A Chinese Mage in US Realm: Self-Intro and Questions

[Akraen]

Uh...I knew I would soon speak something noobish in front of you, Mr. Stasik. I'm so utterly embarrassed now. Still, it's better to know late than not to know, and now I won't have as much pressure to refresh my bombs.

I did some very quick mental calculation just now, though. I believe bombs also benefit from Mana Adept. A extra tick of LB is a little more than 9450 haste, and adds 25% of base (0 haste) damage. The same amount of Mastery would give me 31.5% of total (after haste modification) damage. Is this argument plausible? If it is, does it indicate to enhance bomb damage, an arcane mage should stack mastery in place of haste?

Can you explain how you get the number 31.5% damage and what you mean by "after haste modification?"

I tend to speak specifically for frost, however DoTs function the same for any spec. DPET gains at breakpoints are rather significant though. I actually did a hypothetical analysis based off 5.4.7 stats (so it would be ~item level 580) which takes into consideration stat tradeoffs. DPET gain by moving from 9577 to 18974 with Living Bomb is 14.28%, whereas the jump to 28k was determined to be 12.50% because of the amount of stats one must sacrifice (crit/int) to get to 28k. Here is a chart which represents stat combinations:

You can see that amplification does play a role, and now that we can reach item level 588, I should update this chart to go one more to the right with updated stat combinations (my argument is that it is worth it to go to 28k haste if you do not have to break socket bonuses):


As I try to think of how you came up with 31.5% from mastery gains I also look at some analysis I did about changing mastery levels. For arcane everything is increased by Mana Adept, however the gap between LB ticks is very large and would require giving up too much mastery to go for 18974+frost armor or whatever its equivalent is with mage armor (~23k?). On multidot fights like TFP and Galakras, Arcane can benefit from Nether Tempest breakpoints.

Frost is limited in its benefit from mastery as it only scales with icicle generators. This is a good thing, otherwise frost would only care about mastery. DPET gains from mastery are only on Frostfire Bolt & Frostbolt, and they have a difference of 14-16% depending on your gear (i.e. if you stack mastery those spells will do ~15% more dmg than if you stack haste or crit).

More info on how I analyze those stats can be found here.

[zhengma]

What was on my mind when I typed that paragraph was: Mana Adept is 300 mastery/1%, and it's applied to "all spell damage done", meaning every tick will be benefited. The haste breakpoints are the same for both specs (assuming Frost Armor is used in both cases). The 9400+ haste one needs to go from one LB breakpoint to the next, if put into mastery, could potentially add more DPCT.

However, one thing that many DPCT analysis overlook, as well as my calculation in the last post, is the fact that DPCT is inversely proportional to GCD, and hence linear (of course, assuming fixed % buff) to haste until 14242. As of 5.4.8, it's not very likely to get 14242 haste when stacking mastery, but fortunately, for LB there is only one meaningful breakpoint (the other will be reached however haste is avoided) below 14242. Intuitively I think if an Arcane mage who wants to maximize the DPCT of LB should get 9522 haste and stack mastery like mad, but I need solid experiment/simulation to say that for sure.

The entire second paragraph of my previous reply is strictly limited to the Arcane spec, which I've been studying this fortnight. Icicles, on the other hand, have no interaction with Bombs, which explains Fig. 2 of your reply, that lighter-colored line are always higher, and Haste>Crit>Mastery for bomb damage. However, we're not afflock or spriest, and I still want my icicles hurt, so in practice I'll still put crit last.

I know you don't do much research yourself on Arcane, but if you happen to know any literature on maximization of LB (or Bombs in general) damage for an Arcane mage, experiment or simulation, please recommend. Otherwise I'll just start a small research project myself.

[Komma]

What was on my mind when I typed that paragraph was: Mana Adept is 300 mastery/1%, and it's applied to "all spell damage done", meaning every tick will be benefited. The haste breakpoints are the same for both specs (assuming Frost Armor is used in both cases). The 9400+ haste one needs to go from one LB breakpoint to the next, if put into mastery, could potentially add more DPCT.

