Remove Advertisements

A Call to Arms - Cataclysm Mechanics testing

Warning: Theorycraft inside.

Moderators: Fridmarr, Worldie, Aergis, theckhd

Re: A call to arms - Beta/PTR parses on WoL

Postby Rasmfrackn » Wed Sep 29, 2010 11:27 am

Level 85 Human premade on the Raider's Dummy with only the weapon and shield equipped. 1021 AP, 331 SP.

I've uploaded my log, but a quick min/max/avg set of values seemed useful, assuming these are all uniform distributions. They're slightly off I guess, but I was only checking 1 ability at a time so I may have missed higher or lower values when I was doing other things.

http://www.worldoflogs.com/reports/5z65 ... details/6/

Code: Select all
         min  max  avg
ShoR    1231 1231 1231
CS      1134 1817 1475
JoT      520  520  520
AS      3318 3892 3605
HW      2600 2601 2600
HoW            0
Exor    1298 1410 1354
SoT      167  240  203
SoR       57   57   57
SoJ       24   24   24
Cens    1002 1008 1005
Cons    1640 1650 1645
HotR     181  281  231
HaNova   713  970  841
Melee    601  951  776
Rasmfrackn
Dwarf Paladin
Icecrown Server
Eng/Scribe/Masochist
User avatar
Rasmfrackn
 
Posts: 1014
Joined: Fri Aug 17, 2007 12:11 pm

Re: A call to arms - Beta/PTR parses on WoL

Postby theckhd » Wed Sep 29, 2010 11:42 am

Ok, so the first weird thing is that my model gives me 995 AP and 326 SP with just the weapon and shield equipped. So something is causing a discrepancy of 26 AP and 5 SP...

Could you post your base (naked) values when you get a chance? I'm guessing my base stats are wrong.

If I substitute your AP and SP, here's what I get:

Code: Select all
       Model   Ras   Diff  Err%
ShoR    1225  1231      6     0
CS      1285  1476    191    15
JoT      520   520      0     0
AS      3615  3605    -10    -0
HW      2500  2601    101     4
HoW     1522     0  -1522  -100
Exor    1354  1354      0     0
SoT      199   204      5     3
SoR       53    57      4     8
SoJ       24    24      0     0
Cens     820  1005    185    23
Cons     409  1645   1236   302
HotR     340   231   -109   -32
HaNova   729   842    113    16
Melee    669   776    107    16


ShoR, Judgement, AS, Exor, SoT, SoJ all seem fine. SoR is probably fine also, but I'll hand-check that to be sure. HW seems a bit low, it's only off by 4% but 100 damage is a decent discrepancy. CS, Censure, Consecration, HotR/HammerNova, and Melee are all off by a fair bit though.

Are you excluding glancing hits from the Melee calculations?
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7710
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: A call to arms - Beta/PTR parses on WoL

Postby Rasmfrackn » Wed Sep 29, 2010 11:56 am

theckhd wrote:Could you post your base (naked) values when you get a chance? I'm guessing my base stats are wrong.

Holding mace/shield.
Wearing no gear.

Are you excluding glancing hits from the Melee calculations?

I did exclude glances, but I may not have seen the edges of the range in practice; that hammer has a huge damage range on it. My paperdoll damage range is 849-1416 with mace/shield.
Rasmfrackn
Dwarf Paladin
Icecrown Server
Eng/Scribe/Masochist
User avatar
Rasmfrackn
 
Posts: 1014
Joined: Fri Aug 17, 2007 12:11 pm

Re: A call to arms - Beta/PTR parses on WoL

Postby theckhd » Wed Sep 29, 2010 12:17 pm

Thanks, I was still using the base values for level 80. I have the base human values updated now (though if anyone minds checking the base STR/STA/AGI/INT/SPI for a lvl 85 Dwarf, Draenei, BE, and Tauren I'd appreciate it).

Now my AP & SP values match your paper doll, so that's good.

The consecration data is really far off, would you mind repeating the Cons test naked or with a Dalaran Sword? That will tell me whether they gave it a base damage or whether they tweaked the SP/AP coefficients.

