Fire Mage Guide

[Deleted]

Makes you wonder if windsong is working as intended or if we can expect to see a nerf soon.

[Deleted]

Two questions here:

1. Would 6 extra seconds of all procs from Alter Time not be worth more than going for combustion scenario 1? (Assuming this is right after pull when you actually have everything up)

2. Could i get a link to some calculations about horde racials?

[huth]

Makes you wonder if windsong is working as intended or if we can expect to see a nerf soon.

Hard to tell. It makes sense for Elemental Force to work on a per-target basis, but not for Windsong.

[Kuni Zyrekai]

Tell me stupid but i dont know what u mean.

Channeling evocation is not 100% damage, it's 0%. It's a cast time that results in no damage done. You have 40 seconds of 125%, and 6-haste of 0%.

[Mastamage]

So what you're saying is to let your Ignite DoT fall off while you're channeling Invocation? I have no problem keeping myself in Rune of Power and neither should any Mage. Even on fights with high movement I'm still able to drop one and stay in it. Building a massive Ignite and losing it for an Invocate sounds useless.

[Kuni Zyrekai]

So what you're saying is to let your Ignite DoT fall off while you're channeling Invocation? I have no problem keeping myself in Rune of Power and neither should any Mage. Even on fights with high movement I'm still able to drop one and stay in it. Building a massive Ignite and losing it for an Invocate sounds useless.

If ignite rolled infinitely, yes. It doesn't though.

[mrgreenthump]

Here're the actual number of procs we saw in a 6 minute fight:
- Resto Shaman (keeping healing rain down all the time): 45 procs (150% uptime, yes that means she ALWAYS had 1 of the 3 buffs up, and half the time she had 2 of the 3 buffs up)
- Aff Warlock 1: 38 procs (126% uptime -> 630 crit, 630 haste, 630 mastery)
- Me: 21 procs (70% uptime -> 350 crit, 350 haste, 350 mastery) (Jade Spirit: 1490
Windsong: 2033.5)
- Boomkin 1: 18 procs (60% uptime -> 300 crit, 300 haste, 300 mastery)
- Feral: 16 procs (53% uptime -> 265 crit, 265 haste, 265 mastery)

So, it becomes obvious that when using DoTs, the ability to get a massive # of rolls makes Windsong better than Jade Serpent.

TLDR; USE WINDSONG for Fire.... (Frost is going to be better with Jade Serpent, as they value int much more, and they don't have the DoTs to lead into ridiculous # of rolls).

WoL + gara'jal is pretty much the worst place ever to compare procs and proc times unless you are logging yourself. WoL from this weeks Gara'jal hc kill gave me 6 procs of jade spirit(which might be correct), but it states uptime for it was 154 seconds and 44%, which is clearly not true, so if you are comparing WoL, log yourself and make extra sure the results are correct, at least on Gara'jal.. That said yes windsong is prolly better for multitarget fight at least(if it really procs on absolutely everything), still not totally sure if its like it for single target though, at least till they nerf pyro chaining..

[Annoying]

Some small things: Your AT scenarios should have you getting the Pyro! and Heating up buff. Under heroism, you can cram 3 pyros into an ignite for massive combustion damage. You also shouldn't end AT early. You use AT as a chance to get more crits if the first string doesn't perform well. When I do that and get 2 noncrit pyros, I just fireball once, AT expires, and I try again. Generally one of the two will have a 40k+ ignite, if not much, much higher.

Other things:
1) Use normalized stat weights. Seriously, real hard to compare numbers when your int is up above 4. Even normalizing them, yours are different far different than I'm seeing. Conclusion: Everyone should be doing their own stat weights on simc.
2) You have a major logical fallacy in your windsong vs. JS topic. You're comparing anecdotal evidence to math and concluding "It becomes obvious". Also known as confirmation bias. The math on windsong isn't currently clear anywhere I'm looking. You're free to compare anecdotal evidence to other anecdotal evidence, but please don't mix them.

For example: My average uptime last night on H-Gara'jal for JS was 31.3%. Comparing that to the top parses with windsong puts them almost even on multi-target fights, and puts JS way ahead for single target.
Anecdotally? Sure, it's fine. Mathematically speaking? We know that JS shouldn't be that high, and looking at top parses isn't a good mathematical representation of the actual average behavior of windsong.

TL;DR: We don't know enough about windsong and RPPM to draw a conclusion.

