[Theorycraft] Calculating frostbolt damage exactly

Re: [Theorycraft] Calculating frostbolt damage exactly

by pitfighter » Tue Mar 22, 2016 6:45 pm

Puny wrote:I've made a similar thing:
http://torde.gillamaten.se/wow/

My guideline on spell penetration has been this formula:
Code: Select all
avgresist = (target_resistance - min(target_resistance, spell_penetration) + 15) / 300 * 0.75


Where 15 is the level gap resistance which cannot be penetrated. I've seen some of you mention that this number should be 24, which one is correct?

Also, by this measurement, each 10 points of spell penetration isnt a flat 2.5% increase, which kind of makes sense. The more resistance you have, the more effective it gets. 200 resistance ~= 50% damage reduced; 300 resistance ~= 75% damage reduced, meaning that the last 100 resistance has doubled the effect.

Also, thanks for the info on slowing effects having a 5% penalty, I had no idea about that.


Two thumbs up on the calculator Puny!
pitfighter
Sergeant
Sergeant
 

Re: [Theorycraft] Calculating frostbolt damage exactly

by sirael7 » Wed Mar 23, 2016 11:07 am

I think there is a mistake in your frostbolt calculations puny.

The crit chance shown after calculating assumes that you have a 100% hit chance. You have to multiply that value with (1- chance to miss) to get the correct crit chance. Or you could note that the given crit chance is the crit chance of the spells that hit your target not of the number you cast.

That should also reduce the value of crit in relation to spelldamage. I wrote my own calculator some time ago and it gets the exact same value for Hit as yours but for Crit i get a lower one.

But maybe my calculation is wrong. I found different information on that topic. It comes down to if Crit is calculated before of after Hit chance.

Edit: Nevermind, i found out where the difference in our calculations comes from. I average the crit value over the next 10% of Crit added and Puny takes the value of 1% more Crit. Also my point from the crit calculation above has no Impact on the Value of Crit so it doesnt really matter.
sirael7
Private
Private
 

Re: [Theorycraft] Calculating frostbolt damage exactly

by Youfie » Wed Mar 23, 2016 1:31 pm

sirael7 wrote:I found different information on that topic. It comes down to if Crit is calculated before of after Hit chance.

I'm not sure I understand.

How A*B could be different from B*A? Could you write down an example about what you mean by Crit Chance being calculated before / after Hit chance? Cause it seems to me that as long as there are two rolls, it doesn't matter which one is first, mathematically speaking, since it's a multiplication.
User avatar
Youfie
Knight-Lieutenant
Knight-Lieutenant
 

Re: [Theorycraft] Calculating frostbolt damage exactly

by sirael7 » Thu Mar 24, 2016 11:19 am

Assume we cast 100 Frostbolts, have a hit chance of 90% and a crit chance of 20%.

One example would be:

20 Crits
70 Hits
10 Miss

Another Version could be:

10 Miss
18 Crit
72 Hit

In the first example you crit 20% of all casts but misses can't be crits. In the second example a critcal hit can also miss, resulting in less total crits (and less damage).

I found another thread on this forum discussing that issue. Everyone in that thread was sure the second example is correct. The last poster though said that everything in that thread is wrong and linked to a much longer thread on elitistjerks from 2006 in which people came to the conclusion that the first example was correct back then.

So the question is if crit applies to the number of spells you cast or to the number of spells that hit your target. In the first example crit would be more valuable than in the second.
sirael7
Private
Private
 

Re: [Theorycraft] Calculating frostbolt damage exactly

by Puny » Thu Mar 24, 2016 11:26 am

Youfie wrote:From what I remember from Nostalrius Beta and all, when Devs discussed fairly often with players the scripts and sources to be used, 24 innate resistance was the value used, i.e. 2% average mitigation per level, 6% versus a Boss. This value scales in a linear way, just like resistances in general do, and a level 30 player attacking a 33 Monster would get 12 non-removable / innate / level-based / whatever you call it Resistances, resulting in 6% overall mitigation aswell, and so on.


You're correct about the average mitigation scaling linearly with resistance but that's not what I was trying to say.

I'll try to clarify:
10 points of spell penetration decreses the mitigation by 2.5 percentage points. Let's say the boss target has 60 resistance after curse and talents, resulting in 21% average mitigation, or 79% of damage coming through. Increase with 10 spell penetration => 18.5% average mitigation, or 81.5% of full damage is dealt. The total increase is 81.5 / 79 = 1.0316 (3.16% dps increase) whereas the last 10 effective spell penetration is a 2.73% dps increase.

All I wanted to say is that it's faulty to assume that 10 spell penetration increases your dps by 2.5%.
Puny
Tester
 

Re: [Theorycraft] Calculating frostbolt damage exactly

by sheepstickss » Thu Mar 24, 2016 4:25 pm

sirael7 wrote:Assume we cast 100 Frostbolts, have a hit chance of 90% and a crit chance of 20%.