However, one thing that many DPCT analysis overlook, as well as my calculation in the last post, is the fact that DPCT is inversely proportional to GCD, and hence linear (of course, assuming fixed % buff) to haste until 14242. As of 5.4.8, it's not very likely to get 14242 haste when stacking mastery, but fortunately, for LB there is only one meaningful breakpoint (the other will be reached however haste is avoided) below 14242. Intuitively I think if an Arcane mage who wants to maximize the DPCT of LB should get 9522 haste and stack mastery like mad, but I need solid experiment/simulation to say that for sure.

The entire second paragraph of my previous reply is strictly limited to the Arcane spec, which I've been studying this fortnight. Icicles, on the other hand, have no interaction with Bombs, which explains Fig. 2 of your reply, that lighter-colored line are always higher, and Haste>Crit>Mastery for bomb damage. However, we're not afflock or spriest, and I still want my icicles hurt, so in practice I'll still put crit last.

I know you don't do much research yourself on Arcane, but if you happen to know any literature on maximization of LB (or Bombs in general) damage for an Arcane mage, experiment or simulation, please recommend. Otherwise I'll just start a small research project myself.

I'm not sure I understand what you mean by "DPCT is inversely proportional to GCD, and hence linear (of course, assuming fixed % buff) to haste until 14242". LB has a fixed 1s GCD regardless of your haste, which is why it only gains damage at haste breakpoints, or through mastery (and of course, intellect). It is not inversely proportional. This is one of the main reasons why Frost Armor is not favored for Arcane in most SoO fights, which are not strictly single target.

Like I mentioned earlier, my reasoning for why Frost Armor is so dominant among CN Arcane Mages is because of results from Simc, which runs a default setting of single target, 360s-540s duration. When you shorten the duration, the relative % of time spent under hasted effects such as bloodlust increases, and Mage Armor is more dominant. When you change it to multitarget, Mage Armor again pulls ahead.

[Akraen]

Living Bomb is universally 1 sec GCD, regardless of your haste.

I'm hoping one of our Arcane experts can jump in for your other points

[zhengma]

What was on my mind when I typed that paragraph was: Mana Adept is 300 mastery/1%, and it's applied to "all spell damage done", meaning every tick will be benefited. The haste breakpoints are the same for both specs (assuming Frost Armor is used in both cases). The 9400+ haste one needs to go from one LB breakpoint to the next, if put into mastery, could potentially add more DPCT.

However, one thing that many DPCT analysis overlook, as well as my calculation in the last post, is the fact that DPCT is inversely proportional to GCD, and hence linear (of course, assuming fixed % buff) to haste until 14242. As of 5.4.8, it's not very likely to get 14242 haste when stacking mastery, but fortunately, for LB there is only one meaningful breakpoint (the other will be reached however haste is avoided) below 14242. Intuitively I think if an Arcane mage who wants to maximize the DPCT of LB should get 9522 haste and stack mastery like mad, but I need solid experiment/simulation to say that for sure.

The entire second paragraph of my previous reply is strictly limited to the Arcane spec, which I've been studying this fortnight. Icicles, on the other hand, have no interaction with Bombs, which explains Fig. 2 of your reply, that lighter-colored line are always higher, and Haste>Crit>Mastery for bomb damage. However, we're not afflock or spriest, and I still want my icicles hurt, so in practice I'll still put crit last.

I know you don't do much research yourself on Arcane, but if you happen to know any literature on maximization of LB (or Bombs in general) damage for an Arcane mage, experiment or simulation, please recommend. Otherwise I'll just start a small research project myself.

I'm not sure I understand what you mean by "DPCT is inversely proportional to GCD, and hence linear (of course, assuming fixed % buff) to haste until 14242". LB has a fixed 1s GCD regardless of your haste, which is why it only gains damage at haste breakpoints, or through mastery (and of course, intellect). It is not inversely proportional. This is one of the main reasons why Frost Armor is not favored for Arcane in most SoO fights, which are not strictly single target.

