Current BiS list according to Mr.Robot for ToS

[RayVG]

So, after having a look at Mr.Robot best in slot feature I could not help but notice that havoc DH's want to use 4 set nighthold and CoF over ToS tier set and trinkets there.
I feel like, if this really will turn out to be for our maximum preformance, blizzard has made some bad design choices. However, this post is not about what blizzard is doing right or wrong, I just want to know if this is really true.

[Feral Druid ist Op]

Dont ever use this crap.... Sim yourself or if you to lazy use some discord or spread but avoid MR robot and noxus noxic or whatever that shit Pages name is for love of god.

[Deleted]

Whenever 50% crit on BD equals less damage output than the gain in fury reg by the old t19 4p, this part could really be true.
And ofc the T20 4p gets less good the more crit over 50% you get.
But I doubt fury gen should be an issue with bis gear from tos...

As for COF: The amr template overrates this trinket completely. Even with leg shoulders and relic (2min meta) this comb only tops belt on really long fights (8min+).

Never use amr for sims.
I prefer the best in bags option over pawn though, as pawn doesn't calculate set boni, procs or legendary affixes at all.
But even then you have to know the flaws in amr (like cof value) and aleays double check the dps outcome in simcraft.

[Brewhan]

Whenever 50% crit on BD equals less damage output than the gain in fury reg by the old t19 4p, this part could really be true.
And ofc the T20 4p gets less good the more crit over 50% you get.
But I doubt fury gen should be an issue with bis gear from tos...

As for COF: The amr template overrates this trinket completely. Even with leg shoulders and relic (2min meta) this comb only tops belt on really long fights (8min+).

Never use amr for sims.
I prefer the best in bags option over pawn though, as pawn doesn't calculate set boni, procs or legendary affixes at all.
But even then you have to know the flaws in amr (like cof value) and aleays double check the dps outcome in simcraft.

I have Askmrrobot premium and it's sometimes straight shiet especially with best in bags on my aff lock it only puts me in higher ilvl and stat priority and ignores my trinket rankins and also my 2/4 pieces.. feels bad man. I only use it now for upgrade finder and which m+ dungeon priority to run. I hope someday it gets a little bit smarter in the best in bag algorithm.

[Zoopercat]

Hey, as I mentioned in the other thread, make sure you are using a custom gearing strategy if you want to rank trinkets, set bonuses, etc. Here's a guide on how to do that: http://i.imgur.com/r2gpuBW.jpg

If anyone wants me to run custom setups for them, just drop me your addon string from AMR and preferred consumables and fight length.

Also, for CoF, the length of the fight you sim will dramatically affect it - it will either be great, or meh.

Lastly, if any of you found inaccuracies on the AMR simulator for havoc, just let us know. In fact, we recently did a super in-depth analysis of Havoc on AMR and SimC, and both are accurate, so pick your favorite: http://blog.askmrrobot.com/havoc-keep-simple/

[Extremity]

A "best in slot list" can be a weird thing. You've first got to know if they are looking at each piece individually or not; when they say "these are the best shoulders" are they saying "of all the shoulders, and looking at them independently of your other items, these shoulders have the best stats" or are they saying "when combined with these other items, these shoulders fit the best and contribute the most"?

Either way, as others here have said, you really want to sim yourself and determine your own weights and what is best for you. A best in slot set is great to help you figure out some pieces that you want to shoot for, but you shouldn't blindly follow it in all circumstances -- ESPECIALLY when gearing up, where some pieces on there could be flat-out inaccurate. A trinket on a BiS list might be great when in all of the gear presented, but totally garbage unless you're also wearing items X, Y, and Z.

Just remember that there's really nothing outside of a sim that you should follow religiously, and you shouldn't even follow sims blindly. Remember that these things are references and view them with a grain of salt.

[Zoopercat]

@Brewhan - as mentioned in that other thread - a customized setup gives you great results for your character. As with everything, default weights are a starting point, but when you run a custom setup, all trinkets, set bonuses, legendaries, etc are custom tailored to YOU specifically.

@Extremity makes some good points. Use these tools to test things out and as guidance. But they are a model of the game, not the game itself. So they won't match perfectly. And any setting you change - the length of the fight, the boss script, talents, items, potions, all affect the results. In fact, I was talking with a theorycrafter the other day about some cases we found where his rotation did 0.5% more damage when using Potion of Old War, but another theorycrafter's APL did 0.5% more damage if all you did was swap in potion of prolonged power.

Does this affect your gear decisions? Probably not (but I am still going to test it, for fun).

