I appreciate the test, Pf. It seems like there's a fairly strong case in support of a two-roll system for special attacks. When can we expect an updated spreadsheet incorporating this conclusion? ;-)
Well, if we now assume that a two-roll system does apply to special attacks, then a reevaluation of Surprise Attacks is in order. I'll compare Backstab with dodges to Backstab without dodges and to Backstab without dodges plus 8% damage.
Assume:
30% base crit
level 60 target, at least +5% hit
Backstab average hit damage = 100 (arbitrary, doesn't matter too much since I'm looking for the percentage of change)
5/5 Lethality
Without dodges, plus 8% (i.e. current Surprise Attacks);
0 dodges = 0 energy, 0 damage
60 crits = 3600 energy, 14904 damage
40 hits = 2400 energy, 4320 damage
Total = 6000 energy, 19224 damage
3.20 DPE (~9% increase)
So, even under a two-roll system the original talent was mediocre garbage. After the buff, it amounts to about a 3-5% increase to your overall damage, depending on your gear level (for me it's around 5%, for a Naxx-geared rogue it would be more like 3%).
Doesn't this negate the crit cap that is normally seen when duel-wielding with a high crit rate? Miss 24%, Glance 40%, can only go up to 36% crit? If Miss isn't on the table, does this mean the actual crit cap is higher?
Also, what about block/dodge/parry, are they on the hit/crit table or the miss/hit table?
"Puns are inherently evil, in the "must do evil!" sort of way."
I appreciate the test, Pf. It seems like there's a fairly strong case in support of a two-roll system for special attacks. When can we expect an updated spreadsheet incorporating this conclusion? ;-)
I'll incorporate the change in the next release which will have the preliminary TBC data.
Originally Posted by Harem
Doesn't this negate the crit cap that is normally seen when duel-wielding with a high crit rate? Miss 24%, Glance 40%, can only go up to 36% crit? If Miss isn't on the table, does this mean the actual crit cap is higher?
Also, what about block/dodge/parry, are they on the hit/crit table or the miss/hit table?
This has nothing to do with the crit cap when dual-wielding. It has been proven that white damage follows a single roll system. Now, we know that yellow (special attacks) follow a two roll system. Thus you cannot be crit capped with yellow attacks, only white attacks.
What was your exact gear for the test? +hit and +crit specifically.
I know exactly what you are getting at, so I just repeated the test with +7% hit, 305 dagger skill, 30.34 crit, attacking a lvl 43 mob.
Anyway, 305-215=90, 90x.04 = 3.6 hit
95+7+3.6-50 = 55.6 total hitrate.
30.34+30 = 60.34 backstab critrate.
There is no room whatsoever for overlap in this test. With a single roll system like white attacks work, I would be crit capped and would either crit or miss the mob. But, instead, it only took a few tries and i had a backstab hit just like before. This test is repeatable by anyone, and in my mind, confirms the 2-roll system for special melee attacks. I see no room for doubt or error.
I did a similar test, yesterday we were running AQ to C'Thun.
I especially paid attention to the Anubisath Defenders and their Dust Cloud ability which reduces chance to hit by 85%.
Normally I just stop attacking with specials while under this effect, but now I performed as much backstabs as I could, and I also noticed several hits while under the effect.
Now, it really seems to be a two roll system for special attacks, although I don't like that at all.
It means with 30% crit rate I am losing 60*0.05 = 3% of my actual crit rate for special attacks, and normal hits gain these 3%.
Though that should explain why most times I had noticed a lower crit rate than expected on my backstabs, which always bugged me.
11/23 21:11:35.796 You are afflicted by Dust Cloud.
11/23 21:11:36.796 You miss Anubisath Warder.
11/23 21:11:37.328 Your Backstab missed Anubisath Warder.
11/23 21:11:37.562 You miss Anubisath Warder.
11/23 21:11:38.593 You miss Anubisath Warder.
11/23 21:11:38.843 Your Backstab missed Anubisath Warder.
11/23 21:11:39.281 You miss Anubisath Warder.
11/23 21:11:40.187 Your Backstab missed Anubisath Warder.
11/23 21:11:40.218 You miss Anubisath Warder.
11/23 21:11:41.046 You attack. Anubisath Warder dodges.
11/23 21:11:42.015 You miss Anubisath Warder.
11/23 21:11:42.250 Your Backstab missed Anubisath Warder.
11/23 21:11:43.000 You miss Anubisath Warder.
11/23 21:11:43.406 Your Backstab missed Anubisath Warder.
11/23 21:11:43.625 You miss Anubisath Warder. 11/23 21:11:44.750 Your Backstab hits Anubisath Warder for 649.
