Surrender to Madness Simulation and AMR Stat Weights

[Swol]

Hey all, I'm Swol, the main theorycrafter at AMR. I wanted to stop by and post some information regarding S2M simulation and the new gearing strategy that we are recommending for shadow after the Nov 14th hotfix.

First I'll talk about the gearing strategy. You can find our new weights for Ursoc-style fights here:
http://www.askmrrobot.com/wow/simula...3?weights=true

As you can see, I'm finding Crit, Haste, Intellect, and Mastery to all be very close in value. Crit is slightly edging out Haste right now, which is really the big change/surprise. I've spent about two full days now doing almost nothing besides shadow priest simulations over and over again... and we are in a situation where stat "weights" don't really describe what we want all that well. Lets take a very brief moment to remind ourselves what stat weights really are:

The idea behind stat weights is that your DPS will scale more or less linearly with stats. You can create stat weights a number of different ways. AMR does this by collecting 1000 data points randomly sampled from sets of gear with varying stat allocations, and then we use multiple linear regression to find the actual weights. This generates a linear function that we use to predict DPS in the form of:

weight1 * stat1 + weight2 * stat2 + ...

But, what about a spec like shadow that has a definite non-linear scaling for haste and crit? This predictive model will not do very well across a wide range of gear. So, you will often see TC'ers try to use "soft caps" on the weights to band-aid the results and make them a bit better. In the case of shadow right now... just one soft cap won't help us. There is a constantly varying relationship between crit/haste. Sometimes you will do more damage with them equal, sometimes more by favoring haste, sometimes more by favoring crit. In order to really score a set of gear well, we would need a non-linear function that describes this relationship over a range of values. We're actually working on that at AMR, but, it's not ready yet, so, we need to make due with good old stat weights for now.

Many TC'ers have posted about how the value of Haste drops off dramatically for S2M somewhere in the 11-12k rating on gear range. I'm finding this to still be true. One of the benefits of these latest weights is that by valuing Crit ever so slightly above Haste... you really don't need to worry about a Haste "cap" - you'll realistically never reach it if you are following this gearing advice. But, if you do find yourself swimming in Haste gear, I do suggest that you still keep it below 12k.

My best advice right now is to aim for gear with equal amounts of Crit and Haste. If you are doing more fights where you use bloodlust at the end of the fight during S2M, favor Crit a bit more. If you are doing fights where you lust on the pull more often, favor Haste a bit more. Our default single-target weights are for the Ursoc (H) script, which uses bloodlust during the enrage at the end of the fight, which is why you see us recommending Crit slightly above Haste.

We also have a set of weights for Mythic+ now, which favors Haste heavily:
http://www.askmrrobot.com/wow/simula...4?weights=true

Mythic+ weights are a discussion all to themselves, though.

Now lets talk about S2M simulation itself. First off, I have incorporated numerous optimizations to the shadow rotation for S2M in the past two days. If you have been unhappy with the simulation results for shadow in the past on AMR, give it another try - the DPS is substantially higher now.

The first thing a lot of people will say is: "Hey, I know I can get longer S2M durations than what the simulation shows." This is true. But, your average damage will decrease in a simulator by attempting to go for the longest S2M possible. I'll try to explain, and I apologize if I fail horribly.

The duration of your last voidform can be highly variable, especially as you push it longer and longer. A stretch of zero shadowy apparition procs can create an insanity deficit that drops you out earlier than usual. The slightly fluctuation or delay in your rotation can screw it up, etc. In a simulator, we could set it up to attempt to find the longest possible S2M duration you could get. Lets say it is possible to get up to 150 seconds on a god-like lucky parse. But, then the average duration is, say, 125 seconds and the minimum is 100 seconds. Most of the iterations will be around 125 seconds +/- 5 seconds or so. You'll get a few bad ones that are shorter, and a few really good ones that are longer. But, on average, you are dying 25 seconds before the fight ends... which is awful. So, in a simulator, you want the average S2M duration to be as close as possible to the maximum. This means that, on average, you die just as the fight ends. Being a bit conservative results in much higher average DPS.