Lastly, just like Blizzard has some bugs in game, simulators, spreadsheets, and guides resulting from these tools could all have bugs in them.

So like Extremity said - useful tools, but definitely have to think about the outputs and continue to test theories.

One thing to note about Mr. Robot's BiS lists: we take into account your current gear as an option among other items. Let's say you don't want to look at warforged items for A BiS list because you can't count on getting them... but let's also say that you have a +20 item that normally isn't very good. The BiS list will compare your +20 to the +0 versions of other items and determine what's best.

[Occar]

Leaving for posterity, but I am, in fact, frothing at the mouth from a position of ignorance here. They sort it out later in the thread.

_____________________________
AMR is not reliable at all.

It will suggest bizarre gear combinations, and isn't actually simming anything to provide the results. Instead it's just looking at the stat weights you give it, and some arbitrary weights for given trinkets or special attributes. These arbitrary weights may be correct in one isolated set of gear (this is being generous on my part), but are completely wrong for any other set of gear and/or spec.

I have done the following before:

1) Import my current gear
2) Sim my current gear with AMR and receive a result of X dps.
3) Allow it to choose "best in bag"
4) Sim the suggested gear with AMR and receive a result of around 0.9X

Since we have an AMR rep on this thread, please explain to me how you can suggest people use your site when the gear suggestions are so poor that even the sim engine the site itself uses disagrees with what it spits out? Don't try to blame it on stat weights or the like. You can repeat these types of results with using standard stat weights, the automatic stat weights with targets, or the actual, correct weights that you get from simulationcraft.

Examples of items that AMR fixates on that result in its own sims admitting that its data is bogus:
1) Jeweled Signet of Malandrus
2) DoG and CoF

[Extremity]

AMR is not reliable at all.

It will suggest bizarre gear combinations, and isn't actually simming anything to provide the results. Instead it's just looking at the stat weights you give it, and some arbitrary weights for given trinkets or special attributes. These arbitrary weights may be correct in one isolated set of gear (this is being generous on my part), but are completely wrong for any other set of gear and/or spec.

I have done the following before:

1) Import my current gear
2) Sim my current gear with AMR and receive a result of X dps.
3) Allow it to choose "best in bag"
4) Sim the suggested gear with AMR and receive a result of around 0.9X

Since we have an AMR rep on this thread, please explain to me how you can suggest people use your site when the gear suggestions are so poor that even the sim engine the site itself uses disagrees with what it spits out? Don't try to blame it on stat weights or the like. You can repeat these types of results with using standard stat weights, the automatic stat weights with targets, or the actual, correct weights that you get from simulationcraft.

Examples of items that AMR fixates on that result in its own sims admitting that its data is bogus:
1) Jeweled Signet of Malandrus
2) DoG and CoF

I'm not the right person to respond to this, but how long ago did you run whatever sims you're referring to? Because they are definitely, 100% doing more than just "taking your provided weights and calculating based on that". You can literally see the simulation that was ran and a pseudo-combat log.

[Narabug]

So, after having a look at Mr.Robot best in slot feature I could not help but notice that havoc DH's want to use 4 set nighthold and CoF over ToS tier set and trinkets there.

This is also true for many classes due to titanforging. If the set bonuses remain active into ToS, the NH set bonuses will be far superior to the ToS set bonuses.

AMR is not reliable at all.

As with any tool, its reliability is in its use. Blanket statements such as yours are akin to grouping with a Havoc DH in 865 ilvl gear who does not know the rotation or anything about their build, then stating "Havoc DH are terrible because I saw this one time..."

At its core, AMR is just a tool to simulate your character, similar to SimCraft, but taking a slightly different approach in many ways. If you input your character data in, your rotation in, and the encounter details in, you'll get very precise measurements. If, alternatively, you try to go at it the 'easy' way and just import/run without looking at what is happening, what the settings are, what the encounter is, etc, then you can expect the tool to provide very unreliable results.

Fact of WoW in its current state is that the value of a trinket or talent will change with the gear you have equipped. Additionally, one thing to not on sims with cooldown-reduction items is that you have to actually look at the duration of a simulation and see what is happening. CoF is a great example of this, because, depending on the fight length, that cooldown reduction may result in one additional meta, or nothing at all. Similar considerations could be made for trinkets that have a component that activates on crit - they will become more powerful as your crit increases. Additionally, having them equipped will increase the value of critical strike rating.

Don't blame a tool because you don't know how to use it. Either figure out how to use it, or accurately state, "It was not helpful for me because it was too confusing to set up properly."

[Occar]

Leaving for posterity, but I am, in fact, frothing at the mouth from a position of ignorance here. They sort it out later in the thread.

