spells (Spells), Value: -90"

I did only a short series of tests (thank you, Nyl!) with the following setup:
- level 70 mage (deep fire raid spec) vs. level 70 rogue
- mage uses Molten Armour
- mage has ~27% crit with fire spells
- mage has ~18% hit from gear plus 3% from talents, ~21% in total
- rogue attacks with Cloak of Shadows, with fast daggers and under haste effects
(used Slice and Dice and Blade Flurry, to get a large amount of melee attacks)
- mage has 94% base hit vs. level 70 and 90% increased miss chance from Cloak of Shadows

We did this four times, and I had 16 non-resists, 2 of them were crits and something around ~30 resists of Molten Armour in total. I feel so stupid for not having tracked the exact amount of resists.

Let's look at the possible theoretical numbers:
I have 96% base hit chance, 6% during Cloak of Shadows. With +21% hit I'd have 9% hit if CloS was not affected by spell hit, 27% hit if spell hit can overcome CloS.
I didn't do all the calculation before the tests, so I don't know if my 27% spell crit chance was greater than the possible 27% hits through CloS (not sure about the decimals of the percentages).

If he hava a one-roll system, we should see no hits at all, or next to no hits, just a lot of resists and several crits.
If we have a two-roll system, we should see 27% of the non-resisted molten armour hits as crits, the other 73% should be hits.


Interpreting the test data

From the data we have, it seems that spell hit overcomes CloS. About 1/3 of the procs hit, way more than the possible other 9%.
Out of the 16 non-resists, only 2 were crits. That's is way less than what a 1-roll system would predict (no or <1% non-crits). It's also less than a 27% crit rate among all non-resists, but then again, there is very little information to extract from a test with 16 data points.

So, a result for now, I'd propose the following:
- spell hit can overcome abilites that increase the spell miss chance
- spell hit and crit are on two different rolls

 

dfinberg

Why not find someone with a level 60-65 mage, and send him to play with Dr. Boom without any +hit gear on, but plenty of crit (say 20%). At 83% hit, 20% crit, my rough math shows the overlap region being 2 sigma high (for 20% of hits are crits) and 2 sigma low (20% of casts are crits) at 600 casts. That's pretty tedious, but not that bad right? Quick cast testing with huge miss is the best approach, but I'd go for PvE since some PvP mechanics are somewhat weird, and getting big numbers of casts will be tougher.

1 year old twins means no WoW for me. But you just wait, as soon as they get a little older it will be my own stable of gold farmers.