attack formula of pets

rinuruc

based on the info from that japanese wiki, i found the attack formula for phoenix is second-degree polynomial, (i just used the atk-values in function of the level, and found them to be atk = ax^2 + bx + c
with x = level of phoenix
a=0.37227272727272727272
b=11.7942424242424242
c=19.666666666666
and if i take a third degree polynomial than the coefficient on x^3 is of order 10^(-6) so its really a second degree polynomial model for attack
the errors i get are of the order of one digit after the "," so its supposed to be exact.
this is only for the phoenix, but its possible to calculate it for a few pets and than see how the factor of that growth chart comes into the a b and c from that formula,...
that way it should be possible to calculate for every pet what attack it has at what level

(the link where the attack values are is:
http://translate.google.com/translate?hl=nl&sl=auto&tl=en&u=http%3A%2F%2Fkanbisekai.wikiwiki.jp%2F%3F%25CD%25C5%25C0%25BA%25CD%25D1%25A5%25DA%25A5%25C3%25A5%25C8%252F%25C0%25AE%25C4%25B9%25B5%25AD%25CF%25BF

now all that has to be done is do some quadratic regression on attack values of other pets to find the correct values,...
anyone wnna help on this???

the things i used is the attack value of nix on levels 10 20 30 40 50 60 70 80 90 100 which i got on that site and ofc the levels emself, and let least squares method with a quadratic function do its job,...

for scorpion:
a=0.3042565001
b=9.514168858
c=15.27782647
based on the atk on that site, for level 10 20 40 50 60 70 90 (30 n 80 aint on that site)
anyone wnna do this stuff too, to figure out how it works exactly??? (gnna do walker next, coz that one s also 6 on atk on the growth chart)

walker:
a=0.3040584416
b=8.541320346
c=5.871428571

seems like that a=0.304 is same for pets with 6 on atk,... and the b and c seem to be somewhat diffrent,...
with more data i could figure out the exact values,... if i had like the stats from level 1 to 20 or so, from a pet,...
with that i could get correcter values,... to see if something changes like every 10 levels r so,... coz atm the stats i got r each time from every ten levels,... so that can easily change the results of the values i get

the fact that "a" is higher on nix is logic coz thats the quadratic term,... and the fact of a being approx same for walker/scorpion is also logic coz they got same growthchart for atk,...
anyone else wnna figure out some more of that stuff?

Solandri - Heavens Tear

I suspect there's only two variables since the pet's attack value only has two variables - the growth rate and initial starting value (and level, but that's your x). So if a is dependent entirely on the growth rate, b and c should be related and based on the starting value.

You may want to read up on the work Brael did on this.

http://pwi-forum.perfectworld.com/showthread.php?p=5788561

rinuruc

Solandri - Heavens Tear wrote: »

I suspect there's only two variables since the pet's attack value only has two variables - the growth rate and initial starting value (and level, but that's your x). So if a is dependent entirely on the growth rate, b and c should be related and based on the starting value.

You may want to read up on the work Brael did on this.

http://pwi-forum.perfectworld.com/showthread.php?p=5788561

if brael were still here, we could just use the spreadsheet he had,.... sadly his upload is gone, and nobody knows him irl or so, to contact him coz his upload is gone,...

with that full atk-stats of the nix i might be able to do some more ;-)
but id want attack of scorpion from lvl 1 to 50ish r so, and same for walker,...
i already figured the second degree factor depends purely on the growth chart-index,... n the b and c will depend on starting-level,... just gotta figure out how that works n than its finished

rinuruc

http://pwi-forum.perfectworld.com/showthread.php?t=360982

when using those pet-attack stats for the nix i get
a=0.3721
b=11.8048
c=19.222

with residuals normally distributed so theres nothing that gets added every ten levels or so, or id notice it in the distribution of the residuals,... will see if i can get something out of it with rounding errors etc,...




i guess for those with 7 on growth chart the a=0.372, and for 6 on growth chart its a=0.304,
further than that the b and c will probably depend on the starting-stats,... but will figure that out later,...
so far the fact that a is same for all pets with same number on atk in the growth chart and the fact that its second degree polynomial is the only thing i figured out till now,...
but its already quite nice to have gotten that,... (coz this way im not off by several % like brael was,... but im only off by 0.5 atk maximum,...)

