[5.3] Affliction Warlock and Encounter Specific Guides

[Surutcra]

Concerning the Haste vs. Mastery discussion, you really should state clearly which variable you are trying to "optimise", because the issue you're discussing is not well-defined otherwise !

Average DPS is usually the default measure, and the one used by Simcraft (assuming accurate modelling, which is a different issue here). At constant average DPS, a haste heavy build will have lower DPS standard deviation than a mastery heavy one. This is because even though RPPM remains highly random at high levels of haste, getting more haste does indeed increase the expected number of procs and thus reduce the standard deviation of your DPS.


Now let's look at alternative measures and their effects on your haste/mastery preference :
- Median DPS (you get that DPS or better 50% of the time) : hard to tell, probably close to zero effect.
- 10% percentile DPS (you get that DPS or better 90% of the time) : you're being risk-averse (you want to maximise your DPS in a "worst-case" scenario), this shifts the balance in favor of haste.
- 90% percentile DPS (you get that DPS or better 10% of the time) : you're being risk-seeking (you want to maximise your DPS in a "best-case" scenario), this shifts the balance in favor of mastery (in defavor of haste, in fact).


Now this gets to more of a personal opinion on the matter :
- If you are in a progression context, there is nearly no situation where you want to be risk-seeking. And to be completely fair, your raid as a whole might be risk-averse, but your contribution to the global standard deviation is so small (in 25H at least) that you probably can be risk-neutral in most cases (i.e. listen to Simcraft).
- If you are in a farming context and want to maximize your chance at the best possible WoL rank, you should definitely be risk-seeking, and pray you get lucky ;-)


Important note : the difference in DPS standard deviation for both haste/heavy mastery builds is not insignificant imo, but keep in mind fight specific conditions favoring one or the other stat will flat out make them irrelevant. For example, even if you want to play risk-averse on a fight where mechanics heavily favor mastery over haste (say Durumu H, Destruction), you should definitely stack mastery in spite of everything I just mentioned.

[Sparkuggz]

Concerning the Haste vs. Mastery discussion, you really should state clearly which variable you are trying to "optimise", because the issue you're discussing is not well-defined otherwise !

Average DPS is usually the default measure, and the one used by Simcraft (assuming accurate modelling, which is a different issue here). At constant average DPS, a haste heavy build will have lower DPS standard deviation than a mastery heavy one. This is because even though RPPM remains highly random at high levels of haste, getting more haste does indeed increase the expected number of procs and thus reduce the standard deviation of your DPS.


Now let's look at alternative measures and their effects on your haste/mastery preference :
- Median DPS (you get that DPS or better 50% of the time) : hard to tell, probably close to zero effect.
- 10% percentile DPS (you get that DPS or better 90% of the time) : you're being risk-averse (you want to maximise your DPS in a "worst-case" scenario), this shifts the balance in favor of haste.
- 90% percentile DPS (you get that DPS or better 10% of the time) : you're being risk-seeking (you want to maximise your DPS in a "best-case" scenario), this shifts the balance in favor of mastery (in defavor of haste, in fact).


Now this gets to more of a personal opinion on the matter :
- If you are in a progression context, there is nearly no situation where you want to be risk-seeking. And to be completely fair, your raid as a whole might be risk-averse, but your contribution to the global standard deviation is so small (in 25H at least) that you probably can be risk-neutral in most cases (i.e. listen to Simcraft).
- If you are in a farming context and want to maximize your chance at the best possible WoL rank, you should definitely be risk-seeking, and pray you get lucky ;-)


Important note : the difference in DPS standard deviation for both haste/heavy mastery builds is not insignificant imo, but keep in mind fight specific conditions favoring one or the other stat will flat out make them irrelevant. For example, even if you want to play risk-averse on a fight where mechanics heavily favor mastery over haste (say Durumu H, Destruction), you should definitely stack mastery in spite of everything I just mentioned.

Mm Suru, you are right about this. But thats where i say when our sample size is 20 or 100 attempts even for a whole tier, including RPPM RNG and mechanics there is no stats you can justify being superior or increasing your DPS. Mastery would most likely scale better with the RNG factor of RPPM versus haste. Haste becomes good for 10.000 tries, and RPPM mechanics scales haste up where it doesnt belong. Without RPPM mechanics haste and mastery is close to the same values, meaning haste is still good. But that's where if you have a sample size of 20 attempts instead of 10.000 i'd rather have higher lowest DPS and higher top DPS then flat out doing 200 more DPS over 10.000 tries because the median DPS goes up a bit. Until there is huge shifts of DPS its just within the error margain that simcraft has, thus we cant use it for much. 200 dps out of 220k is too "small" to not be under some errors and simming issues, where as 2k DPS. If you value haste RPPM vs mastery, simc overvalues the haste to the thousands instead of the hundreds it belongs.

I remember simming my warlock with 92xx haste cap vs 9778 haste cap vs more haste cap pre progress, and the DPS difference was within 20-50 DPS, becuase simc valuated the Haste vs Mastery at equal levels, but with RPPM it puts it like God vs human. Then on top of that the fact there is no fights in ToT that really shines haste, since pretty much any fight has a mechanic or an add / adds that would fuck with these numbers further.

