Pure speculation of the most rank variety, and I "almost certainly" overstated it by saying "almost certainly". Will obviously depend on a large number of currently unknown factors.

We need to know the numbers to see if this is the case. I very much doubt it is -- remember, AB is part of efficient rotations too (the major part, I'd guess, although again it depends on what those rotations turn out to be), so buffing its damage doesn't only help the high MPS rotations.

And note that Mana Adept will serve to naturally discourage the use of said high MPS rotations. As Roywyn points out in the post above, there may well be a greater danger that arcane ends up with a boring no-burn rotation that hovers near 100% all the time if they foul up the numbers.

In any event, my point was only that the player does, in fact, have a meaningful way to mitigate the effects of RNG on his or her MPS (and to take advantage of favorable RNG) via rotation shuffling, assuming there are viable rotations to shuffle and cause to do so (which is pretty clearly at least the intention, given Kalgan's post). I actually like this model -- it was the interesting part of WotLK arcane, but it just didn't matter enough to be significant.

See, I affirmatively like this -- you say "issue," I say "feature". If they do it right, it will add skill to the spec. If arcane's MPS were very narrowly distributed, it would be much easier to optimize your play -- there would be certain optimal balances of the various useful rotation, and you'd know ahead of time how much of each one to use. When your mana usage fluctuates, it in effect creates a new optimization problem after every unexpected peak or trough in your MPS, which puts a premium on thinking on your feet. I like.

Perhaps someone should point this out on the BETA forums?

The intention is obviously that there should be a burn cycle. Kalgan agreed with a post that stated as much, and piling more damage into Arcane Blast as well as removing the ability for Arcane Barrage and Missiles to benefit from Blast stacks all indicate that a burn cycle is the idea. If, despite these changes, a burn cycle still isn't a viable option then someone should bring it to the attention of the BETA mage thread with some numbers to back up the fact it isn't working.

If indeed burning isn't an option as Arcane then playing Arcane as a spec would be pretty bizarre. Having the ability to stack Arcane Blast to four, but never stacking it past two, would make for an utterly pointless mechanic. The additional stacks might as well be removed and the damage of Arcane based around the idea of only two stacks.

An artificial way past this is just to ensure Mana Adept doesn't work on a gradient, but rather, has 20% effectiveness down to 50% mana and then drops the bonus entirely. It's pretty boring, but trying to maintain mana at 80% for a six minute fight sounds even more boring.

Not to nitpick but, the 85 premades in gear have a total of 12 mastery points (i.e. 8 base and 4 from gear), giving them a MA bonus of ~18% at max mana.

That being said, the point you raise is valid, especially the one concerning scaling with gear. Though I think Blizzard have already started attacking this problem as well.

As far as I can tell, they are messing with the coefficient of AB somewhat. I will need to confirm it, but it seems they have tweaked the coefficient to have AB scale better then it is currently. (right now, ABs on the beta are packing a wallop like never before, and thats in this cheapish premade gear).
One point that I do think that will play a big part, as you mentioned, will be the increased mana regen in better gear. I think this will play directly into what kind of "pre burn" phase rotations we can use. I would imagine it would be something along the lines of "in entry level gear, use AB1 rotations for preburn but even those are unreliable/unstable; in higher tier gear, your default 'preburn' would be AB2/3, and if something goes wrong (e.g. with RNG), you can shift down to AB1 for more savings".

If this idea comes to pass (and assuming they ensure that at all gear levels a 'burn' phase is desired), I can see the 'reliability of preburn phase cycles' issue that I have been banging on about for a while now, to find a natural solution at higher gear levels. This is not a bad thing overall, imho, and in some sense it does resonate with an idea that Blizzard has stated they hold to in the past, that being that they want higher gear levels to change up what you would be doing, as well as make you stronger.

Maybe it is their design that at entry/low gear levels, Arcane is rather unstable and unpredictable, which then evens out as your gear improves?

I will try to run some tests on my sim to find out at what gear levels does Arcane's AB1 rotations become stable.

I'll second this. It sounds worse than boring -- almost claustrophobic, especially since Logix's point about the variance of even low MPS arcane cycles would make maintaining your mana in this case a matter of pure RNG. It'd be maddening, and there'd be nearly nothing you could do to adjust your consumption.

Why don't they just bite the bullet and turn evocation into a self-innervate? As long as it stops your DPS for N seconds, it's going to be tricky for them to tweak the numbers so that it's better to burn and evocate instead of just using more efficent cycles and only using the instant portion of the ability. If it only takes a global (and maybe it could be free during AP or something a la those other abilities), dropping your mana level seems a lot more attractive.



I think it's always been their intention for arcane mana consumption to be unpredictable -- I can't imagine why they'd implement the combination of AB and missile barrage if they wanted the spec to consume mana at a relatively constant rate.

