Are sims to be trusted?

[Skylarking]

I'm quite a newb with simming. Anyway...

If i sim a spec that says something like 500k, and i sim another spec thats like 400k, does this mean that no matter what i try to do with the weaker spec its just not going to beat the spec that sims the highest?

[Aktec]

It just means that the 500k spec has higher potential dmg. A lot of factors will come into it, player skill, target switching, aoe, fight length etc. But genrally speaking in most scenarios if everything is equal the 500k spec will beat the 400k.

[wregelmann]

Well, there are a few things at play here - they're all pretty intuitive. In theory, the sim is the maximum possible output on a single-target fight with no mechanics. If you play both specs perfectly on this type of fight, then yes, the spec that sims higher will be better. Mechanics (movement) and multiple targets shake things up a bit, and that's harder to sim accurately. On top of that, your level of knowledge and comfort with each spec will dramatically affect your output. Spec A played at 70% will be worse than spec B at 90%.

Best advice would be to try both out in a safe environment (raid finder is great for this). Sims are useful for getting gearing priorities and testing builds, but you won't know for sure how you're going to perform until you do it out in the field.

[Puri]

It completely depends on how and what you are simming.
Usually people use the "Patchwork" simulation, so it only considers a single target with no movement. Obviously this is hardly applicable in actual raids.
So It might very well be that the spec you sim singletarget with 400k is significantly stronger in multitarget fights than the spec you sim with 500k.

You can modify the simulation to your needs, for example by adding more enemies or later in progression sometimes there are raid profile available that try to simulate the current raids as close as possible. But sim is sim, it also just estimates the fights, getting another add-phase or not skipping a certain phase in the actual raid versus the simulated will make the calculated values pointless.

Then there is the "human error", aka skill level. Some might be better (so closer to the sim), others might be worse. And you have additional variables, like latency and reaction time that can hardly be simmed.

Additionally the sims are based on algorithms provided by humans, they also might simply be wrong. Everyone can contribute to simcraft for example, so you can also go and check the formulas.

Simming is a nice tool that gives a good estimation of one's performance, and it might be the best considering the objectivity it delivers, but it will not give you the definite answers to what is best for you, your class or spec in all circumstances.

[Radaney]

No. If I sim 'A' spec in an AoE fight that spec might sim lower than 'B' spec in single target.

Some specs have higher damage potential than others so if I play 'A' spec at 100% no mistakes which is absolutely impossible it might perform better than 'B' spec but for the general population of WoW 'B' spec will be performed higher.

Lets take previous expansions Arcane Mage for example, Pro players could play old Arcane spec very well because they could minimise movement and be affected less by boss mechanics but the typical player would of played Frost because the rest of us are too retarded to play Arcane Mage to it's higher potential.

Another example is when Sub Rogue was mechanically 'difficult' too play the typical player would play Assassination because it's rotation was 100x easier to do well with.

So sure a spec could sim higher but that doesn't mean YOU are going to do better. You'll just have to try it and see.

The only good things to come from Sims are stat priorities and in some circumstances to see which spells should be doing the most of your damage so you can judge if you're playing correctly.

[Deleted]

Sims are mostly trustworthy if you know what they are. As others said, they assume you play the spec at a set level of efficiency (that you can set, it's 95% by default), so if you're confortable with B and shit at A, you'll most likely perform better as B.

Dps variance should also be considered, for example I sim a bit higher as arms than as fury, but I'd rather play fury because it has a lot less variance and thus is more reliable for progress.

And again as others said, it only sims certain fight types, so you have to keep in mind that it won't be trustworthy on stuff like Anomaly, or Gul'dan with its buff and constant target switching. But they're great for knowing what a spec's or gearset's dps potential on a patchwerk or aoe heavy fight will be, and you can derive from that.

[ThrashMetalFtw]

In theory, the sim is the maximum possible output on a single-target fight with no mechanics.

This is wrong. Sims represent the average dmg output if mistakes aren't made. The maximum possible output with better than average RNG (Crits and Trinkets) will be A LOT higher than what the sim tells you.

It's very possible (and realistic) to outperform your average sim.

[Krieleros]

Sims are also stationary, single target bosses. Frost DK is the best performing class in mythic NH because of all the cleave/AoE, but it doesn't sim as high as other classes in SimC. SimC is good to get your stat weights, but I don't put too much faith into the DPS number that gets spit out.

[Raiju]

Well, there are a few things at play here - they're all pretty intuitive. In theory, the sim is the maximum possible output on a single-target fight with no mechanics. If you play both specs perfectly on this type of fight, then yes, the spec that sims higher will be better. Mechanics (movement) and multiple targets shake things up a bit, and that's harder to sim accurately. On top of that, your level of knowledge and comfort with each spec will dramatically affect your output. Spec A played at 70% will be worse than spec B at 90%.

Best advice would be to try both out in a safe environment (raid finder is great for this). Sims are useful for getting gearing priorities and testing builds, but you won't know for sure how you're going to perform until you do it out in the field.

