I thought that as we get closer to the expansion release it might be useful to have some more stat value data for level 90. I ran simulationcraft using the T14N profiles with stat values and had it graph a fairly large range for haste, mastery and crit for all 3 specs. You can see the report at: Simulationcraft Results

Pay particular attention to the graphs of how the stat values change for each spec- SV's stat values relative to each other seem to change sharply, whereas BM's are a lot more consistent. There's also what appears to be a break point in SV haste where it rises sharply in value.

I updated the link in my previous post with a new run which goes for 2400 stat value in either direction (in 160 stat increments). That should be a little more useful. There's a jump in haste for MM I notice too although at a much lower point than SV.

With regards to explosive trap in MM- I suspect shot selection is a big factor- I tested it in different gear sets and it was coming in as a gain, but since it's replacing arcane and saving focus it probably affects the timing of steadies which can impact free aimed shot procs and steady focus uptime. It definitely came out as a dps gain in simc though which should do a better job of evening out the effects of randomness.

Since you are already familiar with them, I was wondeirng if you could do a similar profile of T14H gear - the chart for stats for BM seem to corrospond with what I found through manual testing, but I find it a little curious that all 3 stats converge at the +/- 0 of all stats, and I'm wondering if that is a result of how the stats work.
Analytically, it makes sense for mastery to be better than crit at lower values (mastery giving 2% pet damage increase over crits 1% pet damage, + 1% hunter + dynamic effects), but I think haste is artificially inflated at the low end (due to latency cropping).
I'm also wondering if the initial(?) stat ranking (as I read it, it is the starting point) influences the results - I find it curious that haste is so linear (the jump at around +1400 is clearly a plateau being reached, likely CoS castt ime hitting the 2xCoS+2GCDs between each KC+delay (letting it rise sharply for the delay reduction benefit).

The stats all converge at 0 due to the nature of the analysis. At 0 stat modification for all stats, you are at the baseline DPS value for whatever is the value of each of the stats at that time. Then as you add or remove stats, you see the change in DPS from the baseline.

Also, it is not the relative values of the lines that matter but the slopes. Although the mastery line is higher than crit at modifications less than 0, crit is the better stat since you get more improvement per change. The reason that the mastery line is higher is because it affects your DPS less. Each loss or gain of mastery affects your DPS less than crit. So generally on this plot for the whole range:

crit > haste > mastery

although there is a haste jump. The haste jump appears to be at the 3933 haste rating value. Like for SV, that is the value for a tight rotation with 3 CoSs between a KC. Not really sure why the big jump occurs then since you rarely do 3 CoSs ever.

And yes, the initial stat set does influence the plot due to the interactions amongst the stats. An improvement would be to start with a good guess at the best stat mix and then do the evaluation from that. Obviously, since this graph shows that crit > haste > mastery, a starting stat point with more mastery than crit and haste combined probably is not the best stat point at which to do the analysis. I would suggest redoing it with the stats from the BM Max DPS case.

In an ideal world with plenty of time and resources, data would be generated for a 3D graph that incrementally varies each of the starting stat sets over the whole range of stats and plots the 3D surface so that we can see the overall maximums, local maximums, and the slopes around each to determine the sensitivity to errors for each maximum, but that type of analysis is a little overkill for a game.

Hmm... now that I have looked closely at all 3 specs, it appears that the haste jumps are all around 3900 haste rating. Here is what I see:

BM: Starting haste of 2175 plus about 11 ticks at 150 each = 3900
MM: Starting haste of 5325 minus about 9.5 ticks at 150 each = 3900
SV: Starting haste at 3933 right at the haste jump

I find it very odd that the haste jumps for all three specs occur at around 3900. I originally attributed this for SV to the fact that that is the haste value that has a 1.66s CoS allowing a tight ES cycles with 3 CoSs cast in it. This kind of makes sense, although SVs rarely cast 3 CoSs in a row now.

Then I saw it for BM, and initially wrote it off to the same phenomenom since BM's prime CD is 6s as well. However, not only do BM hunters almost never cast 3 CoSs between KCs, the attack speed of CoS is also affected by FF, which gets the CoS cast down to about 1.28s at this haste level.

Now that I see this same haste jump point for MM, I am very suspicious since MM behaves very differently than the other specs with needing to cast more SSs, having higher haste effects on them statically, and a much longer primary CD at 10s. SS cast time is at 1.45s at this haste level.

This haste jumps may be real for the math for the specs, but I am very suspicious that it may instead be do to some pecularity in the simulation since it is happening at the exact same value for all three specs despite their differences.

EDIT: I was thinking upon this more while getting ready for work. Then I realized the one commonality that all 3 specs have that is affected by haste - autoshot. This could be the relative haste level at which an extra autoshot is performed over this 5 min scenario. I do not have the time at the moment to analyze whether that is the case (have to get to work), but although it is a possibility, I find it unlikely since all 3 specs have a different average autoshot frequency where I would expect that additional autoshots would not be gained at the same haste rating.

Based on your earlier comments, I changed the default steps for stat scaling to 160 (based on the gem bonus amount). The "overkill" 3D model you are asking for can be produced using simc's reforge plot. Simc does not have the ability to render that plot, however, so we haven't generated one for hunters yet.