[Windry]

44 chances that should be at 0,8% procchance each. You'd have a ~1% lower chance of getting a proc in a 10 second window than you would have at 1 chance every 3 seconds, and 6% lower than at the start of the fight(i.e. 10s cap).(Yes, it gets worse as your events per timeframe increase)

You forgot that the chance to proc per event goes down as the number of events increases. What you're describing seems more like a bug in the system that causes the proc chance to not get calculated properly for DoTs. Or maybe it only counts events for a single target, leading to higher proc rates on AoE and multidotting(which would easily explain the Shaman).

I'm confused. I read nothing about "the proc chance going down with more events per second."

I can understand if you're trying to say your average proc chance per event goes down with a lot of events.
However, that does not matter, what matters is your actual proc rate each second because that leads us to the important up-time, or average proc rate per time.

If Blizzard wanted to implement this.... lower each individual chance to proc with increased number of events, it would be the right direction in balance. My logs however show that people with more events per second will have radically more procs per time.

WoL + gara'jal is pretty much the worst place ever to compare procs and proc times unless you are logging yourself. WoL from this weeks Gara'jal hc kill gave me 6 procs of jade spirit(which might be correct), but it states uptime for it was 154 seconds and 44%, which is clearly not true, so if you are comparing WoL, log yourself and make extra sure the results are correct, at least on Gara'jal.. That said yes windsong is prolly better for multitarget fight at least(if it really procs on absolutely everything), still not totally sure if its like it for single target though, at least till they nerf pyro chaining..

The Gara'jal PoV was from a melee who spent 100% time upstairs.

The reason Gara'jal might be a bad fight to examine is because if: you spent 35 seconds upstairs, and 25 seconds downstairs each minute, and you had 12 seconds of proc each minute, it might divinde 12seconds/35 for the uptime.
This problem is easily worked around by... looking for the total number of procs that occurred in during the fight and then dividing by 6 minutes (or however long the fight was).

Some small things: Your AT scenarios should have you getting the Pyro! and Heating up buff. Under heroism, you can cram 3 pyros into an ignite for massive combustion damage. You also shouldn't end AT early. You use AT as a chance to get more crits if the first string doesn't perform well. When I do that and get 2 noncrit pyros, I just fireball once, AT expires, and I try again. Generally one of the two will have a 40k+ ignite, if not much, much higher.

Other things:
1) Use normalized stat weights. Seriously, real hard to compare numbers when your int is up above 4. Even normalizing them, yours are different far different than I'm seeing. Conclusion: Everyone should be doing their own stat weights on simc.
2) You have a major logical fallacy in your windsong vs. JS topic. You're comparing anecdotal evidence to math and concluding "It becomes obvious". Also known as confirmation bias. The math on windsong isn't currently clear anywhere I'm looking. You're free to compare anecdotal evidence to other anecdotal evidence, but please don't mix them.

For example: My average uptime last night on H-Gara'jal for JS was 31.3%. Comparing that to the top parses with windsong puts them almost even on multi-target fights, and puts JS way ahead for single target.
Anecdotally? Sure, it's fine. Mathematically speaking? We know that JS shouldn't be that high, and looking at top parses isn't a good mathematical representation of the actual average behavior of windsong.

TL;DR: We don't know enough about windsong and RPPM to draw a conclusion.

- In your situation of 2 pyros, a 2 second fireball, 2 pyros, combustion, you would be devaluing the ignite from your first 2 pyroblasts.
- Yes, I've been using my own stat weights. I agree with you when you say everyone should be doing their own stat weights in these comparisons.

- You're using anecdotal as a negative connotation, but of course it's anecdotal... since it's my guild's logs, of course it's from my guild's PoV.

If we have: Theory, Anecdotal evidence, and Actual evidence, well:
- Theory is easy, I laid it out
- Anecdotal evidence, well that's from my logs
- The only way to get the true Actual evidence is to go through all logs out there (or go through a very very high number of logs). You looking through 4-5 logs out of 100's of thousands is pretty much as skewed as me looking through my logs.

Now... JS should never have 31.3% uptime. It has a ICD that prevents that from happening, UNLESS JADE SPIRIT IS ALSO BUGGED. Windsong does not have an ICD.

---------- Post added 2012-10-22 at 12:39 PM ----------

Makes you wonder if windsong is working as intended or if we can expect to see a nerf soon.

I expect to see a nerf soon. The way a resto shaman in 25m gets procs from it make it really OP.

[huth]

I'm confused. I read nothing about "the proc chance going down with more events per second."