If you could retest Censure, HotR, CS, and melee at the same time, that would be good too.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7710
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: A call to arms - Beta/PTR parses on WoL

Postby theckhd » Wed Sep 29, 2010 12:45 pm

Aha, found the error for Crusader Strike, and likely for melee. stat_model was using the level 80 armor_c formula, which led to larger physical damage reduction than expected. Here's what I have now:

Code: Select all
       Model   Ras   Diff  %Err
ShoR    1225  1231      6     0
CS      1483  1476     -7    -0
JoT      520   520      0     0
AS      3615  3605    -10    -0
HW      2500  2601    101     4
HoW     1522     0  -1522  -100
Exor    1354  1354      0     0
SoT      199   204      5     3
SoR       53    57      4     8
SoJ       24    24      0     0
Cens     820  1005    185    23
Cons     409  1645   1236   302
HotR     340   231   -109   -32
HaNova   729   842    113    16
Melee    772   776      4     1
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7710
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: A call to arms - Beta/PTR parses on WoL

Postby Arianne » Wed Sep 29, 2010 12:50 pm

A total per cast value of 409 seems low for Consecration. Are you sure that's not a tick value?
Arianne
 
Posts: 502
Joined: Wed Jul 11, 2007 1:22 pm

Re: A call to arms - Beta/PTR parses on WoL

Postby theckhd » Wed Sep 29, 2010 1:06 pm

I figured out what's happening with HotR. It's being mitigated like physical damage. Observe:

base=0.3*player.wdamage=340
mdf.phdmg=0.6818 (same as for CS, melee, etc)

0.3*player.wdamage*mdf.phdmg=231.7

It shows up as holy damage on WoL, does it show up like that in game? If so, I would assume this is an error in the coding of HotR in-game, and should probably be brought to Blizzard's attention.


Arianne wrote:A total per cast value of 409 seems low for Consecration. Are you sure that's not a tick value?

Nope, it's too high for a tick value. The formula that I have is:
(8.*(0+0.027.*player.hsp+0.027.*player.ap))*mdf.HalGro

This differs from what's on mmo-champ and wowhead only in the coefficients - mine are updated to a value announced in a previous patch (21.6% SP and AP). But neither of the formulas make sense:

For 1021 AP, 331 SP, we get
(8*(0+0.027*331 +0.027*1021))*1.4 = 409 total damage over 10 seconds
(8*(0+0.04*331 +0.04*1021))*1.4 = 606 damage over 10 seconds

None of that matches what he observed:
Code: Select all
[13:04:47.765] Rasmtwo casts Consecration
[13:04:48.765] Rasmtwo Consecration Raider's Training Dummy 166 (O: -1)
[13:04:49.968] Rasmtwo Consecration Raider's Training Dummy 166 (O: -1)
[13:04:50.718] Rasmtwo Consecration Raider's Training Dummy Miss
[13:04:51.984] Rasmtwo Consecration Raider's Training Dummy 165 (O: -1)
[13:04:52.796] Rasmtwo Consecration Raider's Training Dummy 165 (O: -1)
[13:04:54.000] Rasmtwo Consecration Raider's Training Dummy 166 (O: -1)
[13:04:54.796] Rasmtwo Consecration Raider's Training Dummy *247* (O: -1)
[13:04:55.812] Rasmtwo Consecration Raider's Training Dummy Miss
[13:04:56.812] Rasmtwo Consecration Raider's Training Dummy 165 (O: -1)
[13:04:57.828] Rasmtwo's Consecration fades
[13:04:58.015] Rasmtwo Consecration Raider's Training Dummy 165 (O: -1)


The ticks he observed were ~165, with one tick every second for a total of 10 ticks. That's a net damage of 1650, like he posted. The numbers from the tooltips and my formula are both too large to be ticks, but way too small to be total damage.

That probably means they added a base damage to the spell and the tooltips on wowhead/mmo haven't been properly updated. If the coefficients are 8*0.04=32% of sp and ap, the base damage is 1217. If the coefficients are 21.6%, the base damage is 1358. Once we know exactly what the scaling coefficients are, we'll be able to estimate the base damage and fix that portion of the code.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7710
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: A call to arms - Beta/PTR parses on WoL

Postby Rasmfrackn » Wed Sep 29, 2010 1:07 pm

Dwarf stats (I just lucked out and procced a race change.)

I'm logging with weapon + shield still, since I need them for CS and HotR. As a dwarf, my censure is definitely doing 168 per tick, with the occasional 169. I get 6 ticks after a refresh when I back off. Consecrate definitely ticks for 165.

Naked, consecrate is 142-143 per tick, 10 ticks. Censure ticks for 94-95.

