A Call to Arms - MoP Mechanics Testing

[theckhd]

It seems that something is up with the str->parry conversion. I'd like some data sets from beta. Required:

Strength
Agility
Dodge
Parry
Dodge rating and pre-DR dodge from the tooltip
Parry rating and pre-DR parry from the tooltip

I need values while naked, and at various gear levels (the more dodge/parry/str the better). Multiple people can perform these tests, as long as they each give individual naked values (so I can adjust for varying base STR).

I'd also like a few data sets wearing agility gear, if possible. I.e. naked, and then only wearing leather agi gear (no str or avoidance rating). A commenter on my blog pointed out that we seem to be getting a small amount of dodge from agility still.

[Klaudandus]

I'll see what I can do...

nvm... read the rest of the post.

[theckhd]

Agility data isn't all that important, it's just something I'm curious about. The parry/str data is more imperative.

[daishan]

Does it need to be on a level 90 or would a level 85 or 86 be ok?

[theckhd]

Level 90. I'm not really interested in determining the exact conversion rate at 85.

[Klaudandus]

Doing the parry data right now, as you can prolly see

[Klaudandus]

Done, and confirmed dodge out of agi.

Theck, you should see the spreadsheet now.

[theckhd]

Thanks, I do see it. I'll take a look at it by tomorrow morning at the latest, depending on how busy I am tonight.

[Klaudandus]

Sure, no prob, just give it a quick glance in case you want me to do a different gearing set.

[Klaudandus]

oh... FFS, new build...

Nerfs, Nerfs everywhere.

Censure Deals ((126 + 11.0% Deals ((107 + 9.4% of SP) * 5) additional Holy damage over 15 sec.
Hammer of Wrath (Holy, Protection) Hurls a magical hammer that strikes an enemy for (1,838 + 161.0% of SP) Holy damage. 1,470 + 128.8% of SP) Holy damage.
Judgment (Protection) A magic attack that unleashes the energy of a Seal to cause (702 + 63.5% 562 + 50.8% of SP) Holy damage and generates one charge of Holy Power.
Seal of Truth Fills you with Holy Light, causing melee attacks to deal 14% 12% additional Holy damage and apply Censure to the target. Censure Deals ((126 + 11.0% 107 + 9.4% of SP) * 5) additional Holy damage over 15 sec.

Eternal Flame Consumes up to 3 Holy Power to place a protective Holy flame on a friendly target, which heals them for (4,260 + 37.7% of SP) and an additional (1,393 + 16.0% 391 + 4.5% of SP) every 3 sec for 30 sec.


Forgive me if I withhold my enthusiasm



These are the only ones that are buffs of sorts.
Sanctity of Battle (Protection) Spell Melee haste effects lower the cooldown and global cooldown of your Judgment, Crusader Strike, Hammer of the Righteous, Consecration, Holy Wrath, and Hammer of Wrath.
Divine Purpose Abilities that cost Holy Power have a 20% 25% chance to cause the Divine Purpose effect. Divine Purpose - Your next Holy Power ability will consume no Holy Power and cast as if 3 Holy Power were consumed. Divine Purpose Your next Holy Power ability will consume no Holy Power and will cast as if 3 Holy Power were consumed. Lasts 8 sec.
Holy Avenger Abilities that generate Holy Power will deal 30% additional damage and healing, and generate 3 charges of Holy Power for the next 15 sec.
Sanctified Wrath Avenging Wrath also increases healing received by 20%.


Will start testing new stuff as soon as the new build is up...