Like I mentioned earlier, my reasoning for why Frost Armor is so dominant among CN Arcane Mages is because of results from Simc, which runs a default setting of single target, 360s-540s duration. When you shorten the duration, the relative % of time spent under hasted effects such as bloodlust increases, and Mage Armor is more dominant. When you change it to multitarget, Mage Armor again pulls ahead.

Thanks for the reminding on LB (as a major counter-example of when you SHOULD overlook haste in DPCT calculation). To be on-topic again, this type of corrections on details, provided without verbal mocking/humiliation, is indeed a major reason I choose to come here instead of Chinese forums.

Now come back off-topic. With Mage Armor on, 5-tick breakpoint becomes 3039 and 6-tick breakpoint becomes 13163. The former is gonna be reached anyway and the latter is too high as of 5.4.8 to be worth it.

Hence, can we conclude that: to maximize LB damage, an Arcane mage should use Mage Armor, ensure 3039 haste and stack mastery?

[Komma]

Now come back off-topic. With Mage Armor on, 5-tick breakpoint becomes 3039 and 6-tick breakpoint becomes 13163. The former is gonna be reached anyway and the latter is too high as of 5.4.8 to be worth it.

Hence, can we conclude that: to maximize LB damage, an Arcane mage should use Mage Armor, ensure 3039 haste and stack mastery?

13163 is not too high to reach in 5.4.8. In fact, I was running >14K even before the new upgrades were released, and I wasn't trying to stack haste either. 3039 is obviously achieved without giving any effort. In that sense, your conclusion is correct. Keep in mind that on single target fights, LB is only ~20-25% of our damage though, compared to AB/AM each being around ~31%.

The only case in which I know 13163 haste is not achieved with best-in-slot gear is with Voltaa's 2T16+3HWF setup, where you replace crit/haste tier pieces with hit/mastery and crit/mastery pieces. This provides the highest bomb damage, but comes at expense of the 4T16 bonus, which single target damage benefits from.

Most of what we've said so far is documented in existing Arcane theorycraft. If you'd like to ask specific questions or request suggestions for your setup, you can make a thread in our "Help me" forums. If you'd like to spend more time on the details of Arcane theory, you might want to make a thread in the Arcane forums instead.

[zhengma]

13163 is not too high to reach in 5.4.8.

When I typed that down, I was having people like me who is 0/14 H and can't find a reliable group to do Heroic progression in any near future (but I'm probably the only such case among all current altered-time users, judging from the armories I've read). With full heroic gear, 13163 is surely easy to achieve.

Keep in mind that on single target fights, LB is only ~20-25% of our damage though, compared to AB/AM each being around ~31%.

But as another way off-topic question, in certain sparse multi-target fights such as Protectors, Norushen and Spoils, is it still advisable to "ensure NT 100% uptime on everything > all else", for a DPCT consideration? Different strategy guides in China and Western give different analysis and arguments on this issue.

The only case in which I know 13163 haste is not achieved with best-in-slot gear is with Voltaa's 2T16+3HWF setup, where you replace crit/haste tier pieces with hit/mastery and crit/mastery pieces. This provides the highest bomb damage, but comes at expense of the 4T16 bonus, which single target damage benefits from.

I have my (Normal Warforged) Fusespark and Dou Dou Chong, but the reason why I don't equip them is: more AM = easier mana management = spending longer time in 4-charge status with a higher mana level to fluctuate around. Without 4T16, my mana will be harder to stabilize, which could negate the benefit from mastery. Of course, there is no theoretical or simulational study to support it, and it has much to do with my mana management skill (and my aesthetic appreciation of the visual effect of AM). If Voltaa or anyone has published solid works on the pros and cons of this build, please link so that I can learn from it.

Most of what we've said so far is documented in existing Arcane theorycraft. If you'd like to ask specific questions or request suggestions for your setup, you can make a thread in our "Help me" forums. If you'd like to spend more time on the details of Arcane theory, you might want to make a thread in the Arcane forums instead.

Thank you for your kindness. Currently I do reforge by hand (while consulting AMR for crude min/max suggestions) and, due to the raiding environment I'm in, I have to work with whatever non-BiS piece I can obtain. But I'd surely ask questions if I need advices on this.