rinuruc

if u guys dont reply with usefull info than i guess i will

if u take
y = attack at level x,
x_0 = level of taming,
y_0 = attack at level x_0,

than u can rewrite it a bit as follows
y(x) = a x^2 + b x + c
y_0 = y(x_0) = a (x_0)^2 + b x_0 + c
assuming y_0 is known, u can write y(x)-y_0 as follows
y(x)-y_0 = a x^2 + b x + c - [ a (x_0)^2 + b x_0 + c]
y(x)-y_0 = a (x^2 - (x_0)^2) + b (x-x_0)
this means
y(x) = a (x^2 - (x_0)^2) + b (x-x_0) + y_0
that means only 2 unknowns now,... ;-)
but that means i gotta know the atk at level of taming,... which i dont most know most of the time,... (i hope pw wiki has that info on it somewhere,... )

and since a is depending on the growth-chart, and that a will be an easy parameter to figure out, this reduces the problem of "what is b"


edit: http://pwi-wiki.perfectworld.com/index.php/Ultimate_Pet_List
there i find the attack at level of taming, and the level of taming of the pets,...
so now only thing to do, is figuring out what is "a" for the 1-7 on growth charts, and figuring out where "b" is derived from,... i assume that it is going to be something depending on the growt-type of pets,... (a parameter that is pet-family-dependent or so,...)




example:
Varicose scorpion: tamed at level 2,
attack at level 2, is 21 according to pwi-wiki,
its 6 on its growth chart for attack,...
so that means
y_0 = 21
a=0,304
x_0 = 2

so y(x) = 0,304 (x^2 - 2^2) + b (x-2) + 21

(havent figured out how to get the "b", but from the second degree polynomial fitting i did, i know its value,... just gotta figure out where it comes from, and what its value is for diffrent pets, and if theres a pattern in there, so just for the example: take b= 9,51

so attack at level 90 becomes
y(90) = 0,304 (90^2 - 4) + 9,51 (90-2) + 21 = 3319,064
and ecatomb says 3336,...
the diffrence i get here is probably due to rounding errors,... (still need to figure out how many digits they take for a and b,... and if a and b are really constant for growth chart of pet, or if they are diffrent for walker and scorpion,... (if the digits i find diffrent are significant or not)) (and if the info of attack on the wiki is with full loyalty or not,... anyone go tame a level 2 scorp n tell me its attack???)

rinuruc

rinuruc wrote: »

http://pwi-forum.perfectworld.com/showthread.php?t=360982

when using those pet-attack stats for the nix i get
a=0.3721
b=11.8048
c=19.222

with residuals normally distributed so theres nothing that gets added every ten levels or so, or id notice it in the distribution of the residuals,... will see if i can get something out of it with rounding errors etc,...




i guess for those with 7 on growth chart the a=0.372, and for 6 on growth chart its a=0.304,
further than that the b and c will probably depend on the starting-stats,... but will figure that out later,...
so far the fact that a is same for all pets with same number on atk in the growth chart and the fact that its second degree polynomial is the only thing i figured out till now,...
but its already quite nice to have gotten that,... (coz this way im not off by several % like brael was,... but im only off by 0.5 atk maximum,...)

-im not spamming btw-

if i use that a=0.3721, and the b= 11.8048, and the fact that at level 17 it has 327 attack,...

i find for level 90 nix
0.3721 (90^2 - 17^2) + 11.8048 (90-17) + 327 = 4095.2235
and ecatomb says its 4096

i guess my formula is right,... (i just used lvl 17 as tame-level and the attack at lvl 17, coz thats what i found in braels topic,... i didnt find the atk on level 1 of nix, at full loyalty,... ;-) )
this proves my formula is good XD
now for other pets with other growthcharts,... ;-)



-now reply someone that i think this is nice or so, coz i feel like im doing this for nothing, when there comez no reaction,... though i m quad-posting and nobody says anything to it, so i guess im doing good??? -

the editbutton will make it a huge wall of text and doesnt show any structure at all,... so nobody will read it XD

sooo i now need a heap of data on pets with diffrent growth charts,... anyone got data of scorpion and walker from level 2 to 50ish or so or for walker from 19 to 60ish,... just to get exact values for "a"

rinuruc

80 views n 1 reply b:cry
nobody wnna help? or nobody know how to help???

Mayfly - Dreamweaver

The figures in the Ultimate Pet List on the wiki are taken (at least the ones I entered) from either newly tamed eggs or the auction house. A Tame level 2 scorpion has 30 attack, so the 21 must be for either Savage or Wild (not sure which has a factor of .7 on the attack.)

The below figures are for a loyal Pollen Scorpion (attack modifier 1.2 included in below figures) which I just leveled up to 35 in Scorpion Valley, mostly on the level 32 Sabertail Scorpid King and his 6 level 30 followers, which can all usually be killed with a single cast of Noxious or Nova at my level (increased life survive and can be dispatched with a single cast of venomous, and occassionally one or two are outside the AOE and must be killed separately.) I only took it up to 35, as I was already getting into the diminishing returns for kills below the level of the pet at that point.

