If: meleeDPS = Damage/BaseSpd*(hit%*1+crit%*2.54+glance%*.75)*(1+haste)*(1+flurry)*(1+IAS1)*(1+IAS2)*...*(1 +IASn) (Quoting Landsoul here)

How should i calculate glance % and quantify flurry?

I'm going to preface this response by saying I have done none of the heroic encounters you have listed. Whenever I put a "rage>=50" execute action(sans raid damage event) ahead of slam, cs, or bt, I get a minor dps loss. My guess is that with the raid damage event, your rage income is so high that the "rage>=50" that high up the priority list is doubling as a rage dump because HS can't handle it adequately. Can't run simcraft at work, makes my machine cry. Also, that much smooth rage income handily eliminates rage starvation. Do all those encounters feature such smooth damage in?

I think simcraft has been fixed where you can just specify a raid damage event and interval, and it will be correctly converted into additional RPS. If you literally specified 15k per second, you might want to change the interval, that could impact the severity of next-GCD starvation. Otherwise, you seem to know what you're doing, so I wouldn't second guess your math. I would like to see it when you get back though. There's a handy stat in the results tab called "Uptimes". If you click on that, it will show the percentage of the fight you were at the rage cap. Try taking the high RPS execute statement out and see how much rage cap you have. I'm betting it's a decent amount.

Outside of heroism and heavy continuous raid damage, good luck judging that not only is your average RPS high enough, so is your instantaneous RPS to warrant changing your priorities on the fly. For the dps gains you listed, doesn't seem worth it, unless the raid damage is just so egregious(Chimaeron). This reminds me of trying to "guess" when to use ferocious bite as a feral.

Well,it's true that outside those factors i stated (heavy raid dmg,heroism),it's hard to use execute in this way (atramedes for example).However heavy and continiuous raid dmg for the execute phases seems to be the rule rather than the exception for this raiding tier.

What is portfolio mix maximization/risk math? Can we see your results?

I'm finding that there is a definite break-even point in damage between execute and BT at a certain rage level. However, the problem is that rage is potential damage (like HS is pre 20%) in the form of variable execute cost-damage. Let's say your break-even point was 23 rage execute equivalent to a BT which in said formula also considered appropriate extra melee damage from IAS. Now consider that extra 3 rage you spent for the same damage. That's 3 rage as damage potential that you lose. Yes, the executioner stack gives you rage, but no where near the 3 rage you didn't save by hitting BT. In fact, the faster executioner stack only gave me about .6 rage that I didn't have before.

I tried adding a rage-weight factor into my equivalence formula which considered the cost difference between BT and said equated execute, and there was no solution. It went up to 250+ rage before it got tired of iterating. This equation was based on a a comparison between the first 2 GCDs when Exe was available, BT-Exe vs Exe-Exe. Maybe I'll compare the first 10 GCD's next.

Thoughts/comments?

Now that 4.1 is actually live, has anyone tried to compare the in-game results with those from SimCraft? That is, using the 'new' suggested Stat Priority of:

Strength -> Crit -> Hit -> Mastery -> Haste (TG)
Strength -> Crit -> Hit -> Haste -> Mastery (SMF)

I know it is dependent on individual gear set-up, but the above priorities are based on BiS and are simply a rough guide.

Being time short before last night's raid, I decided to hold off re-forging Mastery -> Hit until I had more time to do it properly, and surprisingly still saw an increase of around 1 - 2k DPS on most fights (on some even more). I'll hopefully have more time tonight to give the reforging a go (and will repost with the results), however I was curious to see if anyone had taken the leap from a Mastery favoured build into a Hit favoured build, and if so, had it been noticeably beneficial?

I know everyone has spent the past few weeks / months addressing the issue of Mastery v. Hit (or even Hit v. Haste v. Mastery), and it would be interesting to see how that has translated into actual in-game results.

For clarification, I was running with TG; I imagine the results might have been quite different with SMF.

Edit - Since I can't actually test 'in-game' from work, I ran my current build through SimCraft v. a reforged build (into Hit), and it only showed approximately 200DPS difference in favour of Hit. Will still try it in-game, but food for thought nonetheless..

Mastery Build - chardev 8 - WoW Cataclysm
Hit Build - chardev 8 - WoW Cataclysm

I believe we all saw a good > 5 % or so bump to our singletarget dps, and personally I went from a Mastery -> Haste reforge after an old Landsoul post regarding, at the time, current 4.1 sep values he had (presumably pre-Precision ptr change). Simc states this is today pretty much a non-upgrade in terms of stat weights, and yet I still had a good amount higher overall damage than prepatch anyway.