In-game, you have no way to guess the end of the fight at 120+ seconds out with accuracy greater than +/- 10 seconds (or more really). The game is too random. So, your DPS can vary greatly from fight to fight. You can get great parses that are better than what a simulator shows.

I purposefully limit the simulator to a more conservative S2M duration. This increases the average DPS, but, it also cuts out the really high-end (and really low-end) iterations. Besides increasing average simulated DPS, the other reason I do this is because introducing large volatility into simulations makes them ridiculously slow. Slow to the point that it is prohibitive to even do any theorycraft with the results.

Below is a link to my latest rotation and I'll give a few notes on it for those who are interested.
http://www.askmrrobot.com/wow/theory...w&version=live

1.) If you expand the "Learned Variables" you will see two learned values. Those track how long our non-surrender voidforms are and also how long our surrender phase is. Note that the SurrenderDuration variable only learns the minimum and average values. The rotation has to specify a maximum.

2.) There is a function called S2MStartTime that is used to tell the rotation when to start S2M. If you click on that function you can see what it is doing, and it is ugly. What I did was I ran hundreds of simulations at varying levels of stats and found the optimal S2M start times. I stored those in a matrix that I reference from this function, and then layered some rules of thumb on top of that to adjust the start time based on talents, legendaries, bloodlust, etc. I will continue to refine this function over time, but, this is a good start. I couldn't find some "smooth" function to predict when to start so I ended up just going with empirical data instead.

3.) I made improvements to when to use Power Infusion during Surrender. Previously, I had been keying this decision off of a certain number of voidform stacks, but, this really needs to vary with your stats. So, instead I am finding that the best time to use it is when you are about 2/3 of the way through your typical S2M duration.

4.) The logic for using SW: Death with Reaper of Souls is a lot better now - it will make sure you don't sit at max charges, but otherwise save it for times when you are in danger of dropping out of VF during a MB or MF cast.

Another couple of notes:
Some people who really dig into the simulations might ask why the minimum value for S2M duration is so low. I spent a lot of time trying to figure that one out... eventually I managed to get a log of one of these horrible iterations. Basically... it is just possible to have terrible luck and end up with a very short S2M duration, especially since the rotation is saving cooldowns like PI for later in S2M instead of using them to guard against the worst case scenarios. I went through the logs line by line to make sure the simulator was working. These really unlucky cases are very rare - so rare you will probably never encounter them in-game. But, they are possible, which is why you see the low minimum values.

The AMR simulator will have more fluctuation in S2M duration than if you were to use SimC. This is because we have built it to play more like a human (although an insanely good human), who is not perfectly consistent. The relevant piece of this for shadow is when you cast Mind Flay. During Surrender, you will cancel your mind flay when void bolt is ready. A typical pattern you see is like: VB, MB, MF (for 1 GCD/2 ticks), VB. You don't want to clip the second tick of that 1 GCD mind flay. In order to do this, it is not possible to cast VB exactly when the second mind flay tick happens. There is some human reaction time/delay involved because you can't "queue up" the Void Bolt without clipping the tick. We don't use a constant value for that, we use a random value within a realistic range. In a rotation like S2M, very small fluctuations in timing can lead to results that are noticeably different.

Hit me up anytime to talk theorycraft - I'm always open to suggestions for improvement.

[Clrg]

Im not a TC but i did a test with my own stats and current enchant (35.1% haste, 21.1% crit) using the ursoc script and the simulation wasnt able to reach the 3VT with the haste set, if i use the auto gem and enchant option (23.4% crit, 32.5% haste) it reach the 3vt and die after it. Dropping to 29% haste and 25% crit it can also reach the 3VT. Did a few more test with bl at pull and the simulator wasnt able to reach the 3VT with the crit or the haste set, the simulator pretty much need a bl in the execute range like at 60+ stacks to reach the 3vt?

[Swol]

The hotfix makes it much harder to reach the third void torrent consistently unless you have really high level gear or a bloodlust during S2M. Keep in mind that the simulator's rotation is optimized around the highest average DPS. You could certainly bump up the max duration it attempts to achieve and see more 3 VT iterations, but the average DPS would go down due to the factors I mentioned in the OP.

[Aureact]

I'm not that theorycraft savvy, so forgive me if you already explained this. Why did the recent insanity change make crit so much more important? I would have thought haste would become even more important, given the increased difficulty.