I think I've established a basic understanding of the major points regarding the two specs I choose to play, what I do lack is: 1. systematic, quantitative analysis and training to perform such analysis; 2. subtle interaction between class abilities and mechanisms of specific fights. I'll ask well-defined, well-formulated questions if I have any.

I do hope there could be a compilation of major theorycrafting publications for each spec. As a PhD-candidate in real life, I know well where to find papers if I want to read some, but I can't say the same when it comes to mage TC. Such compilation will not only help newbies like me in knowing what has been done, but also, more importantly, inspires us to find what hasn't been done and try to do them. Review article (the type with 100+ citations) will also help.

[Frosted]

I do hope there could be a compilation of major theorycrafting publications for each spec. As a PhD-candidate in real life, I know well where to find papers if I want to read some, but I can't say the same when it comes to mage TC. Such compilation will not only help newbies like me in knowing what has been done, but also, more importantly, inspires us to find what hasn't been done and try to do them. Review article (the type with 100+ citations) will also help.

You won't find this type of material in the mage community. MoP mage TC has been scattered around multiple threads that are sometimes over 100 pages long, with sparsely updated OPs and no compendium of sorts for any spec. Additionally, you will not find any rigorous or thorough math behind any spec, or their mechanics. Almost everything we know right now has been brought about by anecdotal evidence and "common sense" style thinking (which may be wrong, as updated Sims are beginning to show). One of the main reasons I liked the idea of AT.com is so that we could begin to generate the type of information/resources you and I are looking for.

[Komma]

13163 is not too high to reach in 5.4.8.

When I typed that down, I was having people like me who is 0/14 H and can't find a reliable group to do Heroic progression in any near future (but I'm probably the only such case among all current altered-time users, judging from the armories I've read). With full heroic gear, 13163 is surely easy to achieve.

Keep in mind that on single target fights, LB is only ~20-25% of our damage though, compared to AB/AM each being around ~31%.

But as another way off-topic question, in certain sparse multi-target fights such as Protectors, Norushen and Spoils, is it still advisable to "ensure NT 100% uptime on everything > all else", for a DPCT consideration? Different strategy guides in China and Western give different analysis and arguments on this issue.

The 13163 haste mark is attainable even in normal gear - the key item in that case is BBoY. I am not sure whether BBoY attaining the haste breakpoint makes it better than KTT at normal gear levels. And don't worry - there are a good number of non-heroic raiders here!

Ensuring NT is up on everything (that doesn't die within a few seconds) is indeed the strategy on multidotting fights such as Protectors and Spoils.

As Frosted mentioned, there isn't any detailed, citation-included works in the form you described (Probably because you don't get any degrees for it!). Regarding 4T16 vs 2T16 though, the "general consensus" is that 2T16 sacrifices single target damage in favor of high mastery, which benefits heavily on multitarget/AOE fights due to bomb multidotting + higher 4-stack surfing output. There's no "solid work" published - just some forum comments or IRC chat messages here and there.

[Kver]

I do hope there could be a compilation of major theorycrafting publications for each spec. As a PhD-candidate in real life, I know well where to find papers if I want to read some, but I can't say the same when it comes to mage TC. Such compilation will not only help newbies like me in knowing what has been done, but also, more importantly, inspires us to find what hasn't been done and try to do them. Review article (the type with 100+ citations) will also help.

You won't find this type of material in the mage community. MoP mage TC has been scattered around multiple threads that are sometimes over 100 pages long, with sparsely updated OPs and no compendium of sorts for any spec. Additionally, you will not find any rigorous or thorough math behind any spec, or their mechanics. Almost everything we know right now has been brought about by anecdotal evidence and "common sense" style thinking (which may be wrong, as updated Sims are beginning to show). One of the main reasons I liked the idea of AT.com is so that we could begin to generate the type of information/resources you and I are looking for.

This is also one of the reasons why it is great to have founders and administrators who are heavily involved in the theorycrafting scene themselves, rather than owners who are focused on earning money. It is also why we are going to build an archive of sorts, once we figure out the most efficient way to do this.