11/23 21:11:44.750 Your Instant Poison VI hits Anubisath Warder for 122 Nature damage.
11/23 21:11:44.750 You miss Anubisath Warder.
11/23 21:11:45.359 You miss Anubisath Warder.
11/23 21:11:45.859 You are afflicted by Silence.
11/23 21:11:47.859 You miss Anubisath Warder.
11/23 21:11:47.890 You miss Anubisath Warder. 11/23 21:11:47.921 Dust Cloud fades from you.
11/23 21:11:55.046 You are afflicted by Dust Cloud. 11/23 21:11:55.187 Your Backstab hits Anubisath Warder for 685.
11/23 21:11:55.546 You miss Anubisath Warder.
11/23 21:11:55.546 You miss Anubisath Warder.
11/23 21:11:57.187 You miss Anubisath Warder.
11/23 21:11:57.375 You attack. Anubisath Warder dodges.
11/23 21:11:58.906 You miss Anubisath Warder.
11/23 21:11:59.171 You miss Anubisath Warder.
11/23 21:12:00.484 You perform Slice and Dice.
11/23 21:12:00.718 You miss Anubisath Warder.
11/23 21:12:00.734 You gain Slice and Dice.
11/23 21:12:00.765 You gain 25 Energy from Relentless Strikes Effect.
11/23 21:12:00.765 You miss Anubisath Warder.
11/23 21:12:01.765 Your Backstab missed Anubisath Warder.
11/23 21:12:01.828 You miss Anubisath Warder.
11/23 21:12:02.234 You miss Anubisath Warder.
11/23 21:12:03.125 You miss Anubisath Warder.
11/23 21:12:03.468 Your Backstab missed Anubisath Warder.
11/23 21:12:03.703 You miss Anubisath Warder.
11/23 21:12:04.531 You miss Anubisath Warder. 11/23 21:12:04.812 Your Backstab crits Anubisath Warder for 1514.
11/23 21:12:05.031 You miss Anubisath Warder.
11/23 21:12:05.828 You attack. Anubisath Warder dodges.
11/23 21:12:06.359 You miss Anubisath Warder. 11/23 21:12:07.093 Dust Cloud fades from you.
11/23 21:12:13.609 You are afflicted by Dust Cloud.
11/23 21:12:13.875 You miss Anubisath Warder.
11/23 21:12:14.578 You miss Anubisath Warder. 11/23 21:12:14.859 Your Backstab crits Anubisath Warder for 1415.
11/23 21:12:15.093 You miss Anubisath Warder.
11/23 21:12:16.171 You miss Anubisath Warder.
11/23 21:12:16.281 You miss Anubisath Warder.
11/23 21:12:17.468 You miss Anubisath Warder.
11/23 21:12:17.718 You miss Anubisath Warder.
11/23 21:12:18.828 You miss Anubisath Warder.
11/23 21:12:19.015 You miss Anubisath Warder.
11/23 21:12:20.156 You miss Anubisath Warder.
11/23 21:12:20.328 You miss Anubisath Warder.
11/23 21:12:21.562 You hit Anubisath Warder for 159. (glancing)
11/23 21:12:21.734 Your Backstab was dodged by Anubisath Warder.
11/23 21:12:21.734 Your Instant Poison VI hits Anubisath Warder for 136 Nature damage.
11/23 21:12:21.734 You miss Anubisath Warder.
11/23 21:12:22.937 You miss Anubisath Warder.
11/23 21:12:22.953 You miss Anubisath Warder. 11/23 21:12:23.390 Your Backstab hits Anubisath Warder for 622.
11/23 21:12:23.390 Your Instant Poison VI hits Anubisath Warder for 121 Nature damage.
11/23 21:12:24.250 You miss Anubisath Warder.
11/23 21:12:24.453 You miss Anubisath Warder. 11/23 21:12:25.328 Dust Cloud fades from you.
Thanks for pulling all of this data together, guys. It's neat to know that after nearly three years we're still learning things about the mechanics of the game.
Yeah, finally a proven answer to all those blocked crits.
I did some performance testing. With my old linux box I could fill up a table and pull a random result about a million times every second. With that in mind, it doesn't really matter if you are rolling 1 or 3 dices per hit. It's still thousands of times faster than your fastest database query.
So I'd say you can't argue against a two roll system with performance issues. As random number generators are O(1) and work faster than you can measure. :P
Interesting, this finding also questions whenever spells is usings 2 roll system to determine miss/hit/crit outcome.