_____________________________

I'm not the right person to respond to this, but how long ago did you run whatever sims you're referring to? Because they are definitely, 100% doing more than just "taking your provided weights and calculating based on that". You can literally see the simulation that was ran and a pseudo-combat log.

Yes, if you sim it. The gear suggestions do not use the sim engine at all. The "best in bag" functionality doesn't do any sort of sim at all and is just using weights. This would be fine, but there are arbitrarily assigned weights to certain items that result in sub optimal suggestions.

- - - Updated - - -

This is also true for many classes due to titanforging. If the set bonuses remain active into ToS, the NH set bonuses will be far superior to the ToS set bonuses.As with any tool, its reliability is in its use. Blanket statements such as yours are akin to grouping with a Havoc DH in 865 ilvl gear who does not know the rotation or anything about their build, then stating "Havoc DH are terrible because I saw this one time..."

At its core, AMR is just a tool to simulate your character, similar to SimCraft, but taking a slightly different approach in many ways. If you input your character data in, your rotation in, and the encounter details in, you'll get very precise measurements. If, alternatively, you try to go at it the 'easy' way and just import/run without looking at what is happening, what the settings are, what the encounter is, etc, then you can expect the tool to provide very unreliable results.

Fact of WoW in its current state is that the value of a trinket or talent will change with the gear you have equipped. Additionally, one thing to not on sims with cooldown-reduction items is that you have to actually look at the duration of a simulation and see what is happening. CoF is a great example of this, because, depending on the fight length, that cooldown reduction may result in one additional meta, or nothing at all. Similar considerations could be made for trinkets that have a component that activates on crit - they will become more powerful as your crit increases. Additionally, having them equipped will increase the value of critical strike rating.

Don't blame a tool because you don't know how to use it. Either figure out how to use it, or accurately state, "It was not helpful for me because it was too confusing to set up properly."

In your haste to try to blame me, you completely ignore that I am talking about the gear suggestions AMR puts forth. You focus on the sim engine (which is rather feature limited compared to what I can do with simulationcraft), and completely ignore that the "best in bag" function is laughably inaccurate. My whole post was outlining how the best in bag feature gives really poor suggestions. In fact, I state the the sims that AMR itself can put out show that the best in bag feature doesn't provide optimal gear choices. To refute that, you defend the simulation portion? I don't know how to respond to that because, it has nothing to do with my criticism.

[Extremity]

The best in slot feature using stat weights is a pretty obvious limitation and is realistically the only way to be able to offer the service on a massive scale to a multitude of users without running hundreds of simulations at a time for every single person loading a new item set every 2 seconds.

If you want to simulate different items... use the simulator. It's there.

I get what you're saying about item suggestions based on stat weights being less accurate, but nobody's debating that - you put in your stat weights, and the tool simply tells you what gear will give you the highest amount of the stats you've stated you want to emphasize, prioritized by the weights you've provided. It's simply meant to be a guideline, not an end-all-be-all "this is the best no matter what". If you want more accurate, simulated results... you can do that.

I just don't understand the point of somewhat flaming them for having "terrible stat weights that aren't based on a sim" when they literally offer a sim service right there on the same site. You can do the thing you're saying you wish it did, it just takes an additional step.

I guess they could have more dynamic listings, with each item having a hidden 'score'. Any time you run a simulation, the items you happen to be wearing at that time are then given an overall score based on their contribution in that simcraft. As long as you maintain the same weights/whatever you had for that simcraft, the best in slot tool will reference those weights and potentially change suggestions based on your sim results. As you run more sims and do so more often, your best in bag tool could change and slowly become more accurate as it has more saved sim data per-item to reference.

Outside of something like that, though, which would be a pretty significant undertaking from a development perspective... I just feel like you're overemphasizing the problem when in reality the solution is right there, it's just not as convenient as you want it to be. And in terms of the accuracy/options/etc of the AMR sim, read the blog post he linked in earlier in this thread; you can see that, compared to regular SimCraft, it's fairly comparable.

[Occar]

Leaving for posterity, but I am, in fact, frothing at the mouth from a position of ignorance here. They sort it out later in the thread.

_____________________________



If it were flat stat weights, it would be workable.

The issue is that there are certain items that they have arbitrarily assigned such inflated values that if you possess them, it will suggest that you use them regardless of their lack of actual value. Every /use item and every item with a secondary type ability has a hardcoded value that there is nothing you can do to tweak. If the values that they had assigned reflected the actual value of these items it would be ok, but they do not. The people running the site have the insight to know that they do not because their sim engine shows that the items they are suggesting you use are not good gear decisions for you.

