Hit problem is fixed, dumb programming error (used max instead of min for capping the hit at 99%), the trinkets are fixed now too (modelled sextant and silver crescent).
The unrelenting Storms ring is a funny problem, which is actually way deeper. The model is set to 600s, the optimal gear with extreme buffs (without sp) actually allowed the shaman to spam lb for 600s without break and that unrelenting storms ring (the crit) made the differnce. That's why it was that "good". This is a inheritant problem of the modelling system. The easiest way to bypass it is to either deselect some buffs (like mana potions) or add the Shadowpriest MP5 buff (this might change the equipment fairly much though, since without mana problems it will stack haste).

About the equivalence values, these are automatically calculated and read only (fixed that), negative values for mana dependent fights usually show that this attribute might actually lower your output due to the limited resource mana. These values are only for comparison (or internal use), it's not a feature ready for release tbh, since it simply does compare each used attribute in the model with every other used attribute and compares dps increase.

The resulting equipment without sp (with sp it uses some new spellhaste chest instead of t6) should be ok now.

That is just model specific, if you consider a 10 minutes fight the ideal gear is nuking without regeneration, that's why it uses mp5 gems and also the netherstrike belt. That's why you can also select SP as a buff and get "I don't care about fight length" gear. Sooner or later there will be some config options to set fight length etc. but as for now this is rather a feature than a bug. As for gear optimization for a nuker I would always consider the proper group composition, e.g. activating the SP.
The MP5 value displayed in the attributes is fully buffed incl. Mana potion usage (aka 100MP5),

Class: Druid
Build: Feral Combat
Cast/Skill rotation:

Typical Feral DPS rotation is:
* Wait for 80+ energy (skip this on initial contact)
* Rip (skip this on initial contact)
* Mangle
* Shred to 4 or 5 cp
* Repeat

Notes:

The Feral DPS rotation has a few variable factors, and this will affect the actual rotation and how it is carried out. Following values are assuming most common talents.

Rip: 30 energy cost, DoT finisher that requires combo points (cp), lasts 12 seconds
Mangle: 40 energy cost, non-positional instant special that generates 1cp
Shred: 42 energy cost, positional instant special that generates 1cp
Ferocious Bite: 35 energy cost, instant finisher that requires cp, drains all energy beyond the initial requirement of 35 and converts it to damage (used only in special circumstances)

Mangle has a debuff attached to it, that makes the target take 30% extra damage from bleeds and Shred. Therefore, it is applied immediately following the Rip from the previous cycle (before said Rip has had a chance to tick). It is also applied with a decent bank of stored energy (80+ minus the Rip cost) so you can fit in as many Shreds as possible after its application. The goal here is to minimize Mangle usage (1 per cycle ideally), as it is not as energy efficient or time efficient a method of delivering damage as Shred is.

All viable Feral builds will include a talent that gives 2 cp instead of 1 if an instant special crits. Therefore, depending on how many of the instant special attacks in any given cycle crit, and whether Omen of Clarity procs, the Druid might opt to go for a 4cp Rip instead of 5cp. Rip scales with attack power, and the scaling is same for 4cp and 5cp Rips (just the base damage changes), so 4cp Rips are quite usable if you're unable to generate 5cp for any cycle.

Of note is the fact that a finisher that fails to connect (miss, dodge, parry) consumes all energy, whereas cp generating instant specials that fail to connect will refund 80% of the energy, they are commonly modeled as "costing" 8 energy and 1 GCD (GCD for Cats is 1 second).

Ferocious Bite is only ever used if: a) the mob is bleed immune (Hydross, VR and the like) AND b) the Druid has < 42 energy AND c) Ferocious Bite will outdamage their Shred. In cases where c) is unlikely, the Druid will just continue to Shred instead of using a finisher (thereby wasting cp). Ferocious Bite also scales with attack power.

I assume you'll look up the values, scaling etc yourself =x

Finally, powershifting. Powershifting is a technique Druids use to exploit the Furor talent. Furor (at 5/5) has a 100% chance to refund 40 energy every time the Druid shifts to Cat Form. So, all the Druid needs to do is shift to Caster Form, and then back to Cat Form for a free 40 energy. Post patch 2.3, shifting to Caster Form is a client side check, so it's possible (using macros, or the autounshift flag) to powershift literally instantly, shifting to Caster Form incurs no GCD, but shifting back to Cat Form does.

In an actual combat cycle powershifting could be utilized at any time the Druid has very low energy (to minimize energy loss, we lose all energy when shifting to caster form). Since patch 2.3, I'm pretty sure worst case = average case = best case = a net gain of (40 - x) energy, where x is the energy you had prior to the powershift.

Powershifting is also limited by the Druid's mana, I think a typical raid buffed Druid has ~6k mana, and each shift into Cat Form requires a hefty mana investment. There is a talent that reduces shapeshifting cost by up to 30% but it's not common.

Important talents/skills:

See the "Talent point distribution" section.

Set bonus:

2T4 is the big one for Cats, 4% chance to generate 20 energy. This works on autoattack and specials (pretty sure it procs on finishers). This adds another layer of variability to the model. It is commonly believed the power of this set bonus can last you well into tier 6 content, where it becomes superceded by the 4T6 bonus.