The two big commonalities across specs are pets and shared talent abilities. In a prior round of simc profile optimization, stampede was one of the key pivots. The stampede question was whether the stampede pet gets in the last basic attack before desummoning; that seems to depend on focus rate. To determine what's going on here, we run simc with same profile and high and low haste ratings: There are lots of ways to dig into the data. For this particular analysis, I paste the Raw Abilities section into Excel, sort by ability id (and secondarily dps), and subtract every other row (=IF(MOD(ROW(),2)=0,"",RC6-R[-1]C6)). Scanning down the difference column, the only two interesting differences are 798 for dire beast melee and 400 for arcane shot.

TL;DR the haste break point lets each dire beast get in an extra attack, resulting in a step for both damage and focus (leading to more arcane shots as well).

I did some tests with addition of explosive trap on BM rotation on Beta , standardized all my gear to ilvl 463 (heroic dungeon gear) with Crit>Mastery>Haste reforges, all enchanted and using [Tempestuous Longbow] as Orc:

My average explosive trap damage per trap is 25836 and my average Arcane shot NON CRIT damage is 20419. Well. on theory worth put explosive trap on rotation because the damage of trap itself is bigger than Arcane shot non crit.

I made some tests on Shattrath dummys using explosive trap and i had a decrease on DPS, but my damage done goes up, comparing with rotation without explosive trap.

I dont know with better weapons, because Arcane Shot are based on weapon damage, but if your explosive trap does more damage than your arcane shot non crit i think worth put it on rotation.

I usually cast explosive trap right after i cast cobra shot, because on cobra shot cast time i have time to prepare the trap cast.

Some trap facts:

- Dots have chance to crit based on your Crit.
- Does not trigger any of your trinket on cast (not even on dots ticks).

I don't know if haste affects the dot of explosive trap, if affect please let me know.

Thanks for analyzing this. Your result makes tons of sense.

I need to do a better switch of my brain to MoP from Cata. I had thought of all of the old impacts like number of shots between major CDs, extra autoshots, and extra hunter pet attacks, but none of those seemed to sufficiently explain the situation theoretically or from looking at the shot selection results. I totally over looked the possible impacts to the new abilities like Stampede and DB. In FD, I was just comparing the number of DB casts and did not even consider the number of attacks changing per DB cast.

I am so used to our abilities with multiple attacks not being affected by haste. We need to do something similar like what Keldion did for the caster DoTs and find the haste break points for additional Stampede, LR, and DB attacks under the following suggested conditions:

BM/MM/SV with no haste effects
*BM/MM/SV with raid buff
BM/MM/SV with BL
*BM/MM/SV with raid buff and BL
BM with 5 stacks of Frenzy
*BM with 5 stacks of Frenzy and raid buff
BM with 5 stacks of Frenzy and BL
*BM with 5 stacks of Frenzy, raid buff, and BL

I put an "*" in front of the critical ones since we should almost always have the raid haste buff, especially since it is so important to us and in our control.

I left off the haste increasing abilities of FF, RF, and SF since they affect your ranged haste only and do not affect pet melee attacks.

Any volunteers? I am too busy at the moment to do it myself.

Thanks for the clarifications. I had thought that the number of attacks from LR was affected by haste, but I may have misunderstood previous comments. I also didn't think about the fact that Frenzy was specific to the pet. Good point.

However, as far as haste points go, I was only looking to find the points where the new abilities get extra attacks similar to caster DoTs. I was not concerned about the focus benefits from haste effects to Stampede since the Stampede pets do not last that long and I assume that we are normally going to stack Stampede with RF anyway. Plus, adding in the focus impacts just complicate the analysis more. Furthermore, how well the Stampede pets performs depends on whether you are BM with GftT or not or whether you are MM with RR. Thus, analyzing Stampede is probably more trouble than its worth.

So the updated list would be just for DB under the following conditions:

BM/MM/SV with no haste effects
*BM/MM/SV with raid buff
BM/MM/SV with BL
*BM/MM/SV with raid buff and BL

However, even that analysis is probably not worth it since we already know about where the value is when casting DB outside of BL (about 3900 haste rating) and probably do not have another achievable milestone. The only question is what achievable haste rating will cause an extra attack during BL. Since we can at most get the effects of 1.67 DBs in during a BL and at worst just one DB, it probably is not worth analyzing, but it may be beneficial still to do so.

One thing this reminds me about is I'm actually not 100% sure on whether Go for the Throat and Invigoration work with our stampede pets. I think it's likely that Invigoration would work since the pet procs it, but Go for the Throat may not, as it's a transfer from hunter to pet, and I don't believe the stampede pets benefit from Fervor, which would work similarly. Given the long cooldown on stampede it's difficult to test this very well. Does anyone have any details on this based on their own testing?

Just a followup, I did some testing and verified what I expected- Go for the Throat only works for your active pet, Invigoration will proc from Stampede pets. I've updated my site to add the invigoration support.

It's not like having the extra focus is going to make you do less damage- some may just not get utilized. You probably will cap out on focus during that time, but rapid fire will also increase the damage of your stampede (due to the extra focus regen they receive). Holding back on rapid fire generally won't be worthwhile as you'll want to get all your high cooldown abilities used early on and use readiness to reset.