Spell Hit, Spell Crit and mages - Page 2

ziggy · 12/22/06, 11:32 AM

Originally Posted by Eylirria

Spells are believed to work on a 2-roll system, where the first roll determines whether it was a miss/crit/hit and the second roll determines damage lost due to resistances. (Binary spells are on a different single-roll system that takes into account resistances already)

Melee special attacks were believed to be calculated using the same method as white melee, yet this was proved false. I don't think we can make the assumption any more.

If crit caps out as the chance to miss engulfs all hit, then all the casts in both my (though I concede the crit value looks unreliable, something tells me I need to test partially resisted crits. I suspect the level based act differently to the resistance based) and Nal's tests, should have had a 100% crit rate.

It's an easy test to replicate. Low level caster with an ony buff, one mob, 10 levels higher. If the crit rate is capped due to the one roll miss/hit/crit table, you shouldn't be able to land a "hit".

Desemus · 12/22/06, 12:08 PM

Loatheb would be the best place to test the 1 or 2 roll system, or would unavoidable partials distort that?

snape · 12/22/06, 12:31 PM

Originally Posted by Kink

Originally Posted by ziggy

Originally Posted by snape

I disagree with the supposed facts of this sample. At level 20, Priest Smite does between 54-62 Holy Damage. This is what you notice in the 3 Smite hits presented here - except for the 65, but it's close enough to be believable. The partially resisted crit, however, would only have amounted to a total of 70 Holy Damage. Smite crits are 1.5x, so the base would be 46 Holy Damage, which is well outside the spell's non-crit range. Is someone pulling someone's leg, or was the data copy-pasted shoddily?

Interesting. You're absolutely right, and I can't account for the discrepancy, I'd not looked at the damage range before now. Almost all of of the hit's are at the higher end of the damage range as well. I don't recall any buffs/debuffs that would have had that effect.

I'll dig around for some screenshot's later on, as well as the uneditted log.

I can offer a theory.

The crit multiplier was only applied to the NON resisted portion.

If a spell hits for 100 flat out damage, and crits for 150 and you get a 75% resisted crit you would therefore see:

Smite crits for 37.5 (75 resisted).

Totalling them gives 112.5 which makes no sense. But divide the crit portion by 1.5 and only ADD the non crit(i.e. resisted portion) and it comes out to 100.

So looking at the values your friend saw:
Sallurion's Smite crits Young Panther for 23 Holy damage. (47 resisted)

Therefore the full damage would be:

Full damage = (23/1.5)+47 = 62.333

Now factor in a couple points of +damage on his gear and the affect of that +damage with smites casting speed at lvl 20 (no clue about cast time and +damage and their interaction =) and it becomes definately possible.

But this IS guess work, it just looked logical =)

This is a nice try, but sadly it's just not right. :)

My partially resisted crits always add up to a what a real crit would have done (based on months and months of partially resisted crits in Naxx).

Eylirria · 12/22/06, 12:34 PM

Originally Posted by ziggy

It's an easy test to replicate. Low level caster with an ony buff, one mob, 10 levels higher. If the crit rate is capped due to the one roll miss/hit/crit table, you shouldn't be able to land a "hit".

Very true. It's a simple test, as soon as you have the Onyxia buff down. As soon as I can get a friend to offtank some monsters for me and chain-cast for a considerable amount of time and post the results, including my own level, monster level, tooltip crit-rate, and from there we'll know the expected results and what actually happens and try and figure it out.

I guess the reason myself and most people always took 2-rolls with miss/hit/crit all being on the same roll was because of a blue post that refered to +spellhit and making direct mention of 2 rolls, but thinking more about it, I can't recall this post mentioning crit being on the same roll, which could be intentional, or just omition since the goal of that post was to clarify on the +spell hit mechanic.

On my way to Stormwind to park a level 6 mage and camp the ony buff.

Fres · 12/22/06, 12:43 PM

Originally Posted by Eylirria

somewhat related to the old spell system where you actually had to "skill up" your magical schools (ie, @level 60, you have 300 Fire spells

This is definitely related to whatever the old system was. I distinctly remember going into SFK as a lowbie after release, and getting that curse from the ghost guardsmen that randomly applied a -100 skill modified to all skills. Sometimes you would get -100 daggers, sometimes -100 Frost or Shadow. Obviously this isn't a visible part of the system now, but this is definitely how they originally envisioned it.

Kavan · 12/22/06, 1:27 PM

Originally Posted by snape

This is a nice try, but sadly it's just not right. :)

My partially resisted crits always add up to a what a real crit would have done (based on months and months of partially resisted crits in Naxx).

Wrong, resist part in combat log is always displayed before target multipliers and before crit multiplier.