RPPM + small sample size = impossible to justify anything

[Gremory1]

Mm Suru, you are right about this. But thats where i say when our sample size is 20 or 100 attempts even for a whole tier, including RPPM RNG and mechanics there is no stats you can justify being superior or increasing your DPS. Mastery would most likely scale better with the RNG factor of RPPM versus haste. Haste becomes good for 10.000 tries, and RPPM mechanics scales haste up where it doesnt belong. Without RPPM mechanics haste and mastery is close to the same values, meaning haste is still good. But that's where if you have a sample size of 20 attempts instead of 10.000 i'd rather have higher lowest DPS and higher top DPS then flat out doing 200 more DPS over 10.000 tries because the median DPS goes up a bit. Until there is huge shifts of DPS its just within the error margain that simcraft has, thus we cant use it for much. 200 dps out of 220k is too "small" to not be under some errors and simming issues, where as 2k DPS. If you value haste RPPM vs mastery, simc overvalues the haste to the thousands instead of the hundreds it belongs.

I remember simming my warlock with 92xx haste cap vs 9778 haste cap vs more haste cap pre progress, and the DPS difference was within 20-50 DPS, becuase simc valuated the Haste vs Mastery at equal levels, but with RPPM it puts it like God vs human. Then on top of that the fact there is no fights in ToT that really shines haste, since pretty much any fight has a mechanic or an add / adds that would fuck with these numbers further.

RPPM + small sample size = impossible to justify anything

I think saying that since there is random chance in the game, and you only ever kill a boss 20 times, that therefore it's impossible to justify anything and there's no point trying to min-max, is not quite true at least from my POV. Also, if stacking haste increases the median DPS you can do and stacking mastery instead increases the variability of your DPS, then you're not going to have 'higher lowest DPS and higher top DPS' by stacking mastery. You're going to have LOWER lowest DPS and (as you say) higher top DPS.

I'm am, however, completely on board with the concept that fights in ToT don't represent simcraft scenarios well and therefore mastery might actually outperform haste (and I am in fact sticking with the 9778 breakpoint) but just because RPPM increases RNG in a fight (as that rogue trinket picture showed) and you don't end up with a statistically relevant sample size, doesn't mean that you won't do higher average DPS with a setup that sims to higher DPS over 10k tries. If you want to roll the dice, on the other hand, and maximize the potential DPS and hope that the 20 attempts you do are in that lucky few that do huge DPS then that's fine too.

[Sparkuggz]

I think saying that since there is random chance in the game, and you only ever kill a boss 20 times, that therefore it's impossible to justify anything and there's no point trying to min-max, is not quite true at least from my POV. Also, if stacking haste increases the median DPS you can do and stacking mastery instead increases the variability of your DPS, then you're not going to have 'higher lowest DPS and higher top DPS' by stacking mastery. You're going to have LOWER lowest DPS and (as you say) higher top DPS.

I'm am, however, completely on board with the concept that fights in ToT don't represent simcraft scenarios well and therefore mastery might actually outperform haste (and I am in fact sticking with the 9778 breakpoint) but just because RPPM increases RNG in a fight (as that rogue trinket picture showed) and you don't end up with a statistically relevant sample size, doesn't mean that you won't do higher average DPS with a setup that sims to higher DPS over 10k tries. If you want to roll the dice, on the other hand, and maximize the potential DPS and hope that the 20 attempts you do are in that lucky few that do huge DPS then that's fine too.

Mm, i know but you can simulate and compare lowest and top DPS you do with mastery vs haste, the DPS shift is towards the middle curve where you have your average dps tries and not your lowest or highest. It doesnt shift down and give you a lower lowest output of damage, the 10.000 tries play around the middle area and increases your DPS overall slightly in the mid-field average DPS, not the top or bottom burst / damage! That's why having a smaller sample size wont give you a lower min DPS, thus why im saying that it wont be noticeable.

BIS single target is Sac, and Sac exactly scales the way that your top DPS parse is higher and your lowest is higher, but over 10.000 tries you shift very so slightly towards higher overall DPS.

That being said, this is _WITH_ RPPM trinkets in the equation, so if you exclude RPPM trinkets from it all we come back to mastery and haste being equal ~~ within reasons and then all im saying is what do you want? You sim ~50-100 dps higher going full haste vs mastery with 0 RPPM mechanics in the equation and that small DPS increase 50-100 out of 220k+ dps is within an error margain that you cant conlude anything out from. Then let's say Haste is Equalish to mastery, then add RPPM trinkets but not add the +haste value simcraft gives because of 10.000 tries, you will never human possible notice. You still in a smaller amount of tries increases your chance to get top dps tries while haste limits you, but increases your dps by a few hundred if you kept pulling and compared your DPS over 10.000 tries.

Without RPPM mechanics haste = mastery, with mastery bringing higher top parse, higher lower parse but over 10.000 tries give you higher DPS around the average DPS levels.

With RPPM mechanics, Haste > Mastery, because it fakely say on 10.000 tries your trinket will increase with 0,000001% and thus putting your overall DPS higher.


Saying mastery is a risk and will lower the output in amount of tries you have is what im trying to say i dont totally agree with. But yes, if you want to rank on farm, heavy mastery is the way to go for sure, since the game is so RNG at this moment stat weights and min maxing has become impossible.

Anyways point with it all isnt saying mastery is better then haste, but just trying to let people decide what do they want to aim for. Haste giving you top dps parses or show-off on farm is counter productive and that simcraft is slightly odd and cant really be used in Throne of Thunder.