Theck's MATLAB thread - Cataclysm/4.x

[theckhd]

If my memory of Blade Ward parses is correct, the WoL entry for a stacking refresh included the number of stacks. It was something like:
Theck's Windwalk is refreshed.
Theck gains Windwalk(2).

Or some combination. I could be wrong though. We might have to wait until we get some live parses to know for sure.

[tlitp]

Random parse (Dec 4th) :

Code: Select all

[00:05:08.374] Tssmagic's Blade Warding fades
[00:05:43.156] Tssmagic gains Blade Warding from Tssmagic
[00:05:44.903] Tssmagic gains Blade Warding (2) from Tssmagic
[00:05:54.813] Tssmagic's Blade Warding fades
[00:05:55.737] Tssmagic gains Blade Warding from Tssmagic

That's precisely how WW used to work on beta. The second application resets the duration of the first stack.

[theckhd]

Not according to my old parses:

Code: Select all

[15:23:58.575] Theckpretwo's Windwalk fades
[15:24:29.600] Theckpretwo gains Windwalk from Theckpretwo
[15:24:32.989] Theckpretwo's Windwalk is refreshed by Theckpretwo
[15:24:43.021] Theckpretwo's Windwalk fades
[15:24:43.258] Theckpretwo gains Windwalk from Theckpretwo
[15:24:53.256] Theckpretwo's Windwalk fades
[15:24:53.443] Theckpretwo gains Windwalk from Theckpretwo
[15:25:03.474] Theckpretwo's Windwalk fades
[15:26:56.346] Theckpretwo gains Windwalk from Theckpretwo
[15:26:59.769] Theckpretwo's Windwalk is refreshed by Theckpretwo
[15:27:09.767] Theckpretwo's Windwalk fades
[15:28:18.277] Theckpretwo gains Windwalk from Theckpretwo
[15:28:28.275] Theckpretwo's Windwalk fades
[15:30:48.429] Theckpretwo gains Windwalk from Theckpretwo
[15:30:51.785] Theckpretwo's Windwalk is refreshed by Theckpretwo
[15:31:01.817] Theckpretwo's Windwalk fades
[15:32:54.805] Theckpretwo gains Windwalk from Theckpretwo
[15:32:58.195] Theckpretwo's Windwalk is refreshed by Theckpretwo
[15:33:05.007] Theckpretwo's Windwalk is refreshed by Theckpretwo
[15:33:15.039] Theckpretwo's Windwalk fades
[15:34:13.297] Theckpretwo gains Windwalk from Theckpretwo
[15:34:23.228] Theckpretwo's Windwalk fades
[15:35:45.411] Theckpretwo gains Windwalk from Theckpretwo

Those all show a "refresh" event rather than a "add stack" event type, like the Blade Ward one shows. It's also telling that WoL treats these as fulltype 401 (SPELL_AURA_APPLIED) but not as 402 (SPELL_AURA_APPLIED_DOSE). Blade Ward shows up for both fulltypes.

[Treck]

I dont know where else to post it, but im a wonder if anyone else has noticed that when changing towards the Glyph of Divine Protection, it lowers the tooltip of GoAK to 30% dmgreduction.
From a very brief testing it looked like GoAK suddenly reduced less dmg, anyone who can help confirm this?
Or maybe its suppose to be that way might have missed it.
Or maybe it even gives GoAK unintentially -20% physical +20% magical? (Doubt that tho)

[theckhd]

This sort of thing should go in the Call to Arms thread (since you asked).

To test this, we'd ideally want something that hits for a fixed amount of damage. Easiest way is to do this:

1) Find another paladin. Have him use SoT and a weak, fast weapon (Dalaran Sword ideal). Start recording the combat log.
2) Duel him, have him simply auto-attack you and stack Censure to 5.
3) After about 10-15 seconds of taking Censure damage, pop GAnK. After GAnK wears off, wait about 5 seconds and pop Divine Protection.
4) Wait for cooldowns to come up, and repeat step 3 with the DP glyph active (or inactive if you already had it active in step 3).
5) Upload the combat log (either here or to WoL) and let us analyze it.

