Anaxie's Totally Revolutionary SoO Retribution T16 BiS & Theorycraft

[Maegor]

The weights of mastery and crit don't catch up to haste at 40%, it actually happens at several spots over 42ish% haste because of local plateaus. The reason you stop at 40% is because you can't get another tick in of HoW until 50% (only for SW specced). And another thing you are misstating meiffert, is that the more of a stat you have, the less valuable it is compared to the other two (haste is somewhat of an exception because of the previously stated plateaus), which is the opposite of what you said in your post. So stacking mastery makes crit more valuable and vice versa.

The other thing you keep saying is that mastery isn't more consistent. This really depends on your definition of consistent. If consistency = low range, then yes saying crit is more consistent is correct. However, if you are measuring consistency by being able to meet a dps threshold (which in a game where you are killing shit, this should be your measure), then mastery is going to be more consistent. This is because mastery raises your baseline dps, and in the crit heavy and mastery heavy builds, the mastery build has a higher low, avg, and top end dps value.

What we really need to be able to 100% put this argument to bed is data that simc doesn't provide, as stated in my post a few pages back where I ran the sims.

Doesn't matter too much though, in 3-5 months we will be back here arguing about 6 stats instead of 3. Fun times ahead!

[Meiffert]

The weights of mastery and crit don't catch up to haste at 40%, it actually happens at several spots over 42ish% haste

Well it will change depending on how much secondary stats you have overall (the treshold will increase with your ilvl). That being said I believe 42 % falls into "somewhere around 40".
Thanks for specifying the number for close to BiS as well as stating the extra HoWs caps though.

And another thing you are misstating meiffert, is that the more of a stat you have, the less valuable it is compared to the other two (haste is somewhat of an exception because of the previously stated plateaus), which is the opposite of what you said in your post. So stacking mastery makes crit more valuable and vice versa.

This is exactly what I thought I said.

The other thing you keep saying is that mastery isn't more consistent. This really depends on your definition of consistent. If consistency = low range, then yes saying crit is more consistent is correct. However, if you are measuring consistency by being able to meet a dps threshold (which in a game where you are killing shit, this should be your measure), then mastery is going to be more consistent. This is because mastery raises your baseline dps, and in the crit heavy and mastery heavy builds, the mastery build has a higher low, avg, and top end dps value.

Well yes, I agree with your definition of consistency and I know that the heavy crit build you used had weak low end.
But the reason for that was more the build being somewhat terrible in the first place (only around 35 % haste, way overboard on crit). When you look at it closer, the low-end wasn't actually that far away from average. It was only so bad because the average was bad. I don't think that anybody is considering such a build.

I believe that these builds are more reasonable (slightly over 40 % haste, than either mastery or crit):
Mastery


Crit


These are actually reasonable builds where I'm close enough to optimal average dps to consider using them. And the crit build has low-end about 5k closer to average than the mastery build.

Doesn't matter too much though, in 3-5 months we will be back here arguing about 6 stats instead of 3. Fun times ahead!

Looking forward to it!

[macstallion]

Looking at all of this back and forth talk it's simply put at this: Mastery and crit are in all effectiveness of our raids equal. Crit does require a bit more rng (what isnt rng about this game) and mastery in all thoughts is a higher avg. This is being said about gear at a high ilvl 570+. On the other hand me and a fellow ret raider put some testing to 50% haste vs 40% haste and conclusions from our in raid experiments and visual simc to throw into our opinions on the matter are as following:

Conclusion:
Single target haste wins, AOE mastery wins. Immersues and protectors (stack strat) I won. Other single target fights he won where there wasn't much AOE. However, to conclude this RNG being a factor such as galakras he was close to me on meters due to just having more 4 piece procs and lucky crits.
Over a huge pool of simulated fights they'll even out. On the most part for very few pulls on some fights It's completely RNG. If you're going for a parse on X fight a certain build is better and then rng comes into play and you go for the parse.

