I still don't get how you would gain more cycles out of the same amount of energy. Whether you stretch Shallow/Moderate or advance immediately, it still takes the same number of builders. And as you said, Deep Insight is fixed anyway. All I suggest is essentially cutting the time spent in No Insight in favor of time spent in Shallow/Moderate Insight.

yes, the same amount of energy and abilities over the course of the fight, so let's say 50 SS and RvS in the whole fight.

The guy who pools his energy before deep insight does a higher percentage of those during BG30 where they do NOT advance insight. Say he does 20 in BG30 and 30 advancing ones.

The guy who doesn't pool only does 17 during BG30 because he's energy starved every time he hits that state so he ends up doing 33 advancing ones. so he has advanced insight 3 more times than the guy who pooled. Of course, those 3 abilities in question all hit 30% softer than the pooling guy.

So the question is which guy benefits more? the guy who has a higher percentage of his abilities hitting 30% harder, or the guy who is advancing his insight levels a little faster?

I'm pretty sure it's mostly a wash, actually, but if anyone is feeling ambitious then SimC would be the most appropriate tool for this. If you're willing to try and figure out the logic for the priority list that is.

My instinct is that maximizing Deep Insight uptime is probably the way to go, purely because combo point generators are such a small portion of damage done. I mean, making a quick survey of high-end parses on Gara'jal, it looks like the typical Combat rogue gets of order 15% of their damage via Sinister Strike and Revealing Strike - maybe as much as 17% on some parses, but certainly not much more than that. When you factor in DP and MG procs, you might make it up to 20% or so. And taking a hit to 20% of your damage to boost the other 80% seems pretty likely to be worthwhile.

To attempt to quantify this effect: lets assume we're regenerating 16 energy per second. It takes 4 hits to advance a level, which, at 40 energy apiece, means 160 energy. Thus, the "base" cycle is 10 seconds per insight level followed by 15 seconds at deep insight. Total cycle length is thus 45 seconds, and gets an average damage bonus of 10 * 0 + 10 * 10 + 10 * 20 + 30 * 15 = 750 / 45 = 16.7%.

Alternatively, if we save 80 energy at the end of Deep Insight, and thus move two SS from Deep Insight into the 0% buff period, we now only spend 5 seconds at no insight, dropping our cycle to 40 seconds. Our average damage buff is thus boosted to 750 / 40 = 18.75%; however, this only applies to our non-SS damage. We're now spending 10 seconds worth of energy in each of the 4 phases, so our SS damage boost drops to (0 + 10 + 20 + 30) / 4 = 15%. Thus, the net DPS modifier is now 20% * 15% + 80% * 18.75% = 18%.

Hence, in this (immensely simplified) model, we gain a 1.3% damage increase by pooling energy. Obviously this neglects the fact that some energy gets spent on finishers (and that those finishers in turn may move across insight levels). But the general premise of reducing the damage of moves accounting for 20% of your damage output to boost the other 80% seems sound.

Also note that you do want to use Anticipation to shuffle finishers into higher Insight levels - i.e., after you gain 5 CP, its almost certainly desirable to keep SSing until you make the next insight level to get an extra 10% out of the eviscerate. So at 0-20% it seems likely you want to be dumping energy as fast as you can and saving CP, while at 30% you want to be dumping CP and saving energy.

I am still confused. How we should gem? Pure haste in yellow and pure haste in all sockets when the bonus on an item doesn't matter? I checked some top rogues: some of them socket pure haste in some item and agi/haste in other, some socket only agi/haste and get 60 crit rating bonus on an item... One rogue reforged haste to mastery and it was a combat rogue. So yeah didn't really give me any clues. But the thing is i have only like 29k AP while they have 36-39K AP and that's the thing that worries me. Having read these ratings only for T14 rogues in bis gear i assume at that point we will have more agility meaning more AP which would allow us to just stack haste but for now i should just go for agi/haste in every possible socket. Am i right? I am by no means a theorycrafter and since shadowcraft doesn't work atm i am a bit confused as i have already said :P. This is my rogue

By the way another thing i wanted to ask is about trinkets. [Windswept Pages] and [Zen Alchemist Stone]. They are close to each other without reforges. But if we happen to reforge the stone to let's say haste/expertise/hit (without overcapping of course) it will be better right?

I dunno but AGi is 2.7 whilst Haste ist 1.6ish... I still think gemming for Agi is the best way to go.

For gems, secondary stats are twice the amount of the regular stats like agility. On a gem you have 160 agility, but haste gem is 320. or on a hybrid, you have 80 agility and 160 haste. So 2x haste = 3.2ep, which is higher than Agi's 2.7.

re: pooling and deep insight uptime.

I messed around with the action lists in SimC for a while, but no matter what I did with it, I could not produce gains in dps. Now, this may be a limitation of my knowledge of the pooling action in SimC or of SimC itself, but pooling didn't seem to accomplish anything.

When I made the rogue pool at the end of BG30 to push a SS or RvS into BG0, I got the expected small increase in average BG30 uptime (about 1% increase), but the dps went down by a fraction ( about 0.3%).

Alternately, if I pooled at the end of BG20 so there would be more energy to use in BG30, I got the expected decrease in BG uptime, but again the overall dps went down by a (very) small amount.

I looked at a few sample action lists from these runs and it appeared to be pooling properly. The energy loss to overflow was within a couple points of the non-pooling baseline. The dps distributions were a bell curve... no clusters of low dps iterations where it got stuck in some improper logic pooling loop. While it seems intuitive that one method should result in a small increase, both seem to result in a small decrease when simmed to the best of *my* ability.

I suspect that even if that is an incorrect result and there is some pooling method to optimize the BG transition timings for a gain, it will make such a small impact it's probably not worth worrying about compared to the risk of energy capping introduced.

I believe this post sums up everything I have seen. I cannot tell you the gear breakpoints (anyone wanna give it a shot?!), but maybe something like this: low ilvl = <470, mid ilvl 470<x<490, high ilvls >490 (complete guess here). The basic point is that your gemming should most likely change over time.

Here is how I would recommend gemming red/yellow:
Low ilvl range: Focus on Delicate (pure agil).
Middle ilvl range: Focus on deft (Agil/hate) and delicate.
high ilvl range: Focus on quick (pure haste) and deft.

The EP values in the OP are, AFAIK, based on BIS gear, and as we all know, gear does alter those values. If someone has a lot of spare time (I don't right now, hence slow updating on the mechanics thread), it would be nice to get a ep value check at 460, 470, 480, 490, etc. to confirm the breakpoints.

For fights like Stone Guard, where BF is hitting 2+ targets 100% of the time, will EP weights be much different? Energy regen gets docked by 20% so I imagine haste will be devalued by a small amount since it directly contributes to energy regen. Anyone else have some insight on this?

I was wondering, if you enchant dancing steel on both weapons. Do they have each their own proc, so there can be 2 procs up at the same time, or do they just overwrite eachother as one? (hope you know what i mean)
If the second scenario is the case, it would decrease the value of a second dancing steel.
Can anyone confirm one or the other, id like to be sure that it wont be a waste or even a "shot in my foot" before i enchant it