One example would be:

20 Crits
70 Hits
10 Miss

Another Version could be:

10 Miss
18 Crit
72 Hit

In the first example you crit 20% of all casts but misses can't be crits. In the second example a critcal hit can also miss, resulting in less total crits (and less damage).

I found another thread on this forum discussing that issue. Everyone in that thread was sure the second example is correct. The last poster though said that everything in that thread is wrong and linked to a much longer thread on elitistjerks from 2006 in which people came to the conclusion that the first example was correct back then.

So the question is if crit applies to the number of spells you cast or to the number of spells that hit your target. In the first example crit would be more valuable than in the second.


2 table roll.

If you were to read that thread then that they actually disproved it (example 1). You can test it by shooting 1000 frosbolts with 16% hit vs 0% hit and recording crit %. OR you can also test it by having a lvl 30 shoot frostbolts at a lvl 50 or something, and seeing how many of the hits that land crit. With 20% crit and 80% miss rate -> they should have 100% crit chance on all the ones that hit.

Here's my guide from guild forums (posted it previously here but it needed to be changed)
-----------------------------------------------------------------------------------------------------------------
How to calculate your CRIT value:

Your frost bolt does 456 avg dmg. (not the 446 its supposed to be, don't know why)
A solid preraid set of gear gives you spellpower of 400. 2 crit.
Full Raid consumes of +86 spellpower. (36 from Brilliant Wizard Oil and 35 Greater Arcane Elixir and 15 Elixir of Frost Power)
Frostbolt's spellpower coefficient is .814 (3/3.5 = .857 then .857*.95 = .814)
3/3.5 = cast time
95% -> 5% reduction is because it has a slowing effect.

456 + (.814 coefficient * 486) = 852 dmg per frostbolt.
1% crit would over a long average be equal to 1% more damage per frost bolt. 852*.01 = 8.52 dmg.
8.52 / (spell coefficent = .814) = 10.46 Spellpower.
Then we divide it by current crit chance to account for diminishing returns **
10.46 / (1+ [crit chance = .02]) = 10.26 Spellpower.

At this gear level 10.26 spellpower would be equal to 1% crit, and that would be the ratio you tradeoff your items for max theoretical DPS.

Now lets say you have better gear at a level of 450 Spellpower. +86 from Raid consumables.
1% crit = 10.96 SP

TLDR: Crit-value = FB-Dmg/ 100 / .814 / (1 + crit chance)

** Each additional crit has diminishing returns. All of your crit affects your spellpower, so when you add +1% crit, you aren't comparing it to your frostbolt base dmg, but to (frosbolt dmg * current crit chance). So when going from 0-1% crit you do (frostbolt dmg * 1.00). However when going from 99%->100% crit you're comparing it to (frostbolt dmg * 1.99).
-----------------------------------------------------------------------------------------------------
How to value HIT:
You can't go above 16% hit (you get 6% from talents). Base hit chance is 83%. 1% miss is unremovable.

Interestingly, spells are a 2 table roll. First the server rolls to see if you hit, then if you successfully landed your spell; it checks if you made a crit. When calculating hit value, an additional hit can crit and we take that into consideration. Like crit, additional hit makes your spell-power more effective and increasing hit has diminishing returns (due to spell-power becoming worth more).
Crit is not used during hit calculations because it affects the hit and spell-power evenly.

Gear: 400 SP +86 (from consumables) = 486. 0 hit.
852 dmg per frostbolt.
Dmg per frostbolt / Spell coefficient / (.83 + current-hit + .01)
[8.52/.814 = 10.46 dmg] / (.83) = 12.6 sp

At this gear level 12.6 spellpower would be equal to 1% hit, and that would be the ratio you tradeoff your items for max theoretical DPS.

Lets say you have 2% hit?
[8.52/.814 = 10.46 dmg] / (.85) = 12.31 sp

TLDR: Hit-value = FB-Dmg/ 100 / .814 / (hit chance)

The more spellpower you have, the more each crit and hit is worth.
--------------------------------------------------------------------------------------------------------
So you make tradeoffs on your items until you find your max DPS combination when choosing your items.

Good luck!
----------------------------
You can check your gear values if feeling lazy (Save it to drive) + can also verify my math here: https://docs.google.com/spreadsheets/d/1HjDEIyzXBwmH9MQRgQErTuwfJ0plmpy-ZVESJ-75HIY/edit?usp=sharing
sheepstickss
Private
Private
 

Re: [Theorycraft] Calculating frostbolt damage exactly

by Impsie » Sun Mar 27, 2016 4:06 am

Spells are on a two-roll system. It has to Hit first in order to Crit. This has been proven for years and does not need to be debated further.
Impsie
Sergeant
Sergeant
 

Previous

Return to Mage

cron