Personally, I like this model, as long as they give us tools to ride out the fluctuations, but I don't blame you for feeling differently, and I'm open to other ideas about arcane's mana management. I just want to make sure it's a challenge to control your mana levels, not an easy but boring chore. I like the way the RNG forces an element of improvisation.

My gut sense is that without other changes, smoothing out arcane's mana usage would make playing the spec feel like paying a monthly credit card bill -- "oops, down to X% again, time to reach into the cooldown bank and clear the debt." Making on-the-fly decisions about what rotation to use sounds fun to me; watching two cooldowns and a blue bar does not.

But you could absolutely make arcane interesting to me while flatting out some of the annoying MPS fluctuations -- you'd just have to throw something else in to keep the play from getting too repetitive and I'd be happy.

As acrane blast is decoupled with other spells, how does arcane power interact with arcane blast in beta?

E.g. at 4 stacks, the damage with AP is 1.8+0.2 or 1.8x1.2 ?

Its multiplicative.


On a side note, I have updated my simulator (http://manamadesimple.elementfx.com/).

The update does a few things:

• Accurate damage calculations based on input. All damage calculations are now up to date and accurate as of beta patch 12984.

• Added the ability to define exact mana % parameters for a simulation, i.e. you can now test how a specific rotation performs between any mana levels, e.g. test an AB3 AM/ABr rotation between 73% and 45% mana.

• Added crit calculations. Spells damage will now change depending on crit % input as expected

• Added the use/non-use of the Blessing of Kings mana returns.

• Many smaller optimizations

The todo list will be;
- adding support for Arcane Power + trinkets + mana gem, as well as their usage at specific levels, controlled by input.
- maybe adding critical damage on a per crit basis, instead of just adjusting spell damage output with crit, though I can't think of a reason to do this. The sim's mana model accurately simulates procs and mana gains, since I feel those are more important and can't just be 'averaged out'.
- add the ability to 'chain together' multiple different rotations at different mana points within a single simulation, e.g. tell the sim to run AB1 at 100-80%mana, then change to AB2 from 80-60%, then AB4 from 60-40. You can do this now with the mana controls, but you have to run the sim multiple times and manually change the rotations and mana %.

If any of you have any comments, suggestions or input on this sim, or see any bugs, let me know! I would much appreciate it.



Here are some tests that show some interesting output. Done for a mage with 4k int, 12 Mastery points (18% MA bonus), 20% crit, 15% haste, using Mage Armor over 100 iterations.

ABspam from 100% to 40% mana rotation, vs, AB4 AM/ABr
ABspam
AB4 AM/ABr
A few interesting observations with this data:
ABspam + ~5.22 seconds of evocation = 159,737 damage over 26.23 seconds on average. Which is ~6089 DPS on average.

ABx4 AM/ABr is, on average, 5818 DPS over the same mana range.
Over the same time frame (running an ABx4 simulation with a 26 second limit), ABx4 AM/ABr does ~5,842.6 DPS.

Meaning, if you have ~20-25 seconds of free casting time available (i.e. AP), and have Evocation available, you will probably want to choose ABspam over some other rotation.

This may help us get started in deciding what kind of rotations we will want to use given certain situations in the fight.

Logix: That is great work, I started to do similar tests on the PTR (No beta club here) but the results were too buggy due to the spell mechanics.

Could we see similar test results under a few differing scenarios? Can you include your total mana, so we can plot total mana used over time?

1. ABspam / AMproc (I would like to see time vs mana usage if we leave out ABr.)

2. ABX2 /AM -- ABr (I would like to see a mana conservation cycle or two, to measure the dps decrease over a 2 min time period)

Basically if we are looking for theoretical burn cycles, we should try to find the highest, burn cycle, followed by the highest slow burn cycle over time. Then try to find whether a hybrid exists, say burn to 70% and mana conservation to 40%. Because as it stands now, the only true burn phases we will have it the first burn to evocation, and then the last burn phase to boss kill. The rest of the boss fights are mana burn/conservation hybrid phases of some sort or another.

Edit: The AMspam rotation appears to use 60 % of your mana over a 20 second period of time or 3% of your mana per second. This is fine for the first burn phase, Open Boss Fight ----to-----> 20 seconds. Then we are going to Evocate back to 100% and develop a mana burn/conservation cycle that will use 60% of our mana over a 120 seconds period of time, or until evocation is off cool down.

Your second example ABx4/AM-ABr takes place on average over 60 seconds, but has a very large standard deviation, which makes it less predictable. This rotation uses on average 1% mana per second for 60 seconds, but it also used 1.5% mana over 40 seconds and also produced a use of .75% mana over 95 seconds.

Now we are looking for rotations or combination of rotations which yield on average .5% mana per second or will use 60% of the mage's mana over 120 second period of time. So the first challenge is to find a rotation that will use less than or equal to .5% mana per second. I read some of the discussion that an ABX1/AM proc will produce a positive mana consumption rate with mage armor, but the ABX1/ABr rotation produced a negative consumption rate. On average those two rotations seem to balance one another, when you factor in clear cast proccs. So the real answer lies somewhere between, at least for now.