Simulations:
Simmed raiders (Solsacra, Ujx, Anaxie, Macstallion, Lightswrath (guildy)) and just stating everyone is 575+ ilvl. Everyone simmed 360-380k dps from 40% haste + mastery build to 50% haste build. When I was 40% haste + mastery build with my weaker itemization I simmed to 340k dps. I just went 50% haste to play with it on fresh clears. Now simming at 350k dps. However, this does take into account the effect mastery has on increasing AoE dmg on specific fights. I myself have the worst luck on off pieces and a few tier pieces and have been thrown off on itemization compared to the others. I'm using malk neck, normal tier shoulders, normal non WF garrosh legs.

Sidenote:
My inactivity on this forum may show inexperience. However, I do talk to other rets via ingame rather than on a forum and have gone 14/14H with my guild and stream for my guild to demonstrate a ret PoV for the ret community.

[Huntingbear_grimbatol]

Think the mistake often done is that math (simc) more or less never translate to a raiding scenario well. Use simc to find stat weights, sure... then sim if you have multiple targets often, or if your main duty is to nuke a mob every 100sec which you always have wings up for. Not saying that mathy people aren't doing something for this community but know the limitations, iirc Simc for ret was wrong for months early in MoP and it caused so much bad to go around.

also welcome to the thread Macstallion.

[macstallion]

Think the mistake often done is that math (simc) more or less never translate to a raiding scenario well. Use simc to find stat weights, sure... then sim if you have multiple targets often, or if your main duty is to nuke a mob every 100sec which you always have wings up for. Not saying that mathy people aren't doing something for this community but know the limitations, iirc Simc for ret was wrong for months early in MoP and it caused so much bad to go around.

also welcome to the thread Macstallion.

Thanks for the welcome Bear.

Yes this is exactly what my conclusion was saying. I simmed for stat weights. Which shows base dmg, but when in a raid environment where there are different strats (example: Sha of Pride: You can stack up for adds or spread out and deal with them) this can throw a curveball into stats playing a role in this respective fight. Mastery being better for the stacking method cause you do more AoE damage and if you spread out and single target is more important than 50% haste and under my assumption (no math backing this part up) crit would prob be higher.

[Swampmoose]

Think the mistake often done is that math (simc) more or less never translate to a raiding scenario well. Use simc to find stat weights, sure... then sim if you have multiple targets often, or if your main duty is to nuke a mob every 100sec which you always have wings up for. Not saying that mathy people aren't doing something for this community but know the limitations, iirc Simc for ret was wrong for months early in MoP and it caused so much bad to go around.

also welcome to the thread Macstallion.

They were wrong early in Wrath with Strength > Hit because of the assumption of operating on a single roll system and it was found later on that was incorrect. They will be wrong again at some point.

[JemiS]

Ok..... So I'm confused. I hadn't simmed myself in a while, and did a 10k iteration sim last night.

Here's what my scale factors came out as:

Str: 5.56
Crit: 3.48
Mastery: 3.33
Haste: 3.07

And I'm only at about 35% haste. Thought it might just be a scaling thing, but I did a stat scaling plot, and it came out favoring mastery and crit over haste +/- 1000 points.

Am I doing something wrong with simcraft? Or should I be favoring haste less?

Apparently I can't post the plots or links to my profile with this few posts, but it's Jemi / Terenas.

[Deleted]

Ok..... So I'm confused. I hadn't simmed myself in a while, and did a 10k iteration sim last night.

Here's what my scale factors came out as:

Str: 5.56
Crit: 3.48
Mastery: 3.33
Haste: 3.07

And I'm only at about 35% haste. Thought it might just be a scaling thing, but I did a stat scaling plot, and it came out favoring mastery and crit over haste +/- 1000 points.

Am I doing something wrong with simcraft? Or should I be favoring haste less?

Apparently I can't post the plots or links to my profile with this few posts, but it's Jemi / Terenas.

