Ahhh you're right. I was doing the math the long way by adding up the actual numbers from recount (while only taking 94% of pyroblast and fireball damage) and comparing that to the damage I actually did. I must have made a mistake in adding,

2.9% is even better of a result, I guess factoring in ignite and combustion would put it roughly at 4% then. Sorry for the misunderstanding then.

Slot swapping trinkets does not reset the INTERNAL cooldown, it simply triggers a one-minute cooldown where it will not function (whether on-use, or proc).

Would a glyphed Frostfire rotation be worth considering, due to the upcoming Fireball/Pyroblast nerfs?

Fireball and Pyroblast both received large buffs in 4.3. Even with a 6% nerf, they are still more powerful than they were and will still outperform FFB.

I will attempt to make a general but confident assessment of the current situation, having as a reference a random and common example of a bell curve produced by SimCraft on DPS values collected from its multiple runs:



We can graphically see that the distribution of max and min DPS values is very wide compared to predictable specs like Arcane Mage. Other specs do become unpredictable due to crit or other random procs but it appears Fire has increased crit-based unreliability due to reliance on Ignite.

What is most important on the curve is that the distribution is so wide that to reach the max value, one has to go through a plateau, meaning the max value is incredibly rare.

For this we can derive two very general but confident conclusions: One, the log based rankings most players consult with will not be representative of the class (this will include the fact that a kill may not happen at all or a log may not be uploaded if fire mages were unlucky), and that fire spec is very unreliable for end-game raiding or otherwise (in its damage output) .

This is a simple realization, however, it is complex to fully analyse what it means for the end user of the spec and it would probably require a whole book. This is because we could go on here about specific situations affected by the handicaps of Arcane related to movement or mana, or the advantages/disadvantages of any other spec but this is highly depended on specific fights or personal/guild strategies.

What we can keep from this is 1. The fire spec is a very unreliable spec in its damage output, perhaps the most unreliable in the game while Arcane is very reliable comparatively, perhaps the most reliable in the game and 2. One has to now decide if that unreliability is worth to be kept in exchange for the reduced damage, while also factoring in specific fights and guild strategies.

I think you have this wrong slightly. If you look at arcane dps distribution you'll see that it has the exact same shape, the only difference is that fire has slightly higher variance. Arcane still has a large spread compared to other classes though. This is in part due to the legendary staff, but also things like crits during cooldowns.

I wouldn't go so far as to say that log based rankings are unrepresentative. It's true that there will be a slight shift due to different variance, but I don't think the effect is as big as you make it out to be.

Overall I think you're mixing two very different things, dps variance and dps control. In terms of dps control fire and arcane are on far opposites, but on dps variance they're not so far apart.

I should point out that I noticed haste-rich builds from Rawr gave a much smoother and stable distribution. The more crit-based ones that some players started using by intuition and SimCraft supported (that were often also supported by Rawr at a later point) gave the more unpredictable results. Namely I would often see the 'jumps' from a mean of 45k to a 57k for very infrequent samples, if crit was high.

I'd advise you to compare multiple specs and their dps range, instead of calculating it just for Fire Mages and guessing from there that it is much higher than others.

First, use the average_range=0 option when running the simulation, so that the base damage values of direct spells don't get averaged. Then run a Raid sim with the option report_rng=1 and checkout the ranking for it found under Raid Summary. Keep in mind that it will report the range as ( dps_quantile(0.95) - dps_quantile(0.05) )/2 to give more consistent results. Oh and don't forget to run the whole thing with many iterations, maybe 100'000 or more. If you want to really observe max-min range, then you have to look under the statistics analysis of each players, where you'll find it.

My whole point being: fire mages might be the mages with the biggest variance, but there are other specs even worse and others about equally bad. Surely high rng can cause problems, but even the smoothest classes only have about half as much rng as a fire mage.

Your point about rng influencing logs is quite important. This is a fact that most people - unfortunately - always forget or ignore. It is also one of the main reasons of complain that a sim is wrong because eg. the top logs on Wol do x + 4000 dps, where on SimulationCraft they do only x dps, that's clearly a SimC bug and not luck playing a part in wol rankings!

This is a thread about progression raiding, not normals. On Heroic Hagara almost every guild in the world is having all ranged in the center, with zero to little movement. The reason fire is better than arcane on hagara is because of Ice Tomb damage, not because of movement during a phase when ranged barely move.

On a side note, I'm pretty sure a mod is going to come through here soon and give out multiple infractions/warnings to some of players above. This is a progression thread, not a basic question/hand holding thread. Be careful what you ask, and be careful what you answer, or you may, and will receive infractions and warnings. Re-read the rules for posting on these threads.

I wouldn't even say just Ice Tomb damage. The 15 seconds of feedback on Hagara in between phases are enough (bar some really bad RNG) to land a huge Combustion; Arcane burst can't exactly compete especially because during Lightning Phases you should be sitting at 0 stacks of AB when the phase ends.

I'm guessing you're talking about 25 man? Impacting Combustion onto 6 Ice Tombs in 25 man heroic mode would be a lot more dps than Impacting onto only 2 in 10 man heroic mode. Probably still a good idea in 10 man now that you mention it. Previously I'd been thinking that the best time to use it was during Feedback, but that does have problems with missing trinket stacks and the very short amount of time to line up your Combustion in.