My theory is that during the first part of a 120 second phase, the arcane mage can take advantage of the mastery bonus and cast a higher dps rotation with a higher mana usage per second, for X seconds, and then switch to a lower dps rate with a lower mana usage per seconds for Y seconds (Y=120 seconds - X seconds) when the master bonus is not as strong.

After playing with the above simulator I feel that I've confirmed what I had worried about before in this thread, that given the opportunity, arcane mages want to avoid haste like the plague.

Due to the fact that haste's dps increase comes at the cost of more mana spent, and that (and this is assuming the burn, evocate, hover cycle of play) we're limited to finding rotations that maintain x mana for y length of time, haste ends up at the bottom of the ladder.

For example, I used a base stat list of 5000 int, 12 mastery, 20% crit, 15% haste, with a lull phase down to 83% and a burn phase down to 40%. I ran these using 1000 iterations. Yes there is variance but averages are what we have to craft by.

When adding 10 mastery, you get the following numbers
Burn phase DPS: 9936.3
Burn phase Length: 24.11s
Lull phase (3AB, AM whenever it is up was the best dps rotation I could find with these stats that fit the limitation)
Lull phase DPS: 7111
Lull phase maximum average length: 99.77

When adding 14 haste (rating equivilent to 10 mastery), you get these numbers
Burn phase DPS: 10112.8
Burn phase Length: 21.09s
Lull phase (2AB, AM>ABr. Never AM at 1 stack. Again the best rotation for these stats)
Lull phase DPS: 6816
Lull phase maximum average length: 122.27

When playing these out in actual 2 minute segments you get
BurnDPS*BurnLength+LullDPS*(120-6-BurnLength)

With the mastery stats: 7323.1
With the haste stats: 7054.613

To continue on, using crit instead: 7119.767.
And for completion, using int (+896) instead: 7581.253, though I dont think anyone would have doubted we would be gemming for int.

Point for point haste is immediately better whenever adding haste doesn't force you to lower your rotation, but this is very often the case, and so in most circumstances when sticking to rotations it seems to me that going for Mastery+Crit gear (and reforging to Mastery) is the way to go.
That said it may be possible for an amazing player to achieve the optimal amount of dps by going for haste and dynamically altering their lull phase on almost every cast. Even still there is a level of haste (not particularly high, and always surpassed during bloodlust) relative to your mana pool where you run out of any options and are forced to use the 1xAB style, giving no room to game the system, and of course there is a point where even this becomes unsustainable.

In fact the very existence of bloodlust may be a deciding factor in killing haste as a stat entirely as even with no haste on gear at all, finding a stable rotation during bloodlust that doesn't drop our mana to "we can't ever reach 100 again without significantly cutting the burn phase" levels will be difficult enough, if possible at all.

Say what you will about complexity and awkwardness, and theres plenty to say, I find this mastery utterly fascinating and am actually rerolling Arcane Mage in cata just to play with it!

I wonder if anyone has an opinion (or perhaps math!) on gnome (+5% max mana) vs worgen (+1% crit)?
I know gnome will scale to int and mastery but my gut says gnome should come out on top.

Yea. The next version of my sim will include race selection.

From my preliminary investigation, the real front-runners are gnomes and blood elves. Over a 2 min period, gnomes come out on top from a 'raw numbers' perspective, but blood elves work well due to the ability to control when you get the 6% mana back, which helps giving that little bit of precision control with the extra mana (e.g. you can use it during periods of bad RNG to tip you to the mana % point you need to be).

That being said, in the grander scheme of things, the differences between the races is really small.

Hit and Int (in an yet unknown order) will be the best stats. This has always be the case in early tiers, and it's an assumed choice of Blizzard. They specifically want the best "stackable" stat (hit) to have a cap, for more "depth" in the gear choice. And spellpower has always be king in the early tiers, because it's one stat we stack a lot, and as any other stats, it has a negative relative scaling with itself (meaning that the more spellpower you have, the less additional dps (expressed in percentage) 1 more spellpower point will give you.

Now, having a healer as a main, I wanted to emphasize one point: mana consumption, and hence longevity or sustainability of cycles, depends heavily, and in a non-linear way, on the regen you have. That's why healers always start an expansion with stacking regen in the early tiers. A little bit additional regen has "low" effect on your longevity if your deficit is high. Going from a deficit of 1000 MP5 to 900 MP5 does not change a lot (rougly, you gain 11% more longevity). But when you come close to neutral cycles, the same absolute additional regen has a huge effect: going from 200 MP5 deficit to 100 MP5 deficit doubles your longevity, and going from 110 to 10 MP5 deficit multiplies your longevity by 11 !!! And this corresponds to the same absolute gain of 100 MP5. This means that you should really check wether your conclusions of longevity are valid at several points of gearing / intellect / regen. And that the simulation is really correct as far as regen is concerned.