I bought a Mace (item name, damage range 8-16, dps 6.5) and re-did CS and HotR. CS is 144-154, HotR is 18-20 plus 660-906 (I don't understand this range.. the tooltip doesn't even state a range, just a fixed #).

dwarf with premade mace/shield
dwarf with only a white mace with a 8-16 range

-- Interesting... sometimes HotR just fails and doesn't show up in the log. I noticed when no numbers popped up, but it did seem odd somehow... now I realize that it wasn't popping up misses or anything either. Sometimes it just makes the sound and nothing happens.
Last edited by Rasmfrackn on Wed Sep 29, 2010 1:12 pm, edited 1 time in total.
Rasmfrackn
Dwarf Paladin
Icecrown Server
Eng/Scribe/Masochist
User avatar
Rasmfrackn
 
Posts: 1014
Joined: Fri Aug 17, 2007 12:11 pm

Re: A call to arms - Beta/PTR parses on WoL

Postby Sabindeus » Wed Sep 29, 2010 1:09 pm

Rasmfrackn wrote:I bought a Mace (item name, damage range 8-16, dps 6.5) and re-did CS and HotR. CS is 144-154, HotR is 18-20 plus 660-906 (I don't understand this range.. the tooltip doesn't even state a range, just a fixed #).


There's a checkbox in the interface options somewhere for putting averages instead of ranges on tooltips that I believe gets checked by default
Image
Turn In, an NPC interaction automator - http://wow.curse.com/downloads/wow-addo ... rn-in.aspx
User avatar
Sabindeus
Moderator
 
Posts: 10470
Joined: Mon May 14, 2007 9:24 am

Re: A call to arms - Beta/PTR parses on WoL

Postby Rasmfrackn » Wed Sep 29, 2010 1:17 pm

Sabindeus wrote:There's a checkbox in the interface options somewhere for putting averages instead of ranges on tooltips that I believe gets checked by default

So there is. That's new... still has the ! by it and everything. :) Naked range on the HotR holy damage is 654 to 945. With only premade mace/shield, it goes to 703 to 994.

Oh, and to your question theck, HotR shows up as a small physical hit and an AE holy hit. In the combat log that is. The small hit is labeled physical, the nova is labeled holy.
Rasmfrackn
Dwarf Paladin
Icecrown Server
Eng/Scribe/Masochist
User avatar
Rasmfrackn
 
Posts: 1014
Joined: Fri Aug 17, 2007 12:11 pm

Re: A call to arms - Beta/PTR parses on WoL

Postby theckhd » Wed Sep 29, 2010 1:26 pm

Great. I haven't started looking at your dwarf data yet, but I will shortly.

Conclusions from human data:
Holy Wrath doesn't have much variability, it's coming out as 2501 base damage instead of 2401 at level 85. I can only assume that the base damage was buffed by 100, since the SP scaling was pretty clearly laid out by a blue post a little while back.

I can't figure out what's up with the HammerNova damage. There are no other modifiers to buff it, so we're pretty much looking at a damage range issue. Either both max and min were buffed, or the min was increased to reduce the variability. I should be able to figure it out from the Dwarf data.

Consecrate is almost certainly a base damage problem; again, the Dwarf data should clear that up.

I didn't see any Censure data in your Human parse, and given what you've said about the Dwarf data, I'm inclined to write that off to human error. I get ticks of about 165 with my model, but your human data is suggesting ticks of over 200. The dwarf data looks to be right around the model.

I'll hopefully get a chance to scrutinize the Dwarf data in an hour or so.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7710
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: A call to arms - Beta/PTR parses on WoL

Postby Rasmfrackn » Wed Sep 29, 2010 1:43 pm

theckhd wrote:I didn't see any Censure data in your Human parse, and given what you've said about the Dwarf data, I'm inclined to write that off to human error. I get ticks of about 165 with my model, but your human data is suggesting ticks of over 200. The dwarf data looks to be right around the model.

I'll hopefully get a chance to scrutinize the Dwarf data in an hour or so.

Hmm... maybe there was a miscommunication in # of ticks? For my human numbers, I have 167 * 6 as the total damage for censure once i stop attacking. It ticks 6 times, not 5, so maybe that's the confusion. It was definitely 167 for human.
Rasmfrackn
Dwarf Paladin
Icecrown Server
Eng/Scribe/Masochist
User avatar
Rasmfrackn
 
Posts: 1014
Joined: Fri Aug 17, 2007 12:11 pm

Re: A call to arms - Beta/PTR parses on WoL

Postby theckhd » Wed Sep 29, 2010 2:12 pm

Rasmfrackn wrote:
theckhd wrote:I didn't see any Censure data in your Human parse, and given what you've said about the Dwarf data, I'm inclined to write that off to human error. I get ticks of about 165 with my model, but your human data is suggesting ticks of over 200. The dwarf data looks to be right around the model.

I'll hopefully get a chance to scrutinize the Dwarf data in an hour or so.

Hmm... maybe there was a miscommunication in # of ticks? For my human numbers, I have 167 * 6 as the total damage for censure once i stop attacking. It ticks 6 times, not 5, so maybe that's the confusion. It was definitely 167 for human.