Either way, thanks for stopping by to explain

[tiptopmemer]

I'm not that theorycraft savvy, so forgive me if you already explained this. Why did the recent insanity change make crit so much more important? I would have thought haste would become even more important, given the increased difficulty.

Either way, thanks for stopping by to explain

More haste in necessary, yes.

Crit is assumed that you have AS talented and thus the randomness is a little less random with more crit.

Unleash the Shadows? (vt crit artifact trait) is now more valuable, So the relic from Karazhan should be farmed every week.

[Yozu]

More haste in necessary, yes.

Crit is assumed that you have AS talented and thus the randomness is a little less random with more crit.

Unleash the Shadows? (vt crit artifact trait) is now more valuable, So the relic from Karazhan should be farmed every week.

So would double Unleash the Shadows pull ahead of double Touch of Darkness now?

[Djriff]

So, I have a few issues with this new list, especially the methodology that was utilized.

From looking at it yesterday you did 840,000 iterations on a 3:30 fight with some variance for fight length.

From the get go while the new rotation looks better (still room for some small improvements), the analysis on the whole is flawed. The stat weights that we develop using SiMC are across a multitude of different fight types as well as we weigh the stats per current boss encounters to give a more accurate set of weights for current content.

About the 11-12k haste cap that you spoke about, I'm afraid your information is out of date. The haste breakpoint, which was when we noticed a plateau around 10k haste. Using this 10k, we were able to aggregate values around this breakpoint. As you might point out, this is only a snap shot of stats at this set of haste. To combat that we went a few steps further and calculated weights at 7-13k haste using 1k intervals over the same fights, and weighed them. Doing that we were able to observe exactly how haste behaves at the different levels of gear while keeping everything in ratio of each other.

As for minimum vs maximum values, ideally you do want the majority of the actors in the iterations to hit the high end stacks of StM. We noticed that 90% of actors are hitting ~115 stacks, then it starts to decline a little bit. Now if you practice, then in reality you'll be able to hit it, however with the drain change, it's definitely more difficult to hit the 3rd torrent, but still possible.

As for not clipping mind flay in SimC, we currently have it setup to be interrupted by Void Bolt, since you want to void bolt at basically all costs. It's also a DPS + IPS gain to sometimes literally just wait half a second to cast Voidbolt due to the dynamic amount of haste.

Overall, I find the analysis AMR did, ok at best, but flawed in general due to only being for one fight. It's a step in the right direction though.

[Swol]

@Djriff

We have a gearing strategy for both Ursoc style fights and mythic+ style fights on the site. This post, as I stated, is only about the Ursoc style fight. Nowhere in the post did I ever say otherwise. This is not a post about advice across fight types.

You didn't understand what I was saying about interrupting mind flay with void bolt. Of course the rotation is set up to interrupt mind flay with void bolt! I was talking about a nuance of how the game works with regards to spell queuing. Mind Flay channels for the duration of 2 GCDs. You will often want to channel it for only one GCD, which gives you two ticks. A human player cannot cast a Void Bolt at the exact same time the second tick happens. There is a human reaction delay between the second tick and the Void Bolt. AMR randomizes this delay to a degree - as a human would. SimC does not randomize it, resulting in robotic sims that can do the exact same thing on every single iteration.

That 840k number on the gearing strategy page has nothing to do with the analysis that I did to create the shadow rotation, or with this discussion at all. That is just the set of simulations for special effects like trinkets and legendaries that is cached with the gearing strategy when someone runs the simulator in gearing strategy mode. If you would like more detail on what that information is for, we can give more detail on that.

If you think there are more refinements possible in the rotation, I'd be happy to hear your suggestions and we could easily test them out and see if they result in improvements or not.

I don't agree with your methodology for creating stat weights. Simulating different fight types and then aggregating the results doesn't really work, from a mathematical perspective. My approach is to provide weights independently for each type of fight, then let users decide what is most important to them. Right now I have Ursoc style and Mythic+ style weights available pre-calculated. We will be adding more over time.