Eylirria · 12/22/06, 1:36 PM

Originally Posted by Kavan

Originally Posted by snape

This is a nice try, but sadly it's just not right. :)

My partially resisted crits always add up to a what a real crit would have done (based on months and months of partially resisted crits in Naxx).

Wrong, resist part in combat log is always displayed before target multipliers and before crit multiplier.

Let me just make sure I understand this 100%, as despite a lot of raiding done with my mage, I never really sat down to work this out.

Assume that a non-resisted fireball would've crit for 3000 (2000 hit)
Let's assume it was a 50% resist roll.

Your Fireball crits xxxx for 1500 (1000 resisted) [?]

#1 - It would've been a 2000hit (no crit multiplier)
#2 - Remove 50% of that -- Resist Roll.
#3 - Resisted amount displayed = 1000
#4 - 1000 of the hit wasnt resisted, and gets applied the crit modifier, thus 1500 crit.

Is this accurate?

Kavan · 12/22/06, 1:39 PM

Originally Posted by Eylirria

Is this accurate?

Correct.

Evalara · 12/22/06, 1:43 PM

Originally Posted by Kavan

Originally Posted by Eylirria

Is this accurate?

Correct.

Pretty sure I've seen quite a few
Your Fireball crits xxxx for 1500 (1500 resisted)

which would contradict that.

Liar · 12/22/06, 2:04 PM

Originally Posted by Evalara

Originally Posted by Kavan

Originally Posted by Eylirria

Is this accurate?

Correct.

Pretty sure I've seen quite a few
Your Fireball crits xxxx for 1500 (1500 resisted)

which would contradict that.

PvP or PvE?

EDIT: Nvm me; even I remember getting partially resisted crits in MC back in the day with fire.

◊ manly · 12/22/06, 2:27 PM

Well, not to derail things, but I see things apparently very differently from most mages.

My short short version of how to quantify the 'sum' of your +dmg/crit/hit together is to think of it as the volume of a prism. The length is your +dmg, the height is your crit, the depth being your hit. The more volume your prism has, the better your dps is. As such, you don't want to have only one stat and none of the others. Also, this 'model' implies that if you have, for example, fungal bloom debuff, you want to push as much +dmg as you can (ie: remove gear you put as crit and favor +dmg gear piece), since the resulting prism would be larger.

You could make a point and say that, for a fire mage, the height parameter of that prism should be your crit x 2.1, but I wouldn't quite model it that way, because there are some values that would be hard to quantify.

One of them, for example, is the value of hit, which oddly many of you seem to value quite differently than the way I do. Obviously, we assume that we are not hit capped, and comparing which gear to swap-in. Clearcasting only procs when you *hit* a mob. This means if you have no hit gear, and fight a lvl 63 mob, you pass up a lot of opportunities for proccing clearcasting. Additionally, if you specced the talent 'arcane potency', you *want* to max your hit, because you get more chance to proc clearcast as well as proccing arcane potency.

As a rule of thumb, most people say that a frost mage should view hit=crit. I never agreed with this view, mostly because of the way clearcasting works.

Additionally, while situational, I don't think -resistance should be considered useless for pve. I do agree that for a very large part curse of elements will deal with almost all the resistance a mob will have (ignoring 'innate resists', if such thing exists). However, in cases where I know mobs will not have curse of elements on them (ie: trash mobs, or boss with adds, like noth), this is where I definately value putting some -resistance on.

Merlyn · 12/22/06, 2:58 PM

Interesting view there, Manly. It makes total sense to me, as does your assigned value to penetration in certain situations. The stat is not useless, however much hate it gets. If penetration is useless, then so is resistance (I'm talking PvP here!). A buff that increases all resistances by 20, a set piece that increases pet resistance by 60, etc. Why are these bonuses put into the game? To hinder caster damage, and our counter is penetration. Hit is a different mechanic all together.

However I'd like to ask this question of the community here: What, if anything, would be the relationship between a binary spell like Polymorph or Frostbolt, and a certain amount of penetration when compared to +1 hit? More likely, as I'm still a baby theorycrafter (I'm trying!), how would I go about testing this mechanic for myself?

Kavan · 12/22/06, 3:06 PM

normal hit 86
25% resist => 64 hit, 22 resisted
crit => 64*1.75=112 crit, 22 resisted (could have been 87 hit, not sure with the rounding going on)

Rank 1 Missiles against level 61 PVE target.

Until I see a screenshot to the contrary I'll believe what I posted previously.

Eylirria · 12/22/06, 3:13 PM

Originally Posted by Merlyn

Interesting view there, Manly. It makes total sense to me, as does your assigned value to penetration in certain situations. The stat is not useless, however much hate it gets. If penetration is useless, then so is resistance (I'm talking PvP here!). A buff that increases all resistances by 20, a set piece that increases pet resistance by 60, etc. Why are these bonuses put into the game? To hinder caster damage, and our counter is penetration. Hit is a different mechanic all together.