Should take no longer than 10 minutes to do all of this, if anyone has the time and a paladin friend.

[kristoferpally]

Has a lvl 85 range been established for keeping dodge and parry within each other? Right now I have some dodge reforged into parry and am sitting at 12.56% and 11.21% dodge/parry. Is this too much or too little?

[theckhd]

You'd want parry and dodge to be fairly close to one another. The exact equivalence is that parry should be about 1.0295% higher than dodge, at which point 10 points of dodge gives exactly the same amount of avoidance as 10 points of parry.

[Incite.]

Moved to another thread.

[tlitp]

Finally, some code snippets to assess Hurricane (regular 3/9 with CS/SotR, full time-on-target) :

Code: Select all

hu.ppc=gear.swing./60;hu.spc=0.15;hu.pd=12;hu.sicd=45;
htrack=zeros(1,4);
hstore=gear.haste;
for i=1:5
gear.haste=hstore+450.*sum(htrack(i,2:3));
stat_model;ability_model;
hu.q(1)=(1-hu.ppc.*mdf.mehit).^(hu.pd./player.wswing) ...            %AA
    .*(1-hu.ppc.*mdf.mehit).^(hu.pd.*3./(9+1.5.*rot.xtragcd)) ... %CS
    .*(1-hu.ppc).^(hu.pd./(9+1.5.*rot.xtragcd)) ...               %SotR
    .*(1-hu.ppc.*mdf.rahit).^(hu.pd./(9+1.5.*rot.xtragcd)) ...    %J
    .*(1-hu.ppc.*mdf.rahit).^(hu.pd.*P.CS./(9+1.5.*rot.xtragcd)); %AS
hu.p(1)=1-hu.q(1);
hu.q(2)=(1-hu.spc).^(hu.pd./cens.NetTick) ...
    .*(1-hu.spc.*mdf.sphit).^(hu.pd.*max([0.75-P.CS;zeros(size(mdf.mehit))])./(9+1.5.*rot.xtragcd));
hu.p(2)=1-hu.q(2);
hu.tm=zeros(4,4);
hu.tm(1,1)=hu.q(1).*hu.q(2);hu.tm(1,2)=hu.p(1).*hu.q(2);hu.tm(1,3)=hu.p(2).*hu.q(1);hu.tm(1,4)=hu.p(1).*hu.p(2);
hu.tm(2,2)=1;
hu.tm(3,3)=hu.q(1);hu.tm(3,4)=hu.p(1);
hu.tm(4,4)=1;
hu.dd=hu.tm^100;hu.upt=hu.dd(1,:);
hu.upt=(hu.upt.*hu.pd+[hu.q(1) hu.p(1) 0 0].*hu.sicd)./(hu.pd+hu.sicd);
htrack=[htrack;hu.upt];
gear.haste=hstore+900.*htrack(i,4);
stat_model;ability_model;
hu.q(1)=(1-hu.ppc.*mdf.mehit).^(hu.pd./player.wswing) ...            %AA
    .*(1-hu.ppc.*mdf.mehit).^(hu.pd.*3./(9+1.5.*rot.xtragcd)) ... %CS
    .*(1-hu.ppc).^(hu.pd./(9+1.5.*rot.xtragcd)) ...               %SotR
    .*(1-hu.ppc.*mdf.rahit).^(hu.pd./(9+1.5.*rot.xtragcd)) ...    %J
    .*(1-hu.ppc.*mdf.rahit).^(hu.pd.*P.CS./(9+1.5.*rot.xtragcd)); %AS
hu.p(1)=1-hu.q(1);
hu.q(2)=(1-hu.spc).^(hu.pd./cens.NetTick) ...
    .*(1-hu.spc.*mdf.sphit).^(hu.pd.*max([0.75-P.CS;zeros(size(mdf.mehit))])./(9+1.5.*rot.xtragcd));
hu.p(2)=1-hu.q(2);
hu.tm=zeros(4,4);
hu.tm(1,1)=hu.q(1).*hu.q(2);hu.tm(1,2)=hu.p(1).*hu.q(2);hu.tm(1,3)=hu.p(2).*hu.q(1);hu.tm(1,4)=hu.p(1).*hu.p(2);
hu.tm(2,2)=1;
hu.tm(3,3)=hu.q(1);hu.tm(3,4)=hu.p(1);
hu.tm(4,4)=1;
hu.dd=hu.tm^100;hu.upt=hu.dd(1,:);
hu.upt=(hu.upt.*hu.pd+[hu.q(1) hu.p(1) 0 0].*hu.sicd)./(hu.pd+hu.sicd);
htrack(i+1,:)=(hu.upt.*hu.pd+htrack(i+1,:).*hu.sicd)./(hu.pd+hu.sicd);
end
hu.dps=[];
for j=0:450:900
gear.haste=hstore+j;
stat_model;ability_model;rotation_model;
hu.dps=[hu.dps rot.totdps];
end
hu.dps=hu.dps(1).*(htrack(6,1)-1)+hu.dps(2).*sum(htrack(6,2:3))+hu.dps(3).*htrack(6,4);
[htrack(6,:) hu.dps]