You're getting 6 ticks when you disengage because the DoT is present already. But it's only 5 ticks every 15 seconds in the steady-state.

Ex: T=tick, - is a one second wait
---T---T---T---T---T---T---T---T---T---T---T---T---T---T---T
The underlined portion is what happens in 15 seconds. If you include the bolded tick right before the underline in the final section, then you're essentially including the 3-second period that precedes it. Otherwise you're putting 6 ticks in one 15-second period and 4 ticks in the one before it.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7710
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: A call to arms - Beta/PTR parses on WoL

Postby theckhd » Wed Sep 29, 2010 2:47 pm

Consecration

Dwarf Data:
Code: Select all
                        Tick
Gear     AP    SP    Cons   Cens
Naked   573   196   142.5   94.5
Prem   1031   333   165    168

diff    458   137    22.5   73.5


Sanity check: Censure's scaling is (0.013*sp+0.025*ap)*5*1.12 (5 stacks, 1.12 for SotP). It's easy to verify that these AP and SP values give you exactly what's observed in the Tick column for censure. Thus, we feel confident that our AP and SP numbers are right and that our Censure model is right.

If we assume the model for Consecration is 8*(base+c*(ap+sp))*1.4 (HG talent), we can determine the coefficient c by dividing the total damage difference by the sum of the ap and sp differences. I.e.

8*1.4*c=(diff_damage)/(diff_ap+diff_sp) = 22.5*10/(458+137) = 0.3782
c=0.0338

This is somewhere between the tooltip value of 0.04 and the old value of 0.027.

To determine the base value, we can solve the model equation for base and plug in the AP and SP values for the different gear sets:

tick*10=8*1.4*(base+c*(ap+sp))
base=tick*10/(8*1.4)-c*(ap+sp)

plugging in values for tick, c, ap, and sp, we get base=101.2 for both gear sets. So even at 0 ap and sp, Consecration will do about 1133 damage. Since both sets come out to about the same value, we can feel reasonably certain that the model is correct. It's still possible that the scaling with SP and AP use different coefficients, because our AP and SP gains are always proportional in this test. A simple check with an INT or SP trinket should be enough to clarify this detail though.

So the raw damage of Consecrate=8*(101.2+0.0338*(sp+ap))*(1+0.2*points_in_hallowed_ground)
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7710
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: A call to arms - Beta/PTR parses on WoL

Postby theckhd » Wed Sep 29, 2010 3:11 pm

HammerNova


We don't have quite enough data points here to solve the generic model. It's clear from the data that it's gained some sort of scaling coefficient, probably based on SP and AP (though it could use weapon damage for all we know).

If we assume a general form of base+c1*ap+c2*sp, we have three unknowns: base, c1, and c2. To solve that system in three variables, we need 3 equations; unfortunately, we only have two: one from the naked set, one from the premade sword/shield set.

If we assume that c1=c2, we can solve the system and find that c=0.0824 and base=590.6-881.6. But that's only if the underlying assumption of c1=c2 is true. If we get another data point with a different value of AP and SP (maybe via a trinket or ring - a non-STR AP or SP gain would be ideal since it would decouple ap and sp), we can figure out the coefficients.

<edit>Just as a sanity check, I tried using the max base damage from the tooltip (874) to see if it made sense. That gives us the following system of equations, in matrix form:

Code: Select all
| 573 196 945-874 |   | 573 196  71|
|1031 333 994-874 | = |1031 333 120|


Solving that gives us c1=-0.0109 and c2=0.3942. At first this looks like nonsense, but note that 333/196*71=120.6. In other words, subtracting 333/196 times the first equation from the second, and assuming that 120.6 should be rounded down to 120 to account for rounding error, we get:

(1031-974)*c1+0*c2=0
c1=0

In other words, if the tooltip max damage is correct, the damage doesn't scale at all with AP. That would mean that all 49 of the extra damage is due to spellpower, making c2=(120-71)/(333-196)=49/137=0.3577.

Checking this against the min damages seems consistent: 584+196*0.3577=654, and 584+333*0.3577=703. I'd like to get a few more data points to confirm this (again, a non-STR increase to AP or SP would help cement the model), but at least tentatively it looks like HotR does tooltip damage that scales with SP but not with AP.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7710
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

PreviousNext

Return to Advanced Theorycraft and Calculations

Who is online

Users browsing this forum: Google [Bot] and 1 guest


Remove Advertisements

Who is online

In total there are 2 users online :: 1 registered, 0 hidden and 1 guest (based on users active over the past 5 minutes)
Most users ever online was 380 on Tue Oct 14, 2008 6:28 pm

Users browsing this forum: Google [Bot] and 1 guest