2: 36
3: 46
4: 58
5: 70
6: 84
7: 97
8: 111
9: 126
10: 141
11: 157
12: 174
13: 190
14: 208
15: 226
16: 246
17: 265
18: 285
19: 306
20: 327
21: 350
22: 372
23: 396
24: 420
25: 444
26: 469
27: 494
28: 520
29: 547
30: 574
31: 603
32: 632
33: 661
34: 691
35: 721

As an additional data point, my level 80 varicose scorpion (loyal, of course) has an attack of 2724.

rinuruc

ur data gives a=0.303738, b=9.542575, c=15.1847
n with that data, if i calculate for lvl 90, i find 3334.3 atk, and ecatomb says 3336
and for lvl 100, i get 4006.83 n ecatomb says 4008,... so with a bit of tweaking those data, ill find the exact values of "a" and "b"
(can do lots of more stuff like including the data from ecatomb, coz those r influential datapoints etc,... but i leave on holiday for a few weeks, so lots of preparing still to do, so wont be able to figure out much in the next weeks)

Brael - Dreamweaver

rinuruc wrote: »

if brael were still here, we could just use the spreadsheet he had,.... sadly his upload is gone, and nobody knows him irl or so, to contact him coz his upload is gone,...

with that full atk-stats of the nix i might be able to do some more ;-)
but id want attack of scorpion from lvl 1 to 50ish r so, and same for walker,...
i already figured the second degree factor depends purely on the growth chart-index,... n the b and c will depend on starting-level,... just gotta figure out how that works n than its finished

I haven't checked the boards lately but I've been back to the game for a little bit now I came back 2 months ago, quit after 2 weeks, came back again recently... we'll see if I stick around this time. I still have the spreadsheet I believe so I'll put it up again soon.

The formula I have isn't 100% exact but it gets within 2% or so. I know where the error is but I'm not quite sure how to fix it with the method I'm using. However, another poster once figured out a very accurate formula that worked on a scorpion only. My formula to convert between one pet and another works flawless (can convert between different attack speeds and attack growths) so I may simply update to use a more accurate formula on one specific pet, that then converts to whatever pet is being used. Though my goal was more to use the exact formula they use in the game, and from that perspective my current method is closer than the other method.

I've looked at yours a little bit. It looks to me like A is based on attack growth of the pet, then B and C both seem to vary based on attack speed and starting level. One hint I have for you, that I discovered from my own work is that a phoenix has a ridiculous growth level of 22 (think that's right, it's been awhile) so one quick way you can eyeball if your formula is working correctly for other pets is knowing there's a much bigger gap in A between a nix and scorpion than between a scorpion and anything else.

tweakz

Not sure if these are it:

http://opposingdigits.com/pwi

Mayfly - Dreamweaver

Verdant Qingfu (all figures are for Tame status, 1.0 attack modifier):

1: 14
2: 20
3: 25
4: 32
5: 38
6: 45
7: 52
8: 60
9: 68
10: 76
11: 85
12: 93
13: 103
14: 112
15: 122
16: 132
17: 142
18: 153
19: 164
20: 176
21: 188
22: 200
23: 212
24: 225
25: 238
26: 251
27: 265
28: 279
29: 293
30: 308

A couple more data points on the varicose scorpion (loyal on both of these):

45: 1059
66: 1969

Mayfly - Dreamweaver

Heptastral beetle, tame status:

2: 17
3: 23
4: 28
5: 34
6: 41
7: 47
8: 54
9: 61
10: 69
11: 77
12: 85
13: 93
14: 102
15: 111
16: 120
17: 130
18: 139
19: 150
20: 160
21: 171
22: 182
23: 193
24: 205
25: 217
26: 229
27: 241
28: 254
29: 267
30: 281

Taking requests for other low level, non-legendary pets. It takes me about an hour to get them to 30 at tame.

Mayfly - Dreamweaver

Cruel wolfling, tame status:

2: 21
3: 27
4: 34
5: 41
6: 48
7: 56
8: 64
9: 73
10: 81
11: 91
12: 100
13: 110
14: 120
15: 131
16: 142
17: 153
18: 165
19: 177
20: 189
21: 202
22: 215
23: 228
24: 242
25: 256
26: 271
27: 285
28: 301
29: 316
30: 332

Ecatomb - Harshlands

The atk formula seems working. But a,b and c arent the same for all pet
I could have these data for pre-tideborn pet and Shadow Ranger + Glacial Walker

But I need your help for other tideborn pet. So I need the stat of all tideborn pet at starting lvl, lvl40 and lvl60 (so will got better a,b and c).