The major issue I have with your approach is that you generate ALL of your data with the same set of gear as the starting point, which introduces a very large bias to the results. I do my gearing strategies and weights across a range of stats encompassing many different item levels of gear and combinations of stats. You claim that my information for where Haste drops off is out of date, but it is not! You are suffering from only looking at theory generated with that one set of gear at one particular item level as the starting point. I'll give an example to show why this matters:

If you start with a set of gear that has 6000 haste and 9000 crit rating as the reference, and then simulate the DPS increase in 1k Haste increments you get a result like this (which shows a larger drop-off in haste's value at 10k haste):
http://www.askmrrobot.com/wow/simula...3df9fabc215739

But, if you instead start with a set of gear with 6000 haste and only 6000 crit, you will see the larger drop-off in haste's value doesn't happen until after 12k haste:
http://www.askmrrobot.com/wow/simula...d29beb01796fe8

There is no "magic number" for haste that is important. Instead, we need to look at the combination of crit and haste. What really happens is that there is more or less a cap for how long you can really keep S2M going. At some point, adding more stats isn't going to change that. The length of S2M depends on both crit AND haste, so, that value where haste's "weight" starts to go down in value compared to other stats is very dependent on how much crit you have on your gear.

As you can see, you are allowing your starting conditions to inform and bias your results! You can't offer general advice based on tests that are only done using one set of gear at one item level as the starting point.

Also, our gearing strategies use the weights as an approximate scoring function used in conjunction with the cached simulation results for all of the special items. Your methodology of hand-merging the weights makes it impossible to use the weights as a scoring function - the magnitude of the resulting weights will have no relationship to how much DPS you can actually do with a given set of gear on a given type of fight. This makes ranking special items using the weights impossible, and thus not useful for use with a gear optimization website like AMR, which needs to rank more than just items with static stats on them.

So, respectfully, I find that your gearing advice is "flawed in general" to use your words It is a step in the right direction, though.

This is the disclaimer provided in the priest discord:

SimCraft is a tool for theorycrafters. It is not a tool for players. Use Twintops stat values, which were generated over multiple fight types, lengths, and talent setups and which took days to generate. Simming your individual character for one fight type and one fight length (or even multiple) will not give you an accurate picture of stat weights. Even if it did, they would change as soon as you changed any value in the setup (gear, talent, fight type, fight length, actors, latency, etc.). There are only two things you should use SimCraft for -- identifying a target DPS range to make sure you're doing the damage you should be doing and evaluating individual item upgrades/enchants to identify if "A" is an upgrade over "B" and whether it will increase your damage [Disclaimer: not always accurate]"

AMR does not need this disclaimer. It is a tool for theorycrafters and a tool for players. I encourage anyone interested to create custom gearing strategies for the content they care about. I can show people how to do this in a few minutes using AMR, and I can show them how to tailor it a bit to their current gear. We have focused a lot of our energy for Legion on creating a tool that can be used by hardcore and fledgling TC'ers alike.

I don't expect anyone to take my word on anything - so I make it easy to check for yourself if my advice works.

[Djriff]

We have a gearing strategy for both Ursoc style fights and mythic+ style fights on the site. This post, as I stated, is only about the Ursoc style fight. Nowhere in the post did I ever say otherwise. This is not a post about advice across fight types.

Putting bolded statements that say "My best advice right now is to aim for gear with equal amounts of Crit and Haste. If you are doing more fights where you use bloodlust at the end of the fight during S2M, favor Crit a bit more." is definitely saying otherwise. You and I both know that most people skim forums and they'd read that and go do a bunch of crazy things.

You didn't understand what I was saying about interrupting mind flay with void bolt. Of course the rotation is set up to interrupt mind flay with void bolt! I was talking about a nuance of how the game works with regards to spell queuing. Mind Flay channels for the duration of 2 GCDs. You will often want to channel it for only one GCD, which gives you two ticks. A human player cannot cast a Void Bolt at the exact same time the second tick happens. There is a human reaction delay between the second tick and the Void Bolt. AMR randomizes this delay to a degree - as a human would. SimC does not randomize it, resulting in robotic sims that can do the exact same thing on every single iteration.

For getting accurate results, the robotic approach is better. You're introducing more variables where they don't need to be variables. Simulating DPS rotations is suppose to assume perfect play. Yes you can add lag, or delays to spells, but for stuff like stat weights, you want it as close to perfect as possible.