[theckhd]

Parry/Dodge data fitting

I only looked at three of the data sets, since all I need is a test to show whether my previously-calculated DR equations are correct or not.
Fitting file and data here: http://code.google.com/p/matlabadin/sou ... eqns_L90.m

Set #1: Str->Parry, 0 parry rating. X is str-base_str, Y is post-DR parry minus naked parry

Code: Select all

s2p_fit =

     General model:
     s2p_fit(x) = 1/(1/C+0.885/(x/a))
     Coefficients (with 95% confidence bounds):
       C =       239.3  (234.2, 244.4)
       a =       953.2  (952.3, 954)

s2p_gof =

           sse: 2.3377e-004
       rsquare: 1.0000
           dfe: 17
    adjrsquare: 1.0000
          rmse: 0.0037


s2p_fit =

     General model:
     s2p_fit(x) = 1/(1/C+0.885/(x/951.158596))
     Coefficients (with 95% confidence bounds):
       C =       228.5  (226.6, 230.4)

s2p_gof =

           sse: 5.6247e-004
       rsquare: 1.0000
           dfe: 18
    adjrsquare: 1.0000
          rmse: 0.0056


Certainly seems to confirm our equations a is still what GC provided in the Beta Analysis thread, C is still close to our estimate of 232. Note that C won't be very accurate here because we're so low on the DR curve.

Set #2: Parry rating and Str. X is pre-dr parry from rating (tooltip), Y is post-dr parry minus naked parry.

Code: Select all

p2p_fit =

     General model:
     p2p_fit(x) = 1/(1/C+0.885/(x+(9819-176)/a))
     Coefficients (with 95% confidence bounds):
       C =       236.5  (234.5, 238.5)
       a =       952.7  (952.1, 953.3)

p2p_gof =

           sse: 4.8893e-005
       rsquare: 1.0000
           dfe: 11
    adjrsquare: 1.0000
          rmse: 0.0021


p2p_fit =

     General model:
     p2p_fit(x) = 1/(1/C+0.885/(x+(9819-176)/951.158596))
     Coefficients (with 95% confidence bounds):
       C =       231.5  (230.8, 232.2)

p2p_gof =

           sse: 1.9492e-004
       rsquare: 1.0000
           dfe: 12
    adjrsquare: 1.0000
          rmse: 0.0040

Again, very consistent with our previous expression. More accurate bounds on C here - I'll be adding this data set to my surface fits later to see if it helps narrow C down any further.

Set #3: Agility to Dodge, 0 dodge rating. X is agi-base_agi, Y is post-DR dodge minus naked dodge

Code: Select all

a2d_fit =

     General model:
     a2d_fit(x) = 1/(1/65.631440+0.885/(x/a))
     Coefficients (with 95% confidence bounds):
       a =  1.002e+004  (9881, 1.016e+004)

a2d_gof =

           sse: 6.1157e-005
       rsquare: 0.9993
           dfe: 6
    adjrsquare: 0.9993
          rmse: 0.0032

There's no doubt from the data that we're still getting an Agi->dodge conversion. The fit indicates that the conversion factor is roughly 10k. In other words, 100 agi gives you around 0.01 pre-DR dodge percent.

So: confirm agi->dodge, confirm that the parry/str inequity observed by Darielle at L85 is not occurring at L90. Presumably they tweaked the str->parry coefficient scaling at lower levels, but left the L90 end point unchanged. I'm not really concerned about finding the exact DR formula for level 85, but if someone does care enough to put together a data set, I'll look at it.

[theckhd]

Will start testing new stuff as soon as the new build is up...

That's fine. For the ones that changed this patch, naked and set#7 will probably suffice just to re-confirm the bounds are close.

One exception: give me a full rundown on Judgment. Before, it still had hidden AP scaling. I want to see if it still does, or whether that was corrected.

[Klaudandus]

sounds good, i'll work on it as soon as it's physically possible for me to do so.

[Klaudandus]

Considering I can do judgment logs along with SoT, Censure, HotR and EF, I might as well do them all in a single shot for all levels of AP -- question is, would 15 minutes per AP suffice?

[theckhd]

Probably. It's not so much minutes as trials. For abilities that do a static amount of damage (SoT, J, Cons, etc.), one or two casts is enough. For abilities that have a damage range, we need enough casts to get good estimates of min/max damage. I can make do with mean values for this round, since we're mostly going to be confirming equations we already know ahead of time.