Originally Posted by zeidrich
Women's breasts can be modeled as a cone and measured as V = (Db^2*h*.785)/3 and since breasts can be thought of as an amorphous fluid, you just have to worry about containing the volume of the breast.
Think the general sentiment is that first it does a miss/hit/crit roll, then a resist roll.
Yes, however it based on analogy with melee attacks. And theory of single miss/hit/crit roll with special melee attacks has just been disapproved.
As far as i know nobody actually perfomed any tests to prove existance of single miss/crit/hit roll for spells.
Originally Posted by zeidrich
Women's breasts can be modeled as a cone and measured as V = (Db^2*h*.785)/3 and since breasts can be thought of as an amorphous fluid, you just have to worry about containing the volume of the breast.
Thanks for pulling all of this data together, guys. It's neat to know that after nearly three years we're still learning things about the mechanics of the game.
I'd just like to add something on the same note as Wodin:
Really great job testing this out, this thread reminded me of why I started reading these forums in the first place. Has to be one of the best pieces of theorycraft that cropped up in these boards.
Think the general sentiment is that first it does a miss/hit/crit roll, then a resist roll.
Yes, however it based on analogy with melee attacks. And theory of single miss/hit/crit roll with special melee attacks has just been disapproved.
As far as i know nobody actually perfomed any tests to prove existance of single miss/crit/hit roll for spells.
As someone noted before, there was a "blue" post explaining the spell hit mechanic, and this blue post (I believe it was made by Eyonix), pretty much noted that spells worked on a 2-roll system:
Originally Posted by Eyonix
I've seen a number of threads recently on the boards pertaining to +spell hit chance and I've gathered some information for you which I hope will dispel some of the confusion. Please feel free to post your comments in a reply to this thread.
+1% spell hit chance increases the base chance for a spell to land against a target by 1%
If your target is the same level as you, a spell has a base chance to hit of 96%.
If the target is +1 level compared to you: 95%
+2 levels: 94%
+3 levels: 83% if the target is a mob, 87% if the target is a player.
+4 levels: mob: 72% player: 80%
+5 levels: mob: 61% player: 73%
Etc...
If you wear a +1% spell hit chance item, the above percentages will increase by 1. +2% gives +2.
Etc...
For binary spells only, there is an additional modifier for the resistance of the victim to your particular spell school: fire, frost, shadow, nature, arcane. That modifier is multiplied by your hit chance to get your actual chance to land. This is done with binary spells only, because they never do partial damage.
Example:
Eyonix the Mage (level 60) fires a frost bolt at Yeti of Doom (level 63). Eyonix is also wearing a total of +6% spell hit gear. Yeti of Doom has frost resistance such that he takes 50% from level 60 frost attacks. So, heres the hit calculation:
The game will roll a number between 0 and 1, and if its less than 0.445, the frost bolt will hit for full damage. Otherwise, a resist message will appear.
2nd Example:
After the resist, Eyonix decides to fire a fireball at Yeti of Doom. Eyonix still has +6% spell hit. Fireball is not a binary spell. Heres the calculation:
0.83+0.06= 0.89
The game will roll a number between 0 and 1, and if its less than 0.89, the fireball will hit. Otherwise, a resist message will appear. After the fireball lands, the game will then apply spell resistance to determine a partial resist, if any. Assuming the yeti also has 50% fire resistance, on average, 50% of the damage will be resisted.
1st Roll: Did it hit or miss?
2nd Roll: Damage loss through resistance (25%/50%/75%)
It didnt, however, make any mention of where it would be determined that the spell critically hit, and most of us, if not all, just assumed that the Crit factor had to be on the first roll.
Also, something else needs to be pointed out about spellcasting and rolling. This blue thread mentions binary spells being on a single roll with hit/miss/resistance all combined.
edit: Included the original Eyonix explanation of spellhit.
As someone noted before, there was a "blue" post explaining the spell hit mechanic, and this blue post (I believe it was made by Eyonix), pretty much noted that spells worked on a 2-roll system:
I perfectly aware of that post. If you read my post carefully i was questioning an existence of single roll for miss/hit/crit outcome.
Originally Posted by zeidrich
Women's breasts can be modeled as a cone and measured as V = (Db^2*h*.785)/3 and since breasts can be thought of as an amorphous fluid, you just have to worry about containing the volume of the breast.
Surely this is the sort of thing you don't need to test on Live? Assuming that bedrock mechanics like these are unlikely to change, you can do it on a private server. Spawn an unkillable mob (Servant of Allistarj for example), and have a mage toon nuke it non-stop. Heck, you could wrap the client inside a Glider bot and just leave the thing unattended for a day to get some really high number statistics. Then repeat when afflicted by some debuff that reduced spellhit (which on a private server you can do automatically, without having to go looking for mobs that cast one, yes/no?), and see what the results say.