It's because of your haste proc trinket. When the proc is active, you are over the 50% cap (1 sec gcd). That decreases the value of haste because it's much weaker when it doesn't reduce the gcd any more. Solution: Try to get a Thok's tail / an evil eye. Even lfr versions are SO much better.

PS: You reforged +468 exp. Use exp / haste gems instead of str / haste. 160 secondary stats > 80 str.

[Deleted]

They were wrong early in Wrath with Strength > Hit because of the assumption of operating on a single roll system and it was found later on that was incorrect. They will be wrong again at some point.

And how do you figure out the error was discovered? You see a discrepancy between the simulation and what is actually happening. Then you adjust your variables, paramteres and change the theory to include said observation no? You make it sound like Theck, Exemplar and other theorycrafters aren't trying things out ingame. Using the phrase "they were wrong" makes you look ignorant. If Blizzard decided to give us their entire code for how their combat system works, everything could be plotted, but that would also take away some of the fun wouldn't it?

@Meiffert: I'm glad to have you onboard, hope you stick around for the expansion. We could REALLY use some more math geniouses. Most people here seem to simply dismiss it with either "you won't do 50k tries" or "you can't sim this fight". Like getting the highest average performance is a bad thing because average is only good in the long run which is unrealistic. Haven't heard a dumber thing in a long time.

[Swampmoose]

And how do you figure out the error was discovered? You see a discrepancy between the simulation and what is actually happening. Then you adjust your variables, paramteres and change the theory to include said observation no? You make it sound like Theck, Exemplar and other theorycrafters aren't trying things out ingame. Using the phrase "they were wrong" makes you look ignorant. If Blizzard decided to give us their entire code for how their combat system works, everything could be plotted, but that would also take away some of the fun wouldn't it?

@Meiffert: I'm glad to have you onboard, hope you stick around for the expansion. We could REALLY use some more math geniouses. Most people here seem to simply dismiss it with either "you won't do 50k tries" or "you can't sim this fight". Like getting the highest average performance is a bad thing because average is only good in the long run which is unrealistic. Haven't heard a dumber thing in a long time.

I don't know what you're getting or why you seem so angered by my comments but it's not ignorant to state the fact that they are indeed wrong from time to time. I am also not making it sound like they aren't testing things in-game. Please point out where I said anything that would indicate that. I am not faulting anyone for making mistakes. I was simply pointing out another example that they can and will be made.

And yes, as anyone should do, including those crafting such theory, you need to compare simulations vs actual data to pinpoint errors.

[Maegor]

Well yes, I agree with your definition of consistency and I know that the heavy crit build you used had weak low end.
But the reason for that was more the build being somewhat terrible in the first place (only around 35 % haste, way overboard on crit). When you look at it closer, the low-end wasn't actually that far away from average. It was only so bad because the average was bad. I don't think that anybody is considering such a build.

I used BiS gear list to keep things on an even playing field, which is why I also had the haste at the same value for both builds. I could have done true crit and mastery builds by choosing gear to maximize those stats, but that would have been providing false data in reality, because haste would not have been a controlled variable in those situations (if one setup had more haste, it would naturally sim higher). The question was to see if going crit or mastery was better and more consistent. In terms of being able to meet a damage threshold, mastery was shown to be more consistent from the data simc provides (and yes, I wish it spit out more data so we could calculate std dev and variance, instead of the "range" value it spits out). You can't really say it was "way overboard on crit", because that was the test. We know it was way overboard on crit, the mastery build was way overboard on mastery. It is why I showed that a pure haste build sims better than both (something we all knew), and that the stat weight balanced build simmed better than both (something EVERYBODY should have known intuitively).

I also don't think using BiS list is unreasonable either. It is a list that is available so you should be able to reproduce similar results, and it is an obtainable list of gear. Btw does Solsacra have all these pieces? His gear is insane.