However I'd like to ask this question of the community here: What, if anything, would be the relationship between a binary spell like Polymorph or Frostbolt, and a certain amount of penetration when compared to +1 hit? More likely, as I'm still a baby theorycrafter (I'm trying!), how would I go about testing this mechanic for myself?

Think about it like this. When it comes to binary spells, its all or nothing, therefore, you can think of the Resistance score as adding you a chance to miss your spell, if you'd like.

Disregarding base chance to miss, so that we only take into account the benefit from -resists on binary spells (This is assuming your target *HAS* resists to begin with, otherwise useless)

Let's take for instance, Frostbolt.

 
          x              75
 ------------------   *         =    1
 (target_level*5)

X is the absolute power of your penetration gear
75 is the maximum resists acquired through gear
1 is to signal that you're looking for 1% chance

Against a level 60 target, it would look like this:

  
        x      75   
      ---- *        =   1
      300


x = 4

-4 penetration = +1% hit vs a level 60 target, using binary spells, and assuming that the target *HAS* resistances.

Other theorycrafters or just more aware people feel free to point out any mistakes on my math, as I had never done any calculation like this before.

mylek · 12/22/06, 3:14 PM

On the subject of one/two rolls I did another quick test as follows:

L60 mage vs. L70 Unliving Resident (deadwind pass) 16% base hit rate

0% +hit, 18.05% crit (3.05%int, 2% trinket, 6% crit mass, 4% incin, 3% pyromaniac)

Following the two theories either 100% or 18.05% of non resisted scorches will crit. I stopped testing after my first two hits were regular hits.

For the one roll theory to hold true it should be possible to create a scenario where all of your unresisted spells are crits such as this one but it is just not happening. I have yet to see any kind of spell test data showing such an occurrence.

Eylirria · 12/22/06, 3:24 PM

Originally Posted by mylek

On the subject of one/two rolls I did another quick test as follows:

L60 mage vs. L70 Unliving Resident (deadwind pass) 16% base hit rate

0% +hit, 18.05% crit (3.05%int, 2% trinket, 6% crit mass, 4% incin, 3% pyromaniac)

Following the two theories either 100% or 18.05% of non resisted scorches will crit. I stopped testing after my first two hits were regular hits.

For the one roll theory to hold true it should be possible to create a scenario where all of your unresisted spells are crits such as this one but it is just not happening. I have yet to see any kind of spell test data showing such an occurrence.

I just looked through that "Backstab: 2 roll" thread and I read your post about how you were able to get frostbolt hits vs a target with 219 frost resist, and while that missrate was artificially boosted, considering what we know about binary spells, it showed already that a one-roll theory for spells and hit/miss/crit was just not valid for binary spells.

Learning something new everyday.

edited: I'll repost my thoughts on spells and roll-systems when I have something to say that doesnt rely solely on "belief" and "faith"

Aphyrax · 12/22/06, 3:29 PM

You are assuming that the to hit converts misses to hits and that one of those converted spells can never crit. Which, as others in this thread have pointed out, is a contentious issue.

So the accurate statement is, for frost hit is AT LEAST AS GOOD as crit.

Originally Posted by Harem

Also, for frost, hit = crit. Theres no such thing as if you have more, the less each one is worth. For example, lets go with 30% crit, and only 5% chance to miss. Let's say each frostbolt does 1000 damage.

Base damage on a 100 frostbolts
60000 (30% crits)
65000 (65% hits)
0 (5% misses)

delta by adding 1% crit
62000 (31% crits)
64000 (64% hits)
0 (5% misses)

+1000 over base

delta by adding 1% hit
60000 (30% crits)
66000 (66% hits)
0 (4% misses)

+1000 over base

For fire, it's a different story, since with ignite your crits hit for 210%, and MoE is nice, so crit > hit for fire.

Setia · 12/22/06, 6:25 PM

Given two rolls, would that math be correct?

---------

damage_from_crit = average_noncrit_damage*0,01*crit_bonus*hit_chance

damage_from_hit = average_noncrit_damage*(1+(crit_chance*crit_bonus))*0,01

Thus,

crit/hit = (crit_bonus*hit_chance) / (1+(crit_chance*crit_bonus))

Going further, we find that, for a full fire build,

crit/hit >= 1
when
10/11 + crit_chance <= hit_chance.

Since hit_chance is capped at 0,99, whenever crit_chance is greater than 0,99 - 10/11 = 8,09%, it can be guaranteed that hit > crit, when hit is not capped.

