trying to predict spell damage

LTraveller - Dreamweaver · March 2010

Anyone knows how spell damage is calculated?
I'm trying to figure that out; so far, unsuccessfully.

Most spells are "base magic attack + X% of weapon damage + const".
What is "base magic attack"? -- apparently, a range which can be seen in character screen. This one is predictable:
( 1 + ( MAG / 100 ) ) * ( LVL + EQP )
EQP = sum of all +matk in your gear (usually weapon + rings)
Source: http://pwi-forum.perfectworld.com/showthread.php?t=214921
Tested this formula, and it seems correct. And if you unequip the weapon, min and max of the range will become equal, and spells will do exactly the same amount of damage on same mobs.

The damage is reduced by mob's defense; I'm assuming
actual_damage = coeff * predicted_damage; coeff = actual_damage / predicted_damage
coeff depends on mob's def, and probably caster's level and mob's level.

For non-elemental mobs that coeff should be the same for all elemental spells; checked if that's the case by unequipping weapon and rings on my veno (base matk: 229; weapon damage should be 0) and casing some spells on backstabbing scorpions:
coeff = 2856/(229+2716.1) = 1191/(229+966) = 1636/(229+1327.7) = 1521/(229+1270.8 ) = 218/229
coeff = 0.9697 = 0.9966 = 1.050 = 1.014 = 0.9519
(spells: lucky, venomous, ironwood, nox, blazing)
No, coeff is not the same; the difference is almost 10%. Note that the highest coeff corresponds to ironwood scarab, the spell with 300% weapon damage, and lowest -- to blazing's non-DoT part, which doesn't depend on weapon damage at all. Maybe it's possible to solve that as equation for weapon damage?

coeff = 2856/(229+0.85*wdmg+2716) = 1191/(229+wdmg+966) = 1636/(229+3*wdmg+1327.7) = 1521/(229+1.8*wdmg+1270.8) = 218/229
(229+0.85*wdmg+2716)/2856 = 229/218; wdmg = (2856*229/218-229-2716)/0.85 = 64.8
(229+wdmg+966)/1191 = 229/218; wdmg = (1191*229/218-229-966) = 56.1
(229+3*wdmg+1327.7)/1636 = 229/218; wdmg = (1636*229/218-229-1327.7)/3 = 54
(229+1.8*wdmg+1270.8)/1521 = 229/218; wdmg = (1521*229/218-229-1270.8)/1.8 = 54.4

After equipping rings (166 matk; base becomes 979, since rings are amplified by high mag):
3708/(979+2716.1+0.85*wdmg) = 2066/(979+966+1*wdmg) = 2827/(979+1327.7+3*wdmg) = 2524/(979+1270.8+1.8*wdmg) = 934/979
(979+2716.1+0.85*wdmg)/3708 = 979/934; wdmg = (3708*979/934-979-2716.1)/0.85 = 225.35
(979+966+1*wdmg)/2066 = 979/934; wdmg = 2066*979/934-979-966 = 220.53
(979+1327.7+3*wdmg)/2827 = 979/934; wdmg = (2827*979/934-979-1327.7)/3 = 218.83
(979+1270.8+1.8*wdmg)/2524 = 979/934; wdmg = (2524*979/934-979-1270.8)/1.8 = 219.89

And with green FC wand 839-856 [+25] (avg 872.5, min 864, max 881), rings 166;
coeff = 8128/(4896.5+0.85*wdmg+2716) = 6604/(4896.5+wdmg+966) = 9041/(4896.5+3*wdmg+1327.7) = 7729/(4896.5+1.8*wdmg+1270.8) = 4714/4896.5
(4896.5+0.85*wdmg+2716)/8128 = 4896.5/4714; wdmg = (8128*4896.5/4714-4896.5-2716)/0.85 = 976
(4896.5+wdmg+966)/6604 = 4896.5/4714; wdmg = (6604*4896.5/4714-4896.5-966) = 997
(4896.5+3*wdmg+1327.7)/9041 = 4896.5/4714; wdmg = (9041*4896.5/4714-4896.5-1327.7)/3 = 1055
(4896.5+1.8*wdmg+1270.8)/7729 = 4896.5/4714; wdmg = (7729*4896.5/4714-4896.5-1270.8)/1.8 = 1033

...which is actually close, but weapon damage is slightly higher than it should be. So either the structure of damage formula is different, or there is something that adds to your weapon damage even when you don't have anything equipped; + I don't know how damage reduction coefficient is calculated.