4T5 is +75 damage to Shred, which is not inherently terrible, it's just that you need to give up 2T4 for it, which is a poor trade. You can model is like the Everbloom Idol (below) but more as a proof of "don't go 4T5 over 2T4 for DPS" than anything.

2T6 is -5 Mangle cost, which is also ok, it'll save you a bit of energy in the long term, but the fact that Mangle is used only once per cycle keeps it grounded.

4T6 is +15% Rip damage, and is considered the reason to drop 2T4.

Some Druids may choose to drop 2T4 earlier as it adds an element of variability and uncertainty in the DPS cycle, and which may result in wasted procs (e.g. having it proc right as you powershift = wasted, having it proc when you go above 80 while waiting for the next cycle = wasted).

Additional proc items:

All the common melee proc trinkets would apply here, Hourglass of the Unraveller, Darkmoon Card: Crusade, Tsunami Talisman, Madness of the Betrayer, Dragonspine Trophy; as well as the clickies, Bloodlust Brooch, Crystalforged Trinket etc. Our SSC trinket, Living Root of the Wildheart, is terrible.

We don't get proc enchants (Crusader, Mongoose, Executioner), or Windfury totem.

Putting Idols in here, the 3 worth considering:

Everbloom Idol is +88 Shred damage, and is still the best personal DPS Idol. It's estimated that +88 Shred damage is equivalent to adding ~450 or so (?) attack power (to Shred alone of course), so it's quite nice.

Idol of the Raven Goddess is +0.4% crit to everyone in the party (via the LotP aura). In a full melee group, the increase in raid DPS is estimated to be roughly equivalent to the personal DPS increase you get from using Everbloom Idol.

Idol of Terror is 65 Agi for 10s, 85% proc chance on Mangle. Once again, due to Mangle being used minimally, this loses value for DPS. Not nearly as good as Everbloom Idol.

Talent point distribution:
x/4x/1x is pretty much essential. You need 40+ in Feral for Mangle, you need 10+ in Resto for Omen of Clarity. I believe Talent Calculator - World of Warcraft is a good template (as posted in the Feral Druid Megathread) to begin from, and contains all the vital talents. The rest are more preferential than anything.

http://elitistjerks.com/f31/t16902-f...id_megathread/ for reference. Good luck with the project !

Druids, pick apart my post to see if you can't find any ommission, oddness or inaccuracy, cheers.

Regarding melee dps trinket procs, it currently says that for a warrior, tsunami talisman was better than madness of the betrayer. After a glance at your modelling for the relevant trinkets, I've found that some of them seem to be a bit out of line.
DST really procs more than once every minute, since its internal cooldown is only about 25 seconds. 1.5 ppm seems more accurate to me, which averages out to 81.25 passive haste rating. Yes, even after the nerfs it's still the best trinket in game.

The Tsunami Talisman on the other hand seems to be a little overvalued here. With 45 seconds cooldown it procs only like once per minute, so the 75.6 AP are definitely too high. 56.67 average should be more in line with its actual proc rate.

Hi there,

I have attempted to solve the same problem (though didnt expect to scope all classes just enhancement). IMHO the math model for enhancement (cant tell for other classes) is very complicated.

What makes it complicated? How the wf cooldown will interact for both weapons when DW, specially when you start combining flurry charges, different weapon speeds, haste procs, big etc. Not saying it cant be modeled, just saying that its hard as hell and therefore hard to verify its correct.

Imho the best approach to this is a brute force approach, basically setup the demo character and force combat for several minutes (to force averages on long term). Why i am saying this? well mainly and most importantly because we know most if not all the factors in play for hit table formulas so simulating actual combat for long periods of time becomes trivial. It also becomes very hard to mess up, and if we do its fairly easy to track the mistake.

I made a preliminary version of this program in C# and are also suffering from performance issues though it is kinda expected, it currently takes me 1 minute to average to 2 dps error but it can certainly be improved by a ton, its been a while since i did critical c++ apps but i am sure i can do it. However since only like 3-4 ppl i know actually tried and showed interest in it, i never did a version for 2.3 nor did i add the new enchants, etc.

I was also attracted to the idea of building it in C# for an easy way to make an asp.net web interface alternative for the ppl too scared of keyloggers to run a standard application.

If this approach interests you, send me a pm, i can lend you a hand.

Thanks for the info, i will change that in the source


It is set to 250 MP5, which is the Mana regen for a 1000 DPS Shadow Priest.



You might want to post that into the Enhance Shaman class thread, my program needs a closed form expression, since it is basically evaluating several thousand (~20000) different equips per second to find the optimum.
My closed form expression of the enh shaman is fairly good atm, <5% deviation from ingame tests.

I noticed a few items missing on the Elemental Shaman list.
- Blessed Elunite Coverings
- Belt of Blasting
- Avalanche Leggings

-Gems
Chaotic Skyfire Diamond

The Avalanche Leggings are Mail, however, they are from ZA, Lynx boss.

Cool program, a couple of missing buffs that I use are:
Totem of Wrath
Inspiring Presence

I'll be uploading a new version, with many new ZA items and weapon expertise items, including the first Holy Pala model.

Molten Armor is enabled currently.