Given that the talents alone give most fire mages 9% crit... well... yeah. Hit>crit. Although, with MoE, Pyromaniac and Elemental Precision, one crit is also worth between around 1,71*hit_chance mp5 when spamming fireball 12.

Similarly, whenever hit_chance is lower than 90,91%, it can be guaranteed that hit > crit, even if you have 0% crit, which would be impossible given the base value.

For arcane/fire, crit/hit >=1 when 1/1,45 + crit_chance <= hit_chance. Hit > crit for sure whenever you have over 30% crit, or under 69% hit.

For full frost frost, crit/hit >=1 when 1+crit_chance <= hit_chance, which cannot happen.

For arcane/frost, crit/hit >=1 when 1/1,25 + crit_chance <= hit_chance. Hit > crit for sure whenever you have over 19% crit, or under 80% hit.

I just did some Excel charts to illustrate, but I'll upload them later.

------

Oh, and is that formula correct?

average_noncrit__nonresist_damage = ((base_damage + bonus_damage*empowered_modifier*cast_time_modifier)*self_modifier)*debuff_modifier

Where self_modifier includes Fire Power, the Undead Cleansing set bonus and similar bonuses, while debuff_modifier is for Curse of Elements, Improved Scorch, etc.

◊ Bekah · 12/22/06, 9:51 PM

Originally Posted by Liar

I haven't been following Mage mechanics lately, but I remember reading this:

Source

Is this already outdated? Granted, he is talking about physical +Hit/Crit and not Spell Hit/Crit but I'd be surprised if things would be different for them.

Especially this part is interesting:

Originally Posted by Aeus

The way WoW calculates crit rate is over ALL attacks. Crit rate is not based on hits only. In other words, if you have a 5% crit rate, that 5% chance includes misses.

So if this is true, I'd cap my Hit first before stacking Crit. Even if that was not true, I'd go for Hit anyway. Does it really matter if your Frost Nova crits? No. Does it matter if your Frost Nova got resisted? Oh yes! On top of that CS can't even crit, but I am sure you always want CS to land regardless.

Remember: +Hit is affecting all spells minus self buffs. +Crit is affecting all your damaging abililties only.

(edited the longer quote out so this wouldnt' be so long.)

This relates to the fact that folks will also bend their views towards their activities =P If all you ever do is hit level 63 boss mobs, then Hit has some very definite sway. If you're spending your time half and half (and making gearing concessions for, no matter how small) PvPing, on the other hand, the level 60 player/mob hit cap impacts you a little more strongly than your pure-raiding friend. Neither way is inherently better- just more suited to the task at hand.

Same reason why I'm inherently pre-disposed towards +hit on all of my characters. I'm pretty much a pure raider and my shadow priest has absolutely no use for crit- it's all about getting to the hit cap asap. Of course that's also why I'm in this thread- I'm never quite sure how I should balance crit/hit on my mage.

Setia · 12/22/06, 10:17 PM

Here are the charts based on my math.

In the red zone, 1%crit is better. Grey zone, 1% hit is better.

Andrise · 12/23/06, 4:10 AM

Comparing 1% hit to 1% crit is somewhat of a false comparison, because of how Blizzard budgets the two stats. Basically, one rating point of anything is equivalent value, and it only takes 8 hit rating to equal 1%, while it takes 14 crit rating to get 1% crit (both at 60, obviously). So really, 1% hit should be compared to .57% crit if you're comparing equivalent level items. So it would appear that hit gives you "more bang for the buck" item wise, since its only slightly behind the dps gain % for %, and you can get hit a lot more cheaply. However, hit has limited value in PvP, where you only need 3 or 4%, and vs trash mobs, etc. So there's always a tradeoff between the two, and it's not really viable to build around just one or the other, and the choice of items for each slot is generally pretty apparent, at least with current itemization. This may change if there is more varied itemization at each plateau of ilvl in the expansion, however right now there really aren't any slots for a mage where there aren't clear "best items," some being items with hit, some being items with crit.

CHeeSY-CrAfT · 12/24/06, 5:00 AM

Prior to the 2.0 patch, that was a given. But I think even now there is a bit of a discrepancy in terms of what items makeup the best for each slot. Ignoring any of the "raid DPS" boost that 6piece frost fire gave, I believe this setup was a given as the "optimal" for a mage when we had 6% hit on precision:

http://ctprofiles.net/4445087

There were not really choices to make; even if you felt that the %hit on Dark Storm Gauntlets was useless, there wasn't much else to comparitively, save stats on Frostfire Gloves. Now, though, there is 3% hit that is wanted in that gear setup. This is basically the underlying reason for my questions as of late. Is this the best gear setup, or is more hit needed to truly arrive at the optimal setup?