Sims aren't maximum possible damage, they're average damage with an 'optimal' rotation (which is only optimal if the profile is set to be, and largely isn't possible to apply outside of patchwerk)

[chooi]

This is wrong. Sims represent the average dmg output if mistakes aren't made. The maximum possible output with better than average RNG (Crits and Trinkets) will be A LOT higher than what the sim tells you.

It's very possible (and realistic) to outperform your average sim.

SimCraft does show min-max variance in results, though.

[ThePants999]

Sims are also stationary, single target bosses.

SimC is massively configurable. If you want to simulate a fight where you have to run 30 yards every 15 seconds, two adds turn up every 45 seconds with 50M health each and run in a circle around the boss, the boss goes immune for 5 seconds every minute, yadda yadda, you can do exactly that. Just because most people either don't realise you can do more than Patchwerk sims, or don't bother, doesn't mean that's all it can do.

SimC is GIGO. Put in the right conditions and you will get extremely accurate results. Put in unrealistic assumptions and you'll get numbers you can't rely on.

[ComputerNerd]

Sims are to be trusted as much as your understanding of them.
They are in short a number of assumptions.

The more accurately those assumptions model the encounter and the behaviour of the player, the more accurate the results.

[Riverr]

Most of the comments are correct. The only thing I'd add to this is that sims are best used to compare gear (or talents) within the same spec. Comparing 2 different specs (i.e. outlaw vs subtlety), or even worse, 2 different classes, is not as reliable because Simc profiles are made by a number of contributors (generally each class has one or two direct contributors) who may spend more time perfecting one spec than another. Also, certain class specific bugs or features not implemented in Simc can make certain classes/specs have quite disproportionate results when used for large scale comparison, but are still fine if checking to see which item gives a greater number.

So I would say Simc can be trusted when comparing individual gear choices and/or talent choices to find which item gives a greater value. But the actual dps value isn't as important as the relative difference between sims of the same class/spec.

Edit: I suppose for rogues, the anecdotal evidence is that since release, across various iterations, Outlaw and Sub have been simming on par (and often much higher) with Sin, even at lower gear levels but in practice, that hasn't exactly panned out. Although sub seems to be climbing a bit right now in logs (but generally sub players on WCL start at a higher gear level which can be seen using WCL scale graph), you would say on par at best. I suppose I bring this up to illustrate that even though I would say rogues have a very active theorycrafting community and our Simc contributors are fantastic - there is still quite a large disparity between reality and sim when making comparisons across classes and specs.

[Deleted]

No they can't especially the hunter ones.

[Ahovv]

No clue how accurate they currently are, but in 2015 the Arcane inputs were way off; I had to tweak them to get the correct parses which were noticeably higher.

So you sorta have to go in and see if the priorities are correct for your class.

[Cheze]

sim is just taking your stats and plug'n'chugging them into a particular encounter model; if the inputs align realistically with ingame circumstances you should see a corresponding realistic representation of your highest possible performance (within variance.)

it's important to remember that any sim contains a variety of assumptions, and to think about whether they align with real circumstances.

[Tojara]

Depends on the class in question. Some of the classes that sim are very close to the actual results they post, while others, well, not so much because they might be modeled wrong.

Keep in mind a lot of people simply don't know how to use sims at all, and expect the number they are getting from running a standard patchwerk simulation all the time. Krosus for example is the closest you're going to get to patchwerk, and there are still mechanics in that encounter that can/will prevent you from achieving what your sim DPS will show a on a patchwerk like boss. Also there are variances in sims (shown), and the number shown is generally what you would expect as an average. You will be lower and higher on every single attempt most likely.

Like mentioned previously, if you actually take the time to add mechanics when simming to expect things where you won't be able to DPS for a few seconds because of a mechanic, or you have to swap to adds throughout an encounter, you will get a better idea of what your character will pull during an encounter.

Great for stat weights though. But from my experiences with it, there are a couple classes/specs that are modeled slightly off.

[Meowchan]

Sims aren't perfect, and sometimes they are flat out wrong, but they are infinitely better than using no sims at all. If you're serious about min maxing your character, understand the assumptions the sim operate under, and realize what caveats the end result number comes with.

[Wilderness]

Sims are to be trusted as much as your understanding of them.
They are in short a number of assumptions.

The more accurately those assumptions model the encounter and the behaviour of the player, the more accurate the results.

This. Sims are just a computer spitting out numbers based on the variables put into it. Their degree of accuracy depends entirely on those variables. If your class is accurately modeled, if the rest of the classes are accurately modeled, and if the fight conditions match exactly what fight you're trying to compare it to then you will get 100% accurate information out of the sim. Fight conditions can rarely be matched exactly, and classes aren't always modeled perfectly either, so depending on how close those things are is how accurate the sim will be. In general SimC and the people involved with it do a pretty good job, but nothing is perfect.

[Osmeric]

More importantly... do Sims feel pain?