Already trying to make a pet database with that

Brael - Dreamweaver

What tweakz posted is the attack sheet I have, though I'm not sure if it's the most current version.

Here's the version I have, this time I hosted it on my own server so the download doesn't disappear.
http://www.evilgamer.net/utilities/petstats.xls

In addition to that, I have a sheet with various pets at different stages of loyalty, attack growth, and level. Since it doesn't take starting level into consideration though, I would be careful about using any pet other than the herc, jaden qingfu, and scorpion (a correct formula would be off by 1-2 points for the scorpion).
http://www.evilgamer.net/utilities/petattack.xls

That sheet might be useful. Probably the biggest use out of it is that it shows how to convert one pets stats to another (the same method is in my pet attack thread).

Also, rinuruc if you want some more data to look at take a look at Pertia's thread . It's by far the best attack formula so far (that I know of atleast) because it's accurate for the entire family of scorpion pets which means my conversion method can be easily adapted to it allowing for the formula to be used to predict all pets. The only problem with it to me, is that I'm not sure how to get it to work in a spreadsheet though I do have it working on my graphing calculator.

I suppose one solution would be to solve at every point for 1-105 (or higher in the event of future levels), put that into a table, and have a spreadsheet perform a lookup, but I would rather have one universal formula that reflects what the developers actually put into the game. The reason I haven't done that though, is Pertia's formula, while very very good is still off by a couple of points.

Already trying to make a pet database with that

Ecatomb, if you want a pet database download my sheet. It has complete growth stats for every pet in the game that a veno can tame along with a full listing of skills, starting skill levels, and tame levels. Though when it comes to tideborn pets, I went with the name of the mob that's tamed rather than the name of the pet you get so make sure to change those and just to be safe, and I would double check all tideborn pets (what few there are) as I added their stats purely by eyeballing the growth chart rather than leveling them to verify the growth rate.

Infact, it's more accurate than the wiki list which is using data gained from pwdatabase. That data has an error and lists a few pets which don't actually exist in the game. My sheet doesn't include the non existent pets, and goes a step further by including the pets race. All raw stats are under the pet stats tab, other tabs translate those stats to the pets ingame stats.

Ecatomb - Harshlands

Thank you Brael, will check your file.

I have done this, it isn't the final version but it's working:
http://www.ecatomb.net/ms/pet_database.php

Can you check if atk is ok if you got one of these pet?
Thank you.

Brael - Dreamweaver

I checked the pets in my bag and compared them to what you have.
Herc - level 98, 3098 for both
Nix - level 98, 4752 real 4751 for you
Marksman - level 97, 3620 for both
Scorpion - level 76, 2496 for both
Armored Bear - level 84, 1470 real, 1469 for you

The other pets aren't listed on your page.

Ecatomb - Harshlands

So formula is working nice b:victory
Ty rinuruc for it (I think you are the "someone" of guestbook)

Didn't added other pet yet, will be added this week-end.

ps: need to fix the m/s of the page but all other stat should be ok

Brael - Dreamweaver

I came up with a new idea so I'll try it tonight, if it works out I should be able to use just one formula for every pet and have attack within 0.2% (1/10 the margin of error it currently has). The idea I've got isn't as elegant as I would like, but it will work for now I suppose. However, I still think the way to ultimately go is a formula that can figure out exact attack from just current level, starting level, attack rate, and the attack growth.

So formula is working nice
Ty rinuruc for it (I think you are the "someone" of guestbook)

Didn't added other pet yet, will be added this week-end.

ps: need to fix the m/s of the page but all other stat should be ok

Some suggestions for you when you make that page:
1. Convert resistance to damage reduction. Knowing a pet has 70% damage mitigation is much more useful than knowing it has X physical resistance. Alternatively list it in the form of damage taken as that's more understandable.
2. Allow for up to 4 damage skills to be placed on the pet for dps calculations.
3. Allow for the skill level of bash to be adjusted.
4. Remember to account for skill cast time when calculating dps. The cast time is 1.5, you can check some of my old threads if you need the methodology used to get the time.
5. For the sake of simplicity it's probably a bad idea to list every single pet like I did, but it wouldn't be a bad idea to include the lowest and highest level of each pet family as the highest have some advantages that make them worthwhile. An Ethereal Inamorato and Undine Mystic for example make great pullers, the Mystic is a good water pet you don't need to level as well.
6. Sort pets logically somehow. For example by air/land/water then alphabetically or something. Just some way that makes it less jumbled.