It keeps track of the last time it had a chance to proc for that enchant.
It calculates the difference in time since the last chance to proc.
It uses that time to determine the chance for that event to trigger a proc.

Where: Chance to proc = event. The formula given then shows that increasing number of events leads to a lower proc chance.
The rest is simple math.
If you have 44 events in 10 seconds, you get an average chance to proc of 0.8%. If you then calculate the total chance to get a proc over a 10 second window, you will find that it keeps getting lower the more events you have per timeframe.
No idea why it was designed that way, but that's the info Blizz gave us.

Which brings us to...

I can understand if you're trying to say your average proc chance per event goes down with a lot of events.
However, that does not matter, what matters is your actual proc rate each second because that leads us to the important up-time, or average proc rate per time.

If Blizzard wanted to implement this.... lower each individual chance to proc with increased number of events, it would be the right direction in balance. My logs however show that people with more events per second will have radically more procs per time.

That's literally the whole point of the system. Which is why i said your findings might show that there is a bug.

[Windry]

@huth

Oh ok, thanks for the post. Then ya, I would say it's probably bugged.

[rhandric]

@huth: event refers to the damage/healing being done that has a chance to proc the buff. If the math in the blue post is accurate (p=PPM*haste*time since last proc/60 [sec]), then you're not being explicitly penalized for having more events (chances to proc), you by design have a higher chance: 1-(1-p)^n is your actual chance to have a proc with n events over some time with some amount of hast, integrated over time from 0 to some time t up to 10, where p becomes a constant.

So, it's not that you get fewer the more events you have per timeframe, it's that you have a lower chance of getting a proc the more recently a proc has occurred.

[Tickspoon]

For N events/minute, equally spaced, the overall chance would be 1 - (1 - 1/N)^N. Huth is technically right that this function decreases as N increases. It converges pretty quickly though (to 1-1/e =~ 63%) so may as well be constant.

[huth]

I find it somewhat confusing that rhandric is trying to explain to me how i defined the term event in regards to this discussion.

Besides, i never claimed that it drops by huge amounts in the relevant areas. Just that it does drop, which it indeed does.

[rhandric]

Sorry, I misread the blue post and thought you were defining event from it, when it never actually mentions event.

[Jimm3]

btw shouldnt the shado pan revered trinket be in that trinket list somewhere?

[Deleted]

@jimm3 good question. in that same thinking it should be noted that the Jade Magistrate Figurine has the exact same stats. our guild already had 2 drops ==)==)

also along the lines of conquest gear. i find that both wrist cloak and ring is the better choice early. since its going to take a long time to get valor points for all the items, and we are prolly well into next tier before we have rep/valor to buy it. i know the ring is quest item but its mastery.. and i am waay to tired of all the dailies to do klaxi exhalted for that little gain(if any tbh havent done exact math).

Ps. could we please stop using Wol parses for Gara'jal as baseline for any discussion? unless both the one logging and the dps/healer are not in spirit realm at all, it is completely useless and might be why some one saw 30%+ uptime on jade spirit.
its so much easier to just pick any of the other 5 fights ..

[dancm]

If you have 44 events in 10 seconds, you get an average chance to proc of 0.8%.

Really?

Let's have a look...

That's the given formula:

For example, if you have 22% Haste, it was 1.4sec since the last chance to proc, and you’ve got Windsong, then the chance to proc is 2(ppm) * 1.22(haste) * 1.4(time since last chance) / 60 (sec per min) = 5.693%.

Now I'm assuming that those 44 events are evenly distributed across that 10s period. For ease of calculation (I'm lazy) I'm assuming zero haste.

That leads to an event every 10/44 s = 0.22727272... s

So the first event's proc chance will be 2(ppm) * 1 (zero haste) * 0.227 (time rounded down) / 60 = 0.0075666... = 0.75666... %

So the very first event during the 10s period will already have a proc chance almost equal to your average proc chance. Assuming very bad luck you'll reach the final proc chance which will be 2 * 1 * 10 / 60 = 0.333... = 33.3%. I actually don't know how you got an average of 0.8%

[huth]

2 * 10/44/60 = 0.00757575... ~ 0.008 = 0.8%

[dancm]

2 * 10/44/60 = 0.00757575... ~ 0.008 = 0.8%

So you're assuming that all 44 events happen at the same time exactly 10/44s past the last successfull proc and none of these events happens within the rest of the 10s period (430/44s).

Forget what I wrote... My fault. I misread 'time since last chance' as 'time since last proc'. So your average is correct.