Since there *are* private servers out there, I honestly cannot understand why something this basic wasn't worked out to three decimal places years ago.
Same applies to aggro - i.e. working out the precise percentage of Broodlord's knockback. Surely, if you have access to a server, you have the control to be able to test rigorously what effect his knockback has? OK, you'd only know it for that build, so it depends on whether it's changed since then - no point looking at the effects of +skill for example, because we know that's changed. But fundamental stuff like whether it's 1 or 2 rolls per spell cast, or the exact numbers to pull aggro from melee range, or 30 yards, and the coefficients for fade/feint etc. should surely be fairly trivial to find out, if you have access to both client and server. And we do - or at least *someone* out there does!
Women's breasts can be modeled as a cone and measured as V = (Db^2*h*.785)/3 and since breasts can be thought of as an amorphous fluid, you just have to worry about containing the volume of the breast.
Surely this is the sort of thing you don't need to test on Live? Assuming that bedrock mechanics like these are unlikely to change, you can do it on a private server. Spawn an unkillable mob (Servant of Allistarj for example), and have a mage toon nuke it non-stop. Heck, you could wrap the client inside a Glider bot and just leave the thing unattended for a day to get some really high number statistics. Then repeat when afflicted by some debuff that reduced spellhit (which on a private server you can do automatically, without having to go looking for mobs that cast one, yes/no?), and see what the results say.
Since there *are* private servers out there, I honestly cannot understand why something this basic wasn't worked out to three decimal places years ago.
Same applies to aggro - i.e. working out the precise percentage of Broodlord's knockback. Surely, if you have access to a server, you have the control to be able to test rigorously what effect his knockback has? OK, you'd only know it for that build, so it depends on whether it's changed since then - no point looking at the effects of +skill for example, because we know that's changed. But fundamental stuff like whether it's 1 or 2 rolls per spell cast, or the exact numbers to pull aggro from melee range, or 30 yards, and the coefficients for fade/feint etc. should surely be fairly trivial to find out, if you have access to both client and server. And we do - or at least *someone* out there does!
Private servers are reimplementations of wow based on how we think it works. Since none of blizzard's code/algorithms are on private servers, these tests would be useless.
"Puns are inherently evil, in the "must do evil!" sort of way."
Yes.
Calculations are made serverside aren't they? And unless somebody gets their hands on an actual blizzard server, all emulated servers will just be guessing.
Besides the legal issues of course.
Private servers are reimplementations of wow based on how we think it works. Since none of blizzard's code/algorithms are on private servers, these tests would be useless.
Hmm, my mistake - I thought they were based on leaks of the server code.
I ran a simple test with binary magic spells. For this test to prove anything one assumes that a binary spell against a target with high resists is the same thing as having a very high normal miss rate.
Naked mage[60]
6% hit
53% crit (minimum. target always rooted with shatter for +50%. due to winter's chill actual crit is some times higher)
In the expected one roll calculation with 47.25% of my frostbolts hitting and a crit rate of 53% they should all land as crits.
Test data, all frostbolts (27):
Hit: 5 (18.5%)
Crit: 8 (29.6%)
Resist: 14 (51.9%)
Regular hits appearing at all shows that the expected one roll behavior is not occurring in this case. One could repeat this experiment against level +6 mobs (expected 50% miss rate) for definitive proof.
Regular hits appearing at all shows that the expected one roll behavior is not occurring in this case. One could repeat this experiment against level +6 mobs (expected 50% miss rate) for definitive proof.
More controllable than finding +6 mobs would be to get a level 54 mage (or even lower) to go up against the same 60 priest. Got any slightly lower level mage friends?
Looking over the information, it seems that the primary basis for the 2 roll argument on special attacks is that things still occur that should not occur given base math in a single roll system. This is not necessarily indicative of a 2 roll system, however. We could take a given example already known in the game to show that the base math need not apply to more extreme situations: defense vs crits.
Given a warrior with a high enough defense that a normal mob (something without an augmented crit %) would have a 0% chance to crit him, we know this is not the case; you can always be crit. This leads to 2 possible conclusions: minimum hard cap, or diminishing returns. Unfortunately I don't know how exactly the numbers run once you get past a certain point of low crit %. But, it is not unreasonable to say that there could be minimum caps or diminishing returns on other forms, such as chance to normal hit. All other possible outcomes could simply truncate at 95%, 96%, whatever, and a minimum x% is reserved for normal hits.
11/22/06, 10:51 PM
Similar Threads