Mayfly - Dreamweaver

Ocean Chanter, loyal status:

7: 57
8: 66
9: 75
10: 85
11: 94
12: 105
13: 116
14: 127
15: 139
16: 151
17: 163
18: 176
19: (didn't record, but interpolating should be 189 +/- 1)
20: 202
21: 216
22: 230
23: 246
24: 260
25: 276
26: 291
27: 308
28: 324
29: 342
30: 358
31: 376
32: 394
33: 412
34: 432
35: 451
36: 471
37: 490
38: 512
39: 532

Aniella - Harshlands

if u ppl talk about pets w good patk..what about sawfly,kowlin.9tailed foxie. they have hige dg build but low hp/def

tweakz

Aniella - Harshlands wrote: »

if u ppl talk about pets w good patk..what about sawfly,kowlin.9tailed foxie. they have hige dg build but low hp/def

no wee tak bout peenut buttr hav gud flavr

Aniella - Harshlands

tweakz wrote: »

no wee tak bout peenut buttr hav gud flavr

:P Hei ya forgot somthing..read the headline^^ btw, to much peenutbutter make u fat <.<

Mayfly - Dreamweaver

We're not talking about what pets have the best patk, we're talking about how to calculate the patk for any pet.

Aniella - Harshlands

i know..but ya peps use a hard way to find it out..lol..there is a nother way that are faster and only w 6 words not 7000 words^^ maybe easier to understand

tweakz

Aniella - Harshlands wrote: »

:P Hei ya forgot somthing..read the headline^^ btw, to much peenutbutter make u fat <.<

Yew lyk me plumpkin bumpkin! b:victory

Brael - Dreamweaver

Well, some progress on what I've done. I used nix stats to create a table of base attack rather than calculating it. The reason for using a nix is that due to the level 1 tame I don't have to worry about attack loss, and due to the 1 attack/sec I can remove a variable from the attack because the base attack would simply be multiplied by 1. Doing this, I'm a bit closer now on every pet, and fairly confident that base attack is close to correct at every level. However, I'm running into three issues.

1. I'm not accounting for pets level 1 stats. Each pet race seems to have it's own arbitrary level 1 attack. This is nothing new as pets have their own arbitrary amounts of other stats too, for example a phoenix has 52 physical resist at level 1, despite the fact it actually gains 100 per level. Still, that throws calculations off a little.

2. Base attack is using displayed numbers, so decimals aren't being accounted for. Since I had to use 120% attack for many numbers, the base is off by 1-2 points. Which again throws things off a little.

3. I'm not calculating attack lost due to tame level properly. This one makes no sense to me, but it's the conclusion I come to, and the cause of 90% of wrong stats. The higher the tame level of a pet the further off the results are. Maybe someone can tell me what I'm missing. Currently I'm using the attack of the pet at the tame level/10. That doesn't seem correct with my method. A 98 Ethereal Inamorato is off by 300 points for example, yet a herc is within 20 points (off slightly due to issues 1 and 2).

Aniella - Harshlands wrote: »

i know..but ya peps use a hard way to find it out..lol..there is a nother way that are faster and only w 6 words not 7000 words^^ maybe easier to understand

The point is to find a way to know the attack of every pet at every level. That takes thinking and math (or someone leveling up all 304 pets to level 105 and recording the full stats of every pet at every level). It's not simply about which pet is the best in each situation, that's been known for awhile now. If you know a faster way to accomplish that with only 6 words, feel free to post it.

Ecatomb - Harshlands

Mayfly - Dreamweaver wrote: »

Ocean Chanter, loyal status:

...

This pet: http://www.pwdatabase.com/pwi/mob/25092 ?
If yes, can you post full stat for 2 different level? Thank you.

Aniella - Harshlands

tweakz wrote: »

yew Lyk Me Plumpkin Bumpkin! B:victory

Aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaahhhh Ur My Best Friend Rigth!!!!!!!!!!!!!!!!!!!!!!!!!!

Mayfly - Dreamweaver

Ecatomb - Harshlands wrote: »

This pet: http://www.pwdatabase.com/pwi/mob/25092 ?
If yes, can you post full stat for 2 different level? Thank you.

Yes, that one.

Tame, level 7:

HP 248
PAtk 48
PRes 395
MRes 395
Acc 193
Evade 129
Atk 0.8/sec
Speed 5.8 m/sec

Tame, level 39:

HP 1208
PAtk 444
PRes 2528
MRes 2528
Acc 940
Evade 626
Atk 0.8/sec
Speed 6.1 m/sec