I believe that these builds are more reasonable (slightly over 40 % haste, than either mastery or crit):
Mastery


Crit


These are actually reasonable builds where I'm close enough to optimal average dps to consider using them. And the crit build has low-end about 5k closer to average than the mastery build.

How much of a swing of crit and mastery did you have from each of these builds? Did you actually go in and change reforges and gems around or just add to their values in simc like you did previously? 50k sims (since 50k will usually provide the smoothest curves)? Can you post stats with each build?

- - - Updated - - -

And yes, as anyone should do, including those crafting such theory, you need to compare simulations vs actual data to pinpoint errors.

And this is why so many people say to stop at 40% haste. They know more haste sims higher, but in actual practice there are factors both in your hands (rotation crispness) and out of your hands (lag) that cause too much haste to behave less favorable in real world applications. But Felic always bucked at the 40% thing when a lot of people (Bear, Anaxie, Solsacra) said it worked better in game than going full haste.

Not that questioning accepted theory is bad, shit if nobody questioned accepted theory we would still be living on a flat planet in the center of the universe. Just that at some point, you have to go see what works best for you, especially in the case of 40% haste.

[JemiS]

It's because of your haste proc trinket. When the proc is active, you are over the 50% cap (1 sec gcd). That decreases the value of haste because it's much weaker when it doesn't reduce the gcd any more. Solution: Try to get a Thok's tail / an evil eye. Even lfr versions are SO much better.

PS: You reforged +468 exp. Use exp / haste gems instead of str / haste. 160 secondary stats > 80 str.

Yeah, I figured that was the case after the fact. And trinket drops hate me. 14 Gala kills with coins between LFR and Flex, and 9 Thok kills + coin between LFR, Flex and Normal. Gotten every other piece off the loot table (some multiple times), but no trinket on either. 46 chances with no trinket is starting to get really old.

I should go back and try to get the ToT N/H trinkets, I just keep hoping one of the T16 ones will drop for me. I can just swap in an old trinket (Feather or Spark from T15LFR) and sim my stat weights without the proc to get a better idea.

I'm using the reforge instead of haste gems, as it gets me closer to just over the cap with less wasted stats.

3x 160 expertise = 480 instead of 468, meaning I'm another 12 expertise over cap. ATM I'm at 7.53, this would put me at ~7.57 expertise. I've play around with the reforging to get it there, but with 22 expertise "wasted", I was guessing it was better to use the STR/Haste gems.

[Deleted]

Yeah, I figured that was the case after the fact. And trinket drops hate me. 14 Gala kills with coins between LFR and Flex, and 9 Thok kills + coin between LFR, Flex and Normal. Gotten every other piece off the loot table (some multiple times), but no trinket on either. 46 chances with no trinket is starting to get really old.

I should go back and try to get the ToT N/H trinkets, I just keep hoping one of the T16 ones will drop for me. I can just swap in an old trinket (Feather or Spark from T15LFR) and sim my stat weights without the proc to get a better idea.

I'm using the reforge instead of haste gems, as it gets me closer to just over the cap with less wasted stats.

3x 160 expertise = 480 instead of 468, meaning I'm another 12 expertise over cap. ATM I'm at 7.53, this would put me at ~7.57 expertise. I've play around with the reforging to get it there, but with 22 expertise "wasted", I was guessing it was better to use the STR/Haste gems.

160 exp is ~40-50% better than 80 str, so you can go ~50exp over cap per gem.

[Swampmoose]

And this is why so many people say to stop at 40% haste. They know more haste sims higher, but in actual practice there are factors both in your hands (rotation crispness) and out of your hands (lag) that cause too much haste to behave less favorable in real world applications. But Felic always bucked at the 40% thing when a lot of people (Bear, Anaxie, Solsacra) said it worked better in game than going full haste.

Not that questioning accepted theory is bad, shit if nobody questioned accepted theory we would still be living on a flat planet in the center of the universe. Just that at some point, you have to go see what works best for you, especially in the case of 40% haste.