That 840k number on the gearing strategy page has nothing to do with the analysis that I did to create the shadow rotation, or with this discussion at all. That is just the set of simulations for special effects like trinkets and legendaries that is cached with the gearing strategy when someone runs the simulator in gearing strategy mode. If you would like more detail on what that information is for, we can give more detail on that.

Then I either misread it or misconstrued it. It definitely comes across as 840k iterations over 42m (which is fairly fast). Trying to simulate legendaries into a single set of stat weights is also the wrong approach imo. Unless I'm misunderstanding what you said.

If you think there are more refinements possible in the rotation, I'd be happy to hear your suggestions and we could easily test them out and see if they result in improvements or not.

As I said to Zooper when she came into the priest discord to argue with methodology despite not understanding it I would be happy to. I am 100% a fan of giving players tools to better themselves and if I can get AMR to that point I will gladly help. However the problem still lies in the fact that I can't see under the hood about specific things with AMR that I can with SimC. Do you guys publish your code somewhere? Would be interesting to see how it compares on programming level with SimC.

I don't agree with your methodology for creating stat weights. Simulating different fight types and then aggregating the results doesn't really work, from a mathematical perspective. My approach is to provide weights independently for each type of fight, then let users decide what is most important to them. Right now I have Ursoc style and Mythic+ style weights available pre-calculated. We will be adding more over time.

You understand that's exactly what my approach is as well yes? Aggregating results provides minor differences to stat weights and at most will effect your DPS by + or - 100, not really something even min/maxers care about.

The major issue I have with your approach is that you generate ALL of your data with the same set of gear as the starting point, which introduces a very large bias to the results. I do my gearing strategies and weights across a range of stats encompassing many different item levels of gear and combinations of stats. You claim that my information for where Haste drops off is out of date, but it is not! You are suffering from only looking at theory generated with that one set of gear at one particular item level as the starting point. I'll give an example to show why this matters:

Without going into the sheer amount of computation time that simulating several different types of gear sets to an accurate degree would actually take, all of our stat weights are normalized to int. By normalizing to int, you mitigate the ivl discrepancy since secondary stats are completely out of whack in Legion.

If you start with a set of gear that has 6000 haste and 9000 crit rating as the reference, and then simulate the DPS increase in 1k Haste increments you get a result like this (which shows a larger drop-off in haste's value at 10k haste):
http://www.askmrrobot.com/wow/simula...3df9fabc215739

But, if you instead start with a set of gear with 6000 haste and only 6000 crit, you will see the larger drop-off in haste's value doesn't happen until after 12k haste:
http://www.askmrrobot.com/wow/simula...d29beb01796fe8

There is no "magic number" for haste that is important. Instead, we need to look at the combination of crit and haste. What really happens is that there is more or less a cap for how long you can really keep S2M going. At some point, adding more stats isn't going to change that. The length of S2M depends on both crit AND haste, so, that value where haste's "weight" starts to go down in value compared to other stats is very dependent on how much crit you have on your gear.

As you can see, you are allowing your starting conditions to inform and bias your results! You can't offer general advice based on tests that are only done using one set of gear at one item level as the starting point.

You understand that we're not just adding stat values from the void correct? We're not using simc plots(which adds stats from the void, which can be useful data as well), but we're keeping everything in ratio to each other. Again as I said above, by normalizing to int, the issue is mitigated. It allows us to look at the relationship of crit and haste at different levels of each, once we find a good haste breakpoint at X amount of crit, we can then look at crit and see how that curves and if there's a BP for that. While crit and haste do share synergy, the interaction between crit and mastery is interesting to look at as well.

Also, our gearing strategies use the weights as an approximate scoring function used in conjunction with the cached simulation results for all of the special items. Your methodology of hand-merging the weights makes it impossible to use the weights as a scoring function - the magnitude of the resulting weights will have no relationship to how much DPS you can actually do with a given set of gear on a given type of fight. This makes ranking special items using the weights impossible, and thus not useful for use with a gear optimization website like AMR, which needs to rank more than just items with static stats on them.

Special items like legendaries are a completely different ball game. From a TC standpoint, they are a right pain.