[zhengma]

You won't find this type of material in the mage community. MoP mage TC has been scattered around multiple threads that are sometimes over 100 pages long, with sparsely updated OPs and no compendium of sorts for any spec. Additionally, you will not find any rigorous or thorough math behind any spec, or their mechanics. Almost everything we know right now has been brought about by anecdotal evidence and "common sense" style thinking (which may be wrong, as updated Sims are beginning to show). One of the main reasons I liked the idea of AT.com is so that we could begin to generate the type of information/resources you and I are looking for.

That's the whole point, if there's none as of yet, we'll start some. Someone has to initiate something, isn't it?

And the compilation, in contrast to review article, is just a post with dozens of links, linking to various threads scattered around the Internet, so that beginners can at least start somewhere without blindly googling.

Almost all Chinese TC works (for any class) are published on one (or maybe two, though the latter is dominated by immature trolls) forum, and I can reach them with an in-forum search. While that's not quite the case in the western, I'd truly hope AT.com can at least build into a place where people would choose to publish their mage TC works, and existing major TC works from elsewhere are accessible via links in a few posts.

[Aowyn]

So, how is this "Arcane is easy" thing originated? Why are there people who apparently know better buying it?

This originated probably from the Wrath expansion, when Arcane consolidated a rotation that didn't required fire/frost spells to be viable. At the time, the rotation was basically 4 Arcane Blasts -> Arcane Missiles (or 4AB->AM). It didn't even involve Arcane Barrage, unless you were really unlucky with the AM proc (a very rare occurence).

The problem was, and still is, that DPS tracking addons only show the spells and mechanics that dealt damage, not all the relevant information behind the rotation. As such, while you would see 6 to 10 sources of damage for other classes, you would see just the following for an Arcane mage:

1. Arcane Blast 80%
2. Arcane Missiles 19%
3. Mirror Images 1%

This hides all the work for minimizing movement, managing mana and timing cooldown usage that Arcane relied at the time.

Furthermore, Arcane was an spec that was easy to start and friendly with being low ilvl; just from speccing Arcane and following the 4AB->AM rotation, you would already perform decently well. The spec was hard to master; very few achieved the top dps it could output, surpassed only by assassination/combat rogues. But very few players would even see such top performance for any class and would see only the average Arcane mage beating other average players. Few players, thus, had a good notion of how much Arcane's DPS is influenced by minimizing movement, managing mana and timing cooldowns.

This is also why Blizzard changed so much the spec's rotation over Cata and MoP; they wanted to change the spec's rotation to make it feel more complex.

Thank you for your detailed explanation. Now one part of my question, regarding how western WoW players appreciate the sophistication of Arcane, has a satisfactory answer. However, I'd like to ask more regarding another part of my original post, the prejudice against Frost users I witness in CWoW mage community. I admit the fact: if we had to name one spec that's "lagging" in PvE in Patch 5.4, it would most likely be Frost. I know some prominent western mages (like Vykina and Cycobi) also admit it to some extent. I have never, however, seen anyone on US realm saying "Wow dude you have a Water Elemental by your side wtf? How noobish lol! (source)" A few weeks back, a Chinese mage theorycrafter published an entire long essay just to argue "there's not a single fight in SoO where you should spec in Frost, which is for beginners". I'm curious whether there is anything similar in the western mage community?

This may have another problem than the SimC's situation and Frost's soft caps. It is that each spec has a different variation in performance, with Fire having the highest among mage specs. Frost's variation isn't expected to be high, since it isn't affected much by movement (unlike Arcane) nor by crit RNG (unlike Fire). With this, Frost's top variation may not be as high as the top variation of the other two specs.

But most DPS parsing services focus on showing the highest values, instead of the averages (or medians, more robust to outliers) and their confidence intervals. This skews the statistics towards specs that just had people lucky enough to get that top variance. It is an example of statistics badly employed, or badly interpreted.

The western community had this for some time in MoP, having Fire getting absurd DPS parses due to lucky crit sequences on the opener leading to massive Combustions. As such, while Fire was theorycrafted to be within a small margin from Arcane and Frost, many players (particularly those from other classes) saw the opposite.