Natchos is correct. I took all my gear into account so I could have a viable RSV. With that RSV I calculated how much dps those 2 items alone would give me.

I did not take the simulation into consideration but if that is so, then the Ikfirus + T10 Gloves would elevate the dps even more relatively to Aldriana + T10 chest as hit and expertise are much closer to the cap, thus not giving you many energy capping situations.

That's fair. At the same time, I don't think 2.2 vs 2 is 'much better' as a valuation.

The elephant in the room is that agility and mastery simply trump everything else. You can go for high crit rates and that's fine and it will help your DPS. You can go for hit/exp cap and that will also help your damage no matter what; even in cases where DPE is now less, energy regeneration is not the limiting factor any more. The difference between the value of one point of hit vs. one point of crit is the equivalent of the difference of 2 points of hit vs. one point of agility or mastery; at that level it doesn't really matter how you optimize as long as you optimize for agility and mastery first.

Now, if you can find a setup that optimizes crit in such a way that you are not crit capped and it provides significantly more crit while reforging an optimal setup of mastery, that's great; I've seen nothing to indicate otherwise. Again, I'm arguing that it's better to reforge the highest stat on your gear to mastery regardless of what it is. I'll redo the spreadsheet and see if that's the case later today, but I doubt it'll make that significant of a difference; when you're trading 9 points of mastery for the difference in value between 40 points of haste and 40 points of crit, you're going to lose on the crit side.

Please, stop taking the argument from me that I'm saying hit and expertise are amazing stats and must be capped. That is not what I'm saying at all. I'm making the case that it's optimal to cap them anyway because the alternative is having a ton of haste, and they are better in other aspects (such as more reforging stats and more sockets).

As to the formulation, it doesn't appear to have any strategy on when to refresh, which is important. If you do the above strategy such that you have time to refresh if you miss/dodge, you need at least 3 seconds on either rip or rake. That means that their uptime would be higher, but their DPE would be less; 20% less in the case of rake.

It still doesn't take into account things like energy capping, OoC procs, missed ticks on targets that are going to be gone quickly, etc. That is entirely correct.

Wrong. Since 4.0.1, refreshing after the penultimate tick adds the full new DoT duration to the end of the current DoT's final tick, so you do not lose ticks.

Tests with the sim have shown energy capping to be negligible. OoC procs are handled by the Formulation. No idea what you mean by your 3rd point.

There's a reason why Mew offers both Simulation and Formulation. Sure, the Simulation is more accurate than the Formulation, but it's been shown that their results are very close when calculating DPS and RSVs, so we are more than happy to use the Formulation model to make gearing recommendations. The advantages of Formulation will soon unveil itself with some new stuff Mew will offer soon which cannot realistically be done with Simulations.

Please, don't make me say what I never said.
Quoting from my first post on the subject (emphasis added):
If the game is getting the highest number on Mew, I have no doubt that hit/exp to zero is the way to go (and BTW this is exacly what I did on beta at 85, with mastery available on the stuff, hit/exp are rare). But at times I get the feeling that people forget that a simulation is an idealized case, which may (or may not) translate to reality. This is not as irrelevant as it may seem: unless the deviation from the result of the simulation and the values on recount is small (as in a few %), then there are very good reasons to doubt of the simulation's exactness.
Of course I understand using the scaling factors from a simulation as a "best approximation", but as Felhoof said, "I don't think 2.2 vs 2 is 'much better' as a valuation.". In particular when it comes to hit/exp, which compared to other stats can really mess up with the "cycle".

I stand by my initial statement: I'd love to see a 0-hit 0-exp combat log which shows a value within a few % of the Mew result. Then I'll take simulation results and swear by them (extra kudoes if it's not on a training dummy, but something like Saurfang or Festergut). I've never been able to reach simulation values even with hit/exp cap, but I know that I'm not an excellent player.

(BTW I reran Mew and I don't see the standard deviation anywhere, am I to reverse-engineer it from the +- error from the mean?)

So I've been reading this discussion and had some thoughts with respect to the whole "value of not missing" topic.

It seems as though the argument can be boiled down to:

a) The theorycrafting tools (I'll just use "simulator" from now on) show that hit and expertise are not very valuable, compared to their alternatives.

b) Various concerns have been raised about the "idealized" environment within which the simulator runs. Some of the concerns (distribution of dps) seem to have been addressed, but others (short-duration fights, etc.) have not.

So what if we could simulate the less idealized fights also? It strikes me that there are five main fight types:

1) Straight aoe (multiple relatively low-health mobs clustered together)

2) Single short fight, where the mob has relatively low health (e.g., fight lasts 10-30 seconds)

3) Sequence of short fights, where we sequentially churn through relatively low-health mobs one at a time.

4) On/off boss fight, where we dps on a high-health boss, but a certain percentage of the fight is spent outside of melee range.

5) The good ol' Patchwerk/Saurfang style fight.

It seems feasible to be able put together a profile built from these fight types that more accurately describes the environment in which we need to perform (e.g., 5% aoe, 5% short-single, 20% short-seq, 40% long-gaps, 30% long-steady. Contexts (1) - (4) should be as easy to simulate as (5) and it would be trivial to combine them to get an aggregate score.

Some of the considerations I can expect might emerge:

* Bleeds might be even more important for punctuated "in and out" fights

* Hit/exp might have higher importance for shorter fights, but the capping requirements for those stats are likely to be lower (e.g., maybe its really important to hit/exp cap for level 82, but not so much 83)

* Different rotations (priorities) might be appropriate for the different scenarios (e.g., swipe spam for aoe, rake but not rip for shorter fights, etc.)

* etc.

Thoughts?