So, respectfully, I find that your gearing advice is "flawed in general" to use your words It is a step in the right direction, though.

Ok, you're entitled to your opinion.

This is the disclaimer provided in the priest discord:


AMR does not need this disclaimer. It is a tool for theorycrafters and a tool for players. I encourage anyone interested to create custom gearing strategies for the content they care about. I can show people how to do this in a few minutes using AMR, and I can show them how to tailor it a bit to their current gear. We have focused a lot of our energy for Legion on creating a tool that can be used by hardcore and fledgling TC'ers alike.

I don't expect anyone to take my word on anything - so I make it easy to check for yourself if my advice works.

I also encourage players to do their own TC, the knowledge isn't exactly out there for people to learn, and it because of these people that the community constantly gets statements like "my haste values are negative, so haste is bad!". Until I can see under the hood of exactly what makes the simulator tick, I will never trust it for accurate stat weights or gearing strategies.

[Zoopercat]

Jumping in for a bit: For anyone who wants to look under the hood at the AMR simulator, all of the spell mechanics, item data, stat conversions, boss scripts, and rotations are all on the wiki: http://www.askmrrobot.com/wow/theory...w&version=live

The 'simulator' compiles all of that wiki data on demand when you run a simulation. So anything you see there, is what you see happening in a simulation. Hit us up with questions.

[dnasty]

Just going to chime in here in terms of haste, crit, and mastery. I am by no means a huge TCer other than fiddling with the gear actively in my bags and then Mr. Robot and switching things around. With one option, I can go as high as 37% haste and 20% crit with negligible mastery, and on the exact opposite end I run 29% haste (9500ish), 23% crit, and 57% mastery. My sims when running with this gear were roughly 30-40k higher. And in practice I have noticed I have been doing more damage overall as well. It is still entirely possible to have sub 10k haste and consistently hit triple torrents.

[Dawnrage]

Thanks for taking the time to explain things. I personally think it's great there's more tools other than simcraft to compare to. We need more stuff like that, because simc has it's own inaccuracies but due to it's popularity, it's trusted way more than it should be.

I also think it's silly that another theorycrafter would try to downplay what you are trying to do here as not-trustworthy when you are putting forth so much effort to make it a tool worth using - kind of similiar to the state simc was in the past. Really, there's no need to make this a competition, we need more tools worth using.

[Kilee25]

@Dawnrage I agree. I don't have the energy or drive anymore to try to talk sense into people. Some people are just taking crazy pills or something...

[Clozer]

Reaching about 10 stacks more in s2m with amr setup. I dropped from 10k haste down to 7k and went full crit. This is amazing, though in dungeon runs with lotv im doing way less damage. Hopefully those 10 stacks more are worth it.

Still can't get to 3rd VT. Reaching about 100 stacks with ilvl 850. I think more crit/haste in general will help a lot.

[Shardstorm]

Looking at your Surrender to Madness Start Time function. How does this section work:

haste1 = Floor(Min(StatGear(Haste), 12000) / 3000) + 1
haste2 = Ceiling(Min(StatGear(Haste), 12000) / 3000) + 1

&
crit1 = Floor(Min(StatGear(CriticalStrike), 12000) / 3000)
crit2 = Ceiling(Min(StatGear(CriticalStrike), 12000) / 3000)

I'm just not au fait with how the Floor and Ceiling functions work with the Min in there. Or more precisely, what set the min is sampling? With say 9500 Haste and 8000 Crit I'd get:

haste1 = Floor(Min(StatGear(Haste), 12000) / 3000) + 1 = 3
haste2 = Ceiling(Min(StatGear(Haste), 12000) / 3000) + 1 = 5

&
crit1 = Floor(Min(StatGear(CriticalStrike), 12000) / 3000) = 2
crit2 = Ceiling(Min(StatGear(CriticalStrike), 12000) / 3000) = 4?

Is this designed to find a best fit S2M to suit a range of different gear sets to get comparable data?