Agreed. Anytime you have a stat that has a value which can be immediately altered by something like player input or lag, a red flag should go up saying, you probably want to test this yourself. You can run an infinite number of sims but if your abilities as a player are far off from simulation, those sim results don't have much value for a stat like that.

You can lean on sims more for stats like crit which are just a roll of the dice, but even the values of those stats can be altered somewhat by a player if they butcher things enough. You're also assuming the sim is properly calculating damage as the game would (back to Str > hit example).

Its difficult to tell people to see what works best for them when you have stats that are very close to one another in value (crit vs mastery as an example). It's just impossible to do enough self evaluation in a raid setting to reliably say X > Y. Even if you killed Heroic Boss A once every week for a year, you only have 52 tests which you need to split between the two stat priorities.

[Neldarie]

It's actually quite the opposite : Mastery better for ST while Crit, relative to each other (hi cloak, hi no hammerwrath spam) stronger for AoE.

[Elamahpla]

inq ds_4t16_5hp inq tv5 inq how inq tvaw inq cs inq j ds_4t16 inq exo tv3 inq tv4 (e: SW, not DP, but if someone could throw me a sweet DP one thatd be sweet)

what im using for clcinfo. can someone tell (confirm) if this is the proper thing i should be using?

also, wtb more ret paladin friends on btag to talk ret paladin things and be kawaiiiiiiiiiiiiiii together.

[Maegor]

It's actually quite the opposite : Mastery better for ST while Crit, relative to each other (hi cloak, hi no hammerwrath spam) stronger for AoE.

If DS crits does it crit everything?

- - - Updated - - -

inq ds_4t16_5hp inq tv5 inq how inq tvaw inq cs inq j ds_4t16 inq exo tv3 inq tv4 (e: SW, not DP, but if someone could throw me a sweet DP one thatd be sweet)

what im using for clcinfo. can someone tell (confirm) if this is the proper thing i should be using?

also, wtb more ret paladin friends on btag to talk ret paladin things and be kawaiiiiiiiiiiiiiii together.

I would personally clean up the inq's. Clcret is priority based, so if you are at 3 hp and need inq, it will jump it to the front. But that is really more housekeeping than anything. I also move my ds_4pc up in the order to match OP's placement of ds_4t16_aw, and just adjust rotation for when aw isn't active.

[Deleted]

I reforged all my gear to haste40%>crit>mastery for 2 weeks now (still keeping mastery sockets and stuff untouched) and I'm seeing quite funny results in raid. I lost about 10-12% mastery but gained about 8% crit, leading to 35% crit raidbuffed. The first result is that me and my ret buddy in raid still perform almost the same (small differences in gear putten aside). The second result is that on pull I'm able to put out far more burst dps than in mastery form, sometimes even bypassing our rogues for a short time (+800k DPS FINALLY YEAH!). I'm also still doing well on the siegecrafter belt. For me, I don't see any major differences or even disadvantages. I'll stick with crit for now, maybe even consider going full crit with gems and test that.

[cremor]

inq ds_4t16_5hp inq tv5 inq how inq tvaw inq cs inq j ds_4t16 inq exo tv3 inq tv4 (e: SW, not DP, but if someone could throw me a sweet DP one thatd be sweet)

what im using for clcinfo. can someone tell (confirm) if this is the proper thing i should be using?

As already said, all that inqs except the first one are useless. tv4 behind tv3 is also useless.

The following string was build after the priority list used in SimulationCraft and works for both SW and DP:

inq inqdp ds_4t16_5hp tv5 tvdp_2s how ds_4t16_aw tvaw cs j ds_4t16 tvdp exo tv3

You need to rename the rotation.untested file to rotation.lua in the latest version of clcInfo to get the ds_4t16_aw and tvdp_2s commands.

[Weightlifter]

also, wtb more ret paladin friends on btag to talk ret paladin things and be kawaiiiiiiiiiiiiiii together.

I am the most kawaii paladin EU :3