For an example, today, just now, using the last gear I imported in to AMR (probably from a month or so ago), I can sim the gear I was using, with their Krosus heroic fight, and the longest fight length it will allow. The result is 757K dps. I set it to use the much advertised stat targets for weights, and allow it to pick best in bag. It swaps some gear around. I then run the same sim (bear in mind I have chosen the max length it will sim in order to stack the deck in favor of what it is suggesting since it wants to use DoG and CoF) and get a result of 728K. The website tells me to use a certain set of gear, but then admits that the gear it says will do the best damage can't match the set of gear that I have selected to use with far more reliable methods.

If this were an isolated, short term type thing, that might be understandable, but, at least in regards to Havoc, there are so many incredibly poor manually assigned weights to trinkets, legendaries, etc, that the best in bag functionality is laughably inaccurate at best.

[Narabug]

Did you simulate your current talents/gear for stat weights or just use the generic weights that are provided?

[Zoopercat]

@Occar - Our trinkets (and all items with procs) are ranked from simulation data, they are not manually assigned. But let me back up and try to explain a few things:

1. First, it looks like you are using default stat weights, which are based on just one character setup, NOT yours. That would be like using someone else's simc weights for your character - it won't get you the best setup. This is akin to stat weights guides provide - a generic starting point that is not customized.

2. Run a custom gearing strategy - the guide to do that is here: http://i.imgur.com/r2gpuBW.jpg

Now, here's what happens when you run a custom gearing strategy:

1. We simulate various gear setups in order to find you the best stat weights.

2. Trinkets: We don't assign random values. We rank them based on real simulation data. So when you run a gearing strategy, we also sim every single item with a proc, over multiple iLevels and use that data to rank them. This is one of the reasons the AMR simulator has advanced the 'stat weight' tools out there. And since nearly half of your gear might have a proc on it, this is actually a very important piece.

Re: Best in Bags: You're not the only one to run Best in Bags on non-custom stat weights, then click the simulate button and get confused. So we have a few new features and helpful things that will steer people in the right direction. Those should be coming in a few weeks

And you can also browse the theorycraft wiki and edit anything on it You said there was no way to edit anything - I wanted to clarify. Here's an example page, if you edit it, it updates immediately for your simulations



If you post your character's best in bags string, I can run the custom strategy for you, run a few tests, and walk through the results together.

- - - Updated - - -

You focus on the sim engine (which is rather feature limited compared to what I can do with simulationcraft)

If you tell me what you can't find, I'll help you find it. The AMR simulator is actually super powerful, bringing 'batch simulation' to a whole new level, adding in a ton of statistical models, etc. Just ask and I'll point you in the right direction

[Occar]

Leaving for posterity, but I am, in fact, frothing at the mouth from a position of ignorance here. They sort it out later in the thread.

_____________________________



As I have indicated, I have used default stat weights, I have used the "automatic" stat weights, and I have used correct stat weights that I have obtained elsewhere (i.e. simulationcraft). I believe my exact words were "Don't try to blame it on stat weights or the like. You can repeat these types of results with using standard stat weights, the automatic stat weights with targets, or the actual, correct weights that you get from simulationcraft." Additionally, in my most recent example, it isn't even a stat weight issue at all. It's just the bizarre fixation on specific trinkets and legendaries.

Perhaps I should have said weights for trinkets and such are statically assigned instead of manually. What I mean is that you have set these weights, and they may be correct for a single scenario, but they are laughably incorrect for anything else.

However, based on the previous response, I am sure that this post will either a) be ignored or b) meet with another angle of "you're doing it wrong".

[Zoopercat]

So you haven't used customized stat weights from our simulator yet. If you do that, you'll see the best results, that guide is here: http://i.imgur.com/r2gpuBW.jpg

Stat weights can't rank trinkets or items with procs. We are not using stat weights to rank them. We use simulation data to rank them.

So putting in stat weights from simC isn't going to help, since it doesn't rank trinkets. When you run a custom gearing strategy (as the guide walks you through), you ALSO get custom trinkets, legendary, relic and set bonus sims. And that data is bundled in with the stat weight report and that is used to rank the gear. This is the missing link - once you run this custom setup, you'll get custom trinket rankings.

Perhaps I am not explaining it well, which is my fault. Give me your best in bags string (or at least your character name and realm), and I can show you some results to help illustrate the results

[Occar]

I must apologize for my mistake. What you have said actually fixes it. It now just confirms to use what I am already using (which is as I would expect)

[Extremity]

I must apologize for my mistake. What you have said actually fixes it. It now just confirms to use what I am already using (which is as I would expect)

It's good of you to admit you were mistaken -- next time it might be better to simply ask if something is working as intended before making accusations. Your original posts were definitely a little harsh, maybe you could be a nice guy and edit them to let people new to this thread know that there was a resolution to your issue so they don't think that's an accurate assessment?