Following that, I'm not understanding:

if CooldownSecRemaining(Bloodlust) > 0
Result = Result - 7.5
if CooldownSecRemaining(Bloodlust) > 0 and StatGear(Haste) > 6000
Result = Result - ((StatGear(Haste) - 6000) / 6000 * 12.5)

Not having Bloodlust off CD for a 9.5k haste set up means it strips nearly 15 seconds off the timer? That seems extreme to me. The timer I calculated out for the gear I talked about above ends up at 107.7. If you're aiming conservative, I guess that works. But it also explains why I out DPS my AMR Sim by a fair amount.

[Delat7]

Just wanted to clarify something. Are the Mythic+ weights assuming S2M usage? Or would they remain the same regardless of talent swaps.

[socomdan]

Thanks for taking the time to explain things. I personally think it's great there's more tools other than simcraft to compare to. We need more stuff like that, because simc has it's own inaccuracies but due to it's popularity, it's trusted way more than it should be.

I also think it's silly that another theorycrafter would try to downplay what you are trying to do here as not-trustworthy when you are putting forth so much effort to make it a tool worth using - kind of similiar to the state simc was in the past. Really, there's no need to make this a competition, we need more tools worth using.

I do not feel like there is any downplay involved maybe Djriff or Swol can comment. Rather I feel this is a healthy discussion because what we see here are two different approaches to simming our class. By them communicating and sharing information and strategies will/can lead to improvements in their simming techniques which I am happy to say will benefit us the players! I am a chemist rl and this is why scientists publish and discuss their works because it may very well be that they both do not have the complete methodology down yet and/or they could learn from each other as well.

[tiptopmemer]

Just wanted to clarify something. Are the Mythic+ weights assuming S2M usage? Or would they remain the same regardless of talent swaps.

With the multidotting done on trash, its safe to say ilvl > all when it comes to dungeons.

[Swol]

@Shardstorm

The tests I did are stored in a 2 dimensional array. I did tests for all combinations of 0, 3, 6, 9, 12k haste and crit. So, those variables at the top of the function find the two points on either side of your current gear. With 9500 Haste and 8000 Crit, the variables would be:

haste1 = Floor(Min(9500, 12000) / 3000) + 1 = Floor(9500 / 3000) + 1 = Floor(3.167) + 1 = 4
haste2 = .... Ceiling(3.167) + 1 = 5

crit1 = Floor(8000/3000) + Floor(2.67) = 2
crit2 = Ceiling(2.67) = 3

So, it then goes to the lookup SurrenderStartValues and uses (haste1, crit1) for the first data point and (haste2, crit2) for the second.

Looking at this, I could improve the results slightly by doing an interpolation instead of just taking the average value. But, the values are close enough together that it probably won't have much of an effect. I'll try it out and see if it makes much of a difference.

The adjustment for bloodlust looks a little funky, but, the reason for it is because when you have less than 6000 haste, you won't be using dispersion right at the start of S2M, which lowers the total duration of S2M a lot of the time, but ends up getting you more damage overall. When you have more than 6000 haste, most of the time you will use it right away in an attempt to get a second dispersion and/or 3rd Void Torrent. That adjustment is trying to take into account a discontinuity in the surrender duration due to a change in the rotation that occurs. We could try tweaking that adjustment some more to see if it helps.

It's entirely possible that this function is being overly conservative in some cases. If you think that the simulator is not getting enough average damage for your character, we could take a look at it and see if there is something about your particular gear that warrants further refinement. A link to your character and, even better, a link to some logs on Ursoc (not mythic ursoc) and/or Guarm (single target fights) would be great. We could use those to do some investigation.

@Djriff
Still waiting for your suggestions on what might improve the rotation. I don't want to discuss it on discord - post your suggestions here, and then the discussion won't be buried - I want people to be able to come back and read/benefit from the process.

[Djriff]

Djriff[/MENTION]
Still waiting for your suggestions on what might improve the rotation. I don't want to discuss it on discord - post your suggestions here, and then the discussion won't be buried - I want people to be able to come back and read/benefit from the process.

Yeah sorry, didn't mean to disappear for a couple of days, but life happens ya know?

As to what I said above about under the hood, I wasn't speaking about the wiki, that is definitely a useful thing for understanding spells + rotations + boss setups. I was more speaking on the engine side of the house, I'm interested in seeing how it differs from SimC.

Hopefully I get a chance to look into it this weekend.