ShadowsWrath - Heavens Tear · March 2010

Wow just wow

You really over thought that .. just hit the mob already

Magicsaber - Dreamweaver · March 2010

LTraveller - Dreamweaver wrote: »

Anyone knows how spell damage is calculated?
I'm trying to figure that out; so far, unsuccessfully.

Most spells are "base magic attack + X% of weapon damage + const".
What is "base magic attack"? -- apparently, a range which can be seen in character screen. This one is predictable:
( 1 + ( MAG / 100 ) ) * ( LVL + EQP )
EQP = sum of all +matk in your gear (usually weapon + rings)
Source: http://pwi-forum.perfectworld.com/showthread.php?t=214921
...

...which is actually close, but weapon damage is slightly higher than it should be. So either the structure of damage formula is different, or there is something that adds to your weapon damage even when you don't have anything equipped; + I don't know how damage reduction coefficient is calculated.

From random source (I had seen many of them with same formula and checked it ):

damage reduction (in %) = 100*def / (def + 40*lvl)
where:
def = defense value (Edit: - of mob - magical defense against magical attack and physical defense against physical attack)
lvl = level of the character

According formula damage reduction of high-level character is less than damage reduction of low-level character.

Wahida - Harshlands · March 2010

total skill damage = [(base magic attack) + ((equipment magic attack + Character level) * (skill weapon damage % )) + skill bonus damage] * (1+masteries)

^ Should be the correct formula, but meh. Never checked it or w/e

EDIT:
But uhm, didn't calc in the mobs defences whatsoever.
So still a meh

LTraveller - Dreamweaver · March 2010

Ah, masteries. That's what I forgot.
And in damage reduction formula I wasn't sure whose lvl it is: caster's or target's.
Thank you.

Magicsaber - Dreamweaver · March 2010

LTraveller - Dreamweaver wrote: »

...
And in damage reduction formula I wasn't sure whose lvl it is: caster's or target's.

Level of the caster.

Here my post from 12-01-2009 02:54 AM about same damage reduction formula relative to character:

Magicsaber - Dreamweaver wrote: »

In addition to post about real formula above:

Damage reduction in %
__|____def (phys or mag)
lvl|1000_|3000|9000
|
30|__45_|_71_|_88
54|__32_|_58_|_81
90|__22_|_45_|_71

According formula you will get for 30 level:
Damage reduction
|_______________def (phys or mag)
|1000_________|3000_______|9000
|0,454545455__|0,714285714_|_0,882352941 or in %
|45,4545455___|71,4285714__|__88,2352941.

You can check it at http://pwcalc.ru/pwi/.

Kaneharo - Dreamweaver · March 2010

Um, wow. It's amazing what happens when people give thought to something that isn't usually thought of. Not trying to sound offensive, I'm actually amazed.

Asterelle - Sanctuary_1381265973 · March 2010

I did a knowledge dump a while ago on the wiki about this
http://pwi-wiki.perfectworld.com/index.php/Damage

It's not perfect but since it's a wiki feel free to improve it.

One thing I wasn't sure of was how magic mastery stacks with damage amplification...

20% more damage from mastery + 20% amplify damage is 40% more damage or 44%?

LTraveller - Dreamweaver · March 2010

Did more testing on same scorps, without weapon and rings, assuming:
damage = reduction * mastery * (base+(eq_atk+lvl)*%)+const)
reduction = 1-863/(863+85*40)
mastery = 1.2 (level 10)
base = 229
eq_atk = 0
lucky lv7: (1-863/(863+85*40))*1.2*(229+85*0.85+2716.1) = 2887; observed 2856; 1% diff
venomous lv10: (1-863/(863+85*40))*1.2*(229+85*1+966) = 1225; observed 1191; 2.7% diff
ironwood lv10: (1-863/(863+85*40))*1.2*(229+85*3+1327.7) = 1734; observed 1636; 5.6% diff
nox lv8: (1-863/(863+85*40))*1.2*(229+85*1.8+1270.8) = 1582; observed 1521; 3.8% diff
blazing lv10: (1-863/(863+85*40))*1.2*229 = 219; observed 218.

That's closer, but formula structure still seems wrong.

%diff is higher for skills with higher weapon damage % component.
Blazing with 0% weapon damage is predicted correctly, so coeffs for masteries and damage reduction are probably correct too.