I have some questions regarding AMR since you're replying here, @Zoopercat, but I don't really want to hijack this thread. You should really have an "Ask me anything about AMR!" thread. Name it "AMA AMR" to be niche and cool.

I'll ask one question even though it's not what this thread is for, just because you're here; Do you guys have the ability (or the willingness) to implement any sort of "live simming"? Basically, I mean instead of submitting a get/post request and performing the simulation on the backend and loading a results page and that's that, being able to send an ajax call to update that results page once its displayed and show the difference between the two. Basically, something like:

1) I set up a sim I want to run, configuring everything perfectly. For the sim, I'm wearing a 900 Memento of Angerboda and a 900 Convergence of Fates.
2) I submit the request, the sim runs, I get my results page.
3) I'm unhappy with my results, or am just curious; either way, I now want to try the exact same setup, but swapping my Memento for an Arcanocrystal.

Instead of having to run a new sim here (or in another tab) and manually compare them, what it worked like:

4) I hit an "expand" button somewhere, which shows me a customizable character sheet (or whatever UI you use to allow people to select custom gear for a sim; I would assume that something looking like an armory page where you can drag/drop items to each slot would be really cool, but whatever works is fine).
5) Right now, whatever UI element showing the gear/spec/etc used for that Simulation is still showing what I'd originally selected. Within it, I change my Trinket #1 slot from Memento to an Arcanocrystal. A "Re-Sim" button that was greyed out before any changes were made now lights up.
6) An ajax request is sent, and the page dims slightly with a loading icon in the center of the screen while my browser waits for a response.
7) The second sim finishes, and my results page is updated; next to all relevant data from my original sim, I now have the results from the second sim. I can also see the difference between the two, ie "Sim 1: 500k DPS | Sim 2: 485k DPS | Difference: (-15k) DPS

Depending on how you can structure the UI, if you can have it scale accordingly you could effectively have as many compared sims as you want. The only limitation is how much data you can fit on one page; if after 2 or 3 sims it converts into a more compact data-emphasized table with less frills, ie:

                      Sim 1        Sim 2        Sim 3

Total DPS              894k         906k         814k

Attacks

Chaos Strike            52           54           49

First Blood             18           18           15

Throw Glaive            22           21           23

Mastery                 50%         51.4%       53.66%

Crit                   47.82%        31%         49.9%

Obviously that's not a table, it's a terrible code block in bbcode (that I formatted to align properly on my screen but I dunno if it will for you). My point is that as more sims are performed, you might have to reduce the UI to just statistical data as you need to fit more...

..BUT, no matter how many sims are run, they're still never submitting more than one at a time. It doesn't matter if somebody is running one sim or is comparing a new one to the past 6. It's still one set of gear, talents, etc being simmed, the only difference is the data will be displayed next to the prior data, and some basic addition/substraction will take place to show a mathematical difference. It's not as if comparing the sims actually requires any extra processing, because you're never leaving those initial results, never browsing away; you're just adding additional data to the DOM each time.

I sort of went a little overboard with the explanation and got a bit too technical. I hope you understand what I'm saying -- if you guys could add a way to run a simcraft, then make a small change and "refresh" the simcraft, showing the new results right next to the original ones for comparison, that would be a *TREMENDOUS* feature. One of the things that is a pain about simming is comparing different reports, glancing back and forth between two of them on different tabs/different pages and figuring out the precise differences. It's even worse when you need to compare 3-5 of them. I really think this would be incredibly useful and would be something absolutely unique, something that would legitimately be something people would come to AMR to use specifically.

[Zoopercat]

So I shouldn't respond right now, I was out for dinner and drinks with a guest and am slow processing things lol I THINK what you might like is our batch sim.

In the 'sim type' menu, choose batch. An extra set of options show up. One of those options is for items. You can select the items you want and sim the different setups and get the results. If you want to be fancy, you can also compare anything else (like talents + items) in the batch sim. As many different things as you want to compare, you can. But if you get TOOOOO crazy, it could take days to run - so DOUBLE CHECK the estimated sim time before the sim runs, just to be sure you are ok with how long it might take.

Another option is the 'stat' sims under batch. You can set different stat setups, and compare those too.

I'll check back tomorrow morning to see if I completely misunderstood (due to the cocktails), or if you got it working!

See you tomorrow