I personally use Ovale as the move suggester for all my dps and tanking toons. The great thing about Ovale is that you can program your own logic - combining current theorycraft with your own playstyle. For example, on my ret pally I suggest a Word of Glory rather than a Templar's Verdict if my health is <50%. And the default script is usually a great starting point if you are learning a new class or spec.

I have just finished analyzing Ikfirus+T10 Gloves vs. Aldriana+T10 Chest with a BiS setup.

First of all, this is the list that I consider BiS:

Every reforge made was for Mastery.

Head	[Sanctified Lasherweave Headguard] Reforged Haste
Neck	[Sindragosa's Cruel Claw] Reforged Haste
Shoulders	[Sanctified Lasherweave Shoulderpads] Reforged Haste
Cloak	[Shadowvault Slayer's Cloak] Reforged Haste
Chest	[Ikfirus's Sack of Wonder] Reforged Expertise / [Sanctified Lasherweave Raiment] Reforged Haste
Bracers	[Umbrage Armbands] Reforged Haste
Gloves	[Sanctified Lasherweave Handgrips] Reforged Haste / [Aldriana's Gloves of Secrecy] Reforged Haste
Belt	[Astrylian's Sutured Cinch] Reforged Haste
Legs	[Sanctified Lasherweave Legguards] Reforged Crit
Boots	[Frostbitten Fur Boots] Reforged Haste
Ring1	[Ashen Band of Endless Vengeance] Reforged Crit
Ring2	[Signet of Twilight] Reforged Crit
Trinket1	[Sharpened Twilight Scale] Reforged Crit
Trinket2	[Death's Choice]
Weapon	[Oathbinder, Charge of the Ranger-General] Reforged Haste
Idol	[Idol of the Crying Moon] Reforged Haste

Gems used:
Agility/Crit on the Yellow sockets of the T10 chest, Aldriana's Gloves of Secrecy and T10 Legs.
1x Nightmare Tear on Astrylian's Sutured Cinch's blue socket.
Agility on the rest.

As professions I took Jewelcrafting (+42 agility) and Engineering as the 23 agility+parachute tinker on the cloak can stack with the 22 agility enchant.

As weapon enchant I chose Mongoose as it came out on top on both the formulation and simulation with a 20 dps increase with Ikfirus + T10 Gloves and a 7 dps increase with Aldriana + T10 Chest.

In BiS gear as far as RSV goes, I've noticed that crit and haste keep on changing position depending on the reforge. To save a relatively large amount of time I've reforged haste everytime the choice was haste or crit as the dps loss wouldn't be as big. I would recommend reforging crit on pieces such as Astrylian's Sutured Cinch as it has 16 more crit than it has haste.

The stats for BiS with Ikfirus+T10 Gloves after reforging are:

Weapon DPS: 347.8
Agility: 2459
Strength: 114
AP: 889
Crit: 1145
Haste: 470
Mastery: 520
Expertise: 170
Hit: 230

The stats for BiS with Aldriana+T10 Chest after reforging are:

Weapon DPS: 347.8
Agility: 2455
Strength: 114
AP: 889
Crit: 1287
Haste: 529
Mastery: 517
Expertise: 106
Hit: 116

Finally, here's the table that I've come up with.



What's your thoughts on it? Seems like in BiS or not, Ikfirus + T10 Gloves wins.

Are you sure the ashen band of endless vengeance is BiS? I'm getting higher values for frostbrood saphire ring as well as band of the bone colossus.

Furial, that's what I am aiming for, although probably use heroic DBW instead of DC just for availability and ease issues.

You misunderstand...again. Sigh.

The ideal solution for 100% uptime on rip and rake is to always refresh 1 GCD before their end. This ensures they do not drop off ever and accounts for latency causing any issues. If you have hit/exp caps, you can do this reliably. If you do not, you may miss, which would cause you to miss refreshing until they've expired and potentially lose time to redo that attack. Which means instead of reapplying at 1 GCD, you need to reapply at 2 GCDs before they expire to compensate for this and potentially redo another application - which means that they are less DPE efficient.

The points were energy capping, OoC procs and missed ticks on targets that disappear soon. I'll explain again.

I don't know what tests you've done with the sim to show energy capping is negligible, but any time you don't hit your target and wait a GCD that's 1.5*energy regen time that you'll gain in energy. Does this cap energy all the time? My personal testing has shown it to happen quite often with bloodlust or when the subject of berserk, where you can literally not get rid of energy fast enough at times. I do agree this is likely not particularly huge, but it is something to consider.

On OoC procs - this comes into play as a wasted proc when you miss. Does Mew deal with that?

On low-duration times with rip/rakes, I'll try and explain a third time. Let's assume a target is going to live for 20 seconds. If you mangle at 1.0 seconds and rake at 2.0 seconds, you will get a tick at 5, 8, and 11 seconds. Refresh that, and you'll get it at 14, 17, and 20. Rake has caused a total of 6 ticks on the target. If you miss that target and have to rake at 3.0 seconds, you have lost an entire tick of rake - or 16% of your rake damage. Actually, that might be an interesting way to look at it; just run a sim with a 20-second fight time and see what happens.

Furial, it's much better to take blacksmithing as a profession for the extra bonus compared to engineering; while you do gain more agility via the gloves, you lose out on those extra sockets and only gain a haste proc (which isn't that great) in return. BS gives +40 agility straight up.

A) Cat's GCD is 1sec, not 1.5.
B) The ideal time to attempt to reapply Rip/Rake is any time between 1 and 2sec before it falls off, specifically because that gives you time to recast them if they miss, and reapplying any time between the last two ticks produces the exact same uptime and tick frequency.

I cannot think of any targets that live 30 seconds or less and at the same time are level 83. I would actually go for the lvl82 caps as I need to make sure I can reliably nuke these adds, while not giving up on too many raw DPS stats.