I come from an SMF standpoint and three different 10k iterations all show the same stat weight pretty much, with not far from the best gear attainable.

Notes and observations from these 30k iterations with 372 ilvl (Simc Version 410-1 Positive Deltas Only):

Personal SMF stat weights
Strength > Hit = Crit > Haste = Mastery.
Normalized SEP values were 1 > ~0,6 = ~0,6 > ~0,4 = ~0,4

What surprises me the most is the inherent strength of Hit, and how much stronger it is compared to Haste and Mastery (~+50%!), and how even it is to Crit. Only positive deltas were used, but if I recall correctly hit was valued higher than it should have been in Simc in earlier versions (I do not know if it got fixed).

Normalized hit SEP value was around 0,6, meaning gemming +10 str bonuses in blue sockets should be optimal (~+2 sep each bonus).

With these stat values, [License to Slay]. seem to surpass [Crushing Weight].



Note that all this above assumes inflated hit values in Simc got fixed.

A quick question for some of the pros. I haven't seen much discussion about the value of Weapon DPS in all the discussions so I don't know how to handle this. DPS is my offspec so it puts me into a weird situation.

Basically I have a pair of Soul Blades but i tank with a 372 Mace of Acrid Death. I'm wondering if due to the extra DPS on the weapon, if the Mace of Acrid Death would win out over a Soul Blade for DPS assuming that the Soul Blade has Landslide on it and the Mace of Acrid Death has Windwalk (useless) and the Parry is not reforged....

I'd assumed that the Mace would win out due to the extra strength and weapon DPS but I figured I would see about how to go about looking at the problem. I'd turn to Simcraft, but its been giving me really wonky results lately so I figured I would ask the real experts.

I'm gonna give Sim a try, Sim just seems a bit wonky for Fury though. I guess it all makes sense though since so many of our attacks are AP based and our weapon damage based attack ate it pretty hard over the last few hotfixes and the patch. I thought the additional strength would help cover some of the AP difference but I guess the uptime of landslide is good enough that it outweighs it in the end.

To sum up every thread regarding fury right now:

Question -
"Does anyone know if hit is better than haste now? Has anyone sim'd it? etc....."

Answer -
No. No one has really taken the time yet. All we have to go on is Aylene's post in this thread, and the fact that Landsoul is putting hit over haste on his own gear.

I personally tried both, a haste set up, and a hit set up, and hit was higher dps for me. It wasn't scientific by any means, so take it with a huge grain of salt, but just thought I would share.

Edit: I am SMF btw.

Has your rotation changed? With my higher DPS on my haste build over hit build I do the same thing, just prioritizing slam over RB. I'm SMF as well.

Back on this topic, and you discuss Markov Chain.
I am really not conviced by the approach of slotting time in n intervals, and letting n tends to infinity.

Why not use continuous time Markov Process ? I know that they assume that events occur occur independently, according to a Poisson process. But anyway, I guess that you assume this independence in your slotted time Markov model (or the transition matrix won't be nice at all).

The good point with continuous time MP is that you can either directly find the steady-state distribution (just find any steady-state distribution, and that's the one), or use the embedded Markov Chain (ie. the chain of the state after each transition). Both steady-state distribution are related, with weights being the mean sojourn time in each state (and then renormalization). You can find details easily on Internet (wikipedia for example: Continuous-time Markov process - Wikipedia, the free encyclopedia).

Now, to make things even more complicated, you have to decide in fact precisely which quantity you want to measure (based on game mechanism). Is it the proportion of time you spend with flurry up ? Or the proportion of swings you start with flurry up ? Both are not strictly equivalent (you have more time between swings when flurry is down). You then need to modify accordingly your MC / CTMP, or you need to use Palm theory to invert between both (but that will be even more tricky I guess). In short, if you want to consider time-proportion, then go with the CTMP "classical" route. If you are interested with swings proportion, my guess is that the easiest way is to consider the MC at swing times, and correctly write transition probabilities including yellow hits (either based on CTMP or directly).

The last point I don't know how to modelize without spending time on it is the fact that yellow hit themselves are not evenly distribution. They arrive at fixed (GCD) intervals, and then, they are also clustered (depending on whether RB is available or not, etc.). MC / CTMP with time varying (though stationary) transition matrices are a pain !