After seeing an interesting point made in the BiS thread that after capping exp and hit the stat weightings would be str > crit > mastery > hit (27%) > haste, and seeing this same point made a while ago , I decided to run a few experiments to determine the possible difference between stacking hit past the 8% cap or reforging for crit. To do this I conducted a series of 10 tests, of 10 million damage each on the heroic target dummies, once with a high hit build and again with a high crit build.

The spec I used was the standard fury raiding spec.

During the tests, no sunders were applied and no cooldowns were used. The standard rotation was followed throughout. I have the 2 set T11 bonus. For your benefit, only the averages of all the tests I conducted have been posted. If you wish for more in depth details to be published then by all means ask and I'll post them for you.


Hit Build Stats and Results Average

Hit - 17.10%
Crit - 14.27%
AP - 11145 (self buffed)
Haste - 3.4%
Mastery - 14.24 (8 + 6.24)


Bloodthirst - 28.9% of damage done - 3 million damage of total

Melee - 25.8% of damage done - 2.64 million damage of total

Raging Blow - 20.1% of damage done - 2.06 million damage of total

Heroic Strike - 13.5% of damage done - 1.38 million damage of total

Slam - 4.7% of damage done - 481k damage of total

Deep Wounds - 4.2% of damage done - 426k damage of total

Colossus Smash - 2.7% of damage done - 272k damage of total

Total average damage done - 10.26 million - 11586 DPS



Crit Build Stats and Results Average

Hit - 9.70%
Crit - 17.26%
AP - 11145 (self buffed)
Haste - 3.93%
Mastery - 15.83 (8 + 7.83)


Bloodthirst - 28.9% of damage done - 3 million damage of total

Melee - 26.8% of damage done - 2.78 million damage of total

Raging Blow - 21.3% of damage done - 2.2 million damage of total

Heroic Strike - 11.7% of damage done - 1.21 million damage of total

Deep Wounds - 4.7% of damage done - 482k damage of total

Slam - 3.5% of damage done - 358k damage of total

Colossus Smash - 3.1% of damage done - 324k damage of total


Total average damage done - 10.35 million - 11689 DPS


As you can see the high crit build is ahead, but only by a small amount. This leads me to believe that given the buffs to Mastery, Raging Blow and Bloodthirst on the PTRs, reforging for crit > mastery over hit both is and will be better for DPS.

I'm really surprised the stat differences worked out the way they did for the 4.0.6 build Landsoul suggested, and your numbers Mildcorma. I like most am stacking hit like no other, and this is a huge shift in distribution.

I'd also like to note if crit and mastery are so awesome for us, and we only need 8% hit...this makes our BiS pretty flexible; I could see builds with Liscence to Slay and the axes off Cho'gall also being high up there, along with SMF builds (one thing I found SMF to hold it back was just a lack of ratings - hard to get high hit %'s if you don't have 2 friggen weapons with like 200 hit rating)

Yes, for all the tests I only used heroic strike above 65 rage to ensure I was never short, however being human in the few cases I did make a mistake the amount of data collected should make these errors a non-issue.


I was also quite surprised, but then even this kind of parse is a relatively small data set and my gear is a very unique snapshot of the entire situation. If someone else were to run the same tests, I would not be surprised to see maybe not a large shift in distribution but certainly enough of one to validate hit over crit for example. I will be raiding with the high crit build next reset to see if there is any noticeable difference in a raiding environment. If it's not the best build for everyone now, then there is no doubt that it will be.

Tested Landsouls suggested 4.0.6 Crit/Mastery-Build (roughly 10% hit, 23% crit, 14 Mastery) today on Maloriak, Chimaeron, Nef and a few Conclave-HC tries.

The results were pretty solid but sometimes a little spiky and rng just like foretold by shha (if i remember correctly).
Rage gen didn't feel that much worse then with a high-hit build and overall my dmg wasn't lower, ofc "Fury of the Angerforge" isn't fixed yet and it was just a frickin pain when those stacks fell of.

I think i can get the time tomorrow to test the 2 builds on a dummy with Recount or Skada screenies up (your choice), cuz i don't really know which WoLogs-parses you can see on our Guilds site..



Edit: ah and i need to know if you want that tests with 3/3 DW or 3/3 Incite. I don't really see, as for now, why we should take 3/3 incite with the 4.0.6 HS-changes.