Unlike Av, both LS/Hu do scale with weapon ilvl. Moreover, Hu scales heavily with Vengeance (whilst LS/Av are Vengeance-invariant). Let's see how Hu performs in the most favorable scenario (raid gear set with an ilvl 372 weapon, Veng=1.0) :

Code: Select all

ph=physical hit (%)
exp=expertise (skill)

               LS.dps   Av.dps   Hu.dps   Hu.dps@Veng=0.3
ph=0,exp=0        176      146       78       46
ph=8,exp=0        210      158       86       50
ph=8,exp=26       256      175       99       58

As expected, a lower Vengeance value absolutely butchers Hurricane's effectiveness.


Good grace, why is Hurricane that bad for Protection ? Well, there are some key differences between Ret/Hu and Prot/Hu :

• Ret uses slower weapons (2H versus 1H), and (generally) higher hit/exp values; both factors contribute to a higher uptime.
• haste has interesting side effects for Ret (AoW, SoB, HoL), boosting abilities with way higher DPET than regular autoattacks; for Prot it only has minor interactions with Reckoning.

[towelliee]

Anyone look at the new weapon enchants? Landslide procs like every other second. It is amazing.

[theckhd]

Anyone look at the new weapon enchants? Landslide procs like every other second. It is amazing.

I'd love to see a long auto-attack parse with Landslide to confirm its mechanics, if you have access to the enchant. A slow 2-hander would be better, but given how expensive the enchant is right now I'll take what I can get. I'm hoping to finish off the enchant comparison this week, now that the holiday blitz is calming down.

[tlitp]

Random parse (Dec 26th) :

• 383 triggers, counting only the connected ones (179 AA/105 CS/70 TV/29 J); 3.6 base swing time
• expected procs : mean 22.98, stdev 4.6477
• observed procs : 21

"Landslide procs like every other second."

[theothersteve7]

Am I reading this right? I see 1000 AP with 40% uptime. Compared to 600 dodge with 13% uptime for Windwalk. Even on a tank, I'm not sure I'd trade 400 AP for 100 dodge. That's double the itemization points.

[theckhd]

Am I reading this right? I see 1000 AP with 40% uptime. Compared to 600 dodge with 13% uptime for Windwalk. Even on a tank, I'm not sure I'd trade 400 AP for 100 dodge. That's double the itemization points.

Yes, you're reading it right. I'd still trade 400 AP for 100 dodge in current content though.

[theothersteve7]

Yeah, I guess you're right, the dodge is better, but it's reasonably close. Avalanche looks very nice on low vengeance levels, particularly. It's comparable to the dodge's weight in expertise, at least. Though I suppose I'd prefer expertise for the reliability.

I'm a little disappointed, that would have been a pretty funny discovery.