Deceptistar - Sanctuary · March 2010

its impressive but is it even useful when the weapons themselves give a minimum AND a maximum range of attack? cause in the end it just turns into a range of numbers. Unless your trying to calculate some mass DoT kiting project....

LTraveller - Dreamweaver · March 2010

That range can be small. For example, my base is 4858-4935 atm.
With larger ranges it can be useful for choosing best gear for DDing at bosses.
Understanding the structure of these formulas can be useful when choosing between, say, +mag and +matk gear.

Deceptistar - Sanctuary · March 2010

isnt it easier to just use pwi calc? if your goal is to choose which is better to use.

unless you are searching for specific numbers for some other reason.

LTraveller - Dreamweaver · March 2010

PWI calc will calc your base damage for any set of equipment, but will spell damage grow linearly with base matk? Seems not.

Deceptistar - Sanctuary · March 2010

LTraveller - Dreamweaver wrote: »

PWI calc will calc your base damage for any set of equipment, but will spell damage grow linearly with base matk? Seems not.

mm its kinda confusing because it seemed that you were searching more for the method than the results (which could be summed up by the calc)

...which is actually close, but weapon damage is slightly higher than it should be. So either the structure of damage formula is different, or there is something that adds to your weapon damage even when you don't have anything equipped; + I don't know how damage reduction coefficient is calculated.

One thought.
Our stats come from our 4 main VIT MAG DEX STR + gear/mastery. Assuming the mobs and bosses themselves could have an invisible 4 stat (which can possibly be calculated by going backwards on a lv150 boss using database def/attck numbers maybe?) could that possibly affect the decrease or increase in the resistance of def or attack on the mobs as a separate factor? (when it clashes with ours)

Magicsaber - Dreamweaver · March 2010

Deceptistar - Sanctuary wrote: »

mm its kinda confusing because it seemed that you were searching more for the method than the results (which could be summed up by the calc)

One thought.
Our stats come from our 4 main VIT MAG DEX STR + gear/mastery. Assuming the mobs and bosses themselves could have an invisible 4 stat (which can possibly be calculated by going backwards on a lv150 boss using database def/attck numbers maybe?) could that possibly affect the decrease or increase in the resistance of def or attack on the mobs as a separate factor? (when it clashes with ours)

We can get real Magic Defense of any mob or boss from pwdatabase.com, for example,
for Nien Beast Mighty (life - 118 584 802):
Physical Defense

1545

Magic Defense

Metal: 2650

Wood: 2650

Water: 2650

Fire: 2650

Earth: 2650.

Deceptistar - Sanctuary · March 2010

Magicsaber - Dreamweaver wrote: »

We can get real Magic Defense of any mob or boss from pwdatabase.com, for example,
for Nien Beast Mighty (life - 118 584 802):
Physical Defense

1545

Magic Defense
Metal: 2650
Wood: 2650
Water: 2650
Fire: 2650
Earth: 2650.

i know, but i meant on the 4 stats lvl

LTraveller - Dreamweaver · March 2010

Deceptistar - Sanctuary wrote: »

mm its kinda confusing because it seemed that you were searching more for the method than the results (which could be summed up by the calc)

Actually the question is: what's better to have in gear? mag, matk or -chan? Maybe a mix?
Why mix? Because base apparently includes mag*matk, so it might be more effective to add to both than to max one and disregard the other. E.g. you have 1 mag, 1 matk, and can increase either both by 1, or one of them by 2. In first case mag*matk goes from 1*1 to 2*2, in second -- only to 1*3.
Also, what's better: necklace which increases your mag, thus increasing base by X, or shard which increases your weapon damage, and thus base, and base increases by Y<X? Shard may be better, because it increases both base and weapon damage, and if you look at wizards spells, their weapon damage multipliers go up to 5.
Ideally, I'd like to see an addition to the calc which would predict damage for all spells (at all levels) for any set of gear.

Deceptistar - Sanctuary wrote: »

One thought.
Our stats come from our 4 main VIT MAG DEX STR + gear/mastery. Assuming the mobs and bosses themselves could have an invisible 4 stat (which can possibly be calculated by going backwards on a lv150 boss using database def/attck numbers maybe?) could that possibly affect the decrease or increase in the resistance of def or attack on the mobs as a separate factor? (when it clashes with ours)

It is possible that formula for damage reduction is different from what I'm using, but it's not important for now. observed_damage/predicted_damage should still remain the same for different spells; in practice, it doesn't: see post #9.

trying to predict spell damage

Comments

Categories