So, here are the results of my testing today. Live-Server 2 Tests:
One with a High-Crit/Mastery Build (had to copy&paste the stats because i forgot to blend it in when i made the shot.. :<)
ImageShack® - Online Photo and Video Hosting


Second Test was with a High-Hit Build:
ImageShack® - Online Photo and Video Hosting

Edit: didn't cleared out: used HS only with >60Rage or with Battle Trance / Incite Procs, DW on cooldown

Okay, but what's the point in testing those on live? If anything, you should be doing that on the PTR.
Until 4.0.6 rolls in on live, it seems pointless to me.

My tests were made in regards to MildComa's tests, because if i don't got it all wrong he did the same tests on Live.

But what you can see with those is that even on live the DPS is not that big of a difference. So with a fixed Fury of the Angerforge my used High Crit/Mastery Build would be ahead of the High-Hit Build, thats what i wanted to point out for MilComa's stated

Right, so I've done some training dummy sessions to compare live to the newest PTR (including slam buffs). All tests were done using Battle Shout only, No sunder armor on the dummy. Tested each 2 times. These are to be taken with a grain of salt, as I'm no expert when it comes to this stuff, and I'm simply sharing the results I got.

First off here's what I manage to get on live on the raider's training dummy:




Here's exactly the same thing, on the PTR:




Here's PTR, with hit reforged to mastery everywhere I could, while maintaining expertise and crit, I was under 26 by 8 expertise rating. Overall not a huge change, since there wasn't all that much I could reforge:




To sum it up. I'm doing higher dps on live then on PTR as of now. Not by a huge amount, but I am.

With the lowered buff to RB (120% instead of 145%) the mastery/crit test came out worse then the "avoid all mastery" ones. Unfortunately I can't get my hands on a "8% hit, rest into mastery/crit" gear so I have no idea how that would do.

Just to compare this with the previous PTR (145% RB, no Slam buffs, no screen-shots of that, sorry) going mastery/crit (same as this time) yelded me 12k dps, which was a very clear boost.

Also, the tooltip of Heroic Strike changed once again. I'm not really sure if they didn't ended up nerfing it again (60% AP down from 93% according to mmo-champion), or if that simply was a tooltip fix.

It seems to me all that Blizzard is doing is simply shifting the damage from HS to other abilities, not flat out buffing us. They never said they find Fury DPS too low as far as I remember.

It's just a tooltip fix(93% -> 60%), blue already commented on that.

Also, realize that the heroic strike nerf is 30.5%. The buff to just bloodthirst is tad under 30%. So, in order for you to be doing less dps on the PTR, your bloodthirst contribution would have to be at least lower than your heroic strike contribution on live. And since that's obviously not the case in either place(live or ptr), it must be something else.

Were you doing the same ability priority? Slam did not get buffed enough for it to push ahead of bloodthirst in the queue.

I'm still using the same priority. Slam was still the last in prio. HS was used a little more aggressively on live, as opposed to a last resort rage dump on PTR. I've had a slightly higher latency on the PTR so that might have screwed up the entire thing, but overall, it's pretty much the same as live dps wise.

I'll give PTR another try a bit later, the latency should be lower.

You are confused because this kind of thing is not a good experiment. As a comparative example, what you are looking at is similar to trying to compare morning highway commutes to work with different cars and attributing a more effective commute to the better car. Priorities and habits may be the same in different trials, but what is discounted are the facts that there are other drivers and inconsistent traffic, varying streetlight patterns, and even more marginal variables such as wind direction, temperature, and your alertness based on how well the driver slept the previous night.

This kind of research simply does not provide any reliable results at all to be able to come to solid conclusions. This kind of experimenting takes hours of operation and attention to produce any kind of reliable data to source conclusions from. And to be honest, this kind of thing is not even worth doing because (1) it takes so long and (2) figuring out what build is better (the purpose of the experiment) can be done much easier by knowing exactly the mechanics of each ability, which is already given.

You can't just simply do 5 minute dps trials with the same behavior but different gear and compare them. To be able to make any conclusions about something as marginal as secondary stat selection you must do at least dozens of these tests with extremely controlled behavior, and record them with miscellaneous notes such as: "rotation felt very smooth and correct and was consistent throughout the entirety of the test. Was able to stack X and Y cooldown at time Z for about T seconds." Each trial can be worlds different. Only certain trials are fair to be compared with each other. There are so many variables that throw off readable correlation. For example: different rage lulls at different times, different enrage and flurry uptimes, different positioning of enrage effects with relaiton to the RB cooldown, varying periods of rage surplus with relation to the CS cooldown and random trinket procs and landslide etc, noticable variance of successes of crits on varying abilities, and the list goes on.

A conclusion is only as good as the experiment performed to reach that conclusion.