Bear stat weights - Mr. Robot wants your input

[Deathfairy]

That i sall cute and stuff, but i know one thing for sure. Last night 4 hours of working on HM chimerion few <2% attempts etc. NOT ONCE did we wipe because healers went oom. However 50% of wipes were caused by tank death, same goes for HM Halfus (which is a joke now btw). So all this sta is not important is BOGUS, it might be true for normals but once you have good geared healers and doing HM sta is still amazing.

[Yellowfive]

I'm sure that there are fights where more health is more helpful. But why gear for it when you don't have to? For most fights, getting more dps/avoidance stats is better. For the fights that you mention, getting more stamina is probably better. So gear for your own situation -- that's what you guys are advocating anyway, right?

Another thing that I plan to put into the Mr. Robot site soon is a minimum threshold for stamina, so that people can tweak sets of gear to have at least X amount of total health.

If you get down to it, there are only two major ways to increase your stamina: gems/enchants, and trinkets. Otherwise, your stamina is pretty fixed based on the overall item level of your gear. It's really easy to make a stat-weight based tool recommend stamina trinkets/gems/enchants instead of avoidance-based alternatives... I think that you guys are making things too difficult. There just aren't that many choices in the stamina department.

And thanks tangedyn, I am aware that there are some inaccuracies in the math of the original post. It was really meant as a rough draft to kick around on our own forum, but zoop got a little zealous and posted it here All of that will be irrelevant shortly though: the simulator that I am finishing will surpass the accuracy that any spreadsheet-based approach could give me.

Also katarn -- I'm not starting from scratch here, obviously (writing a simulator in one day would be a tall order). I have these different spreadsheets as a point of reference, and the simulator engine itself is well-tested. It served as the basis for our very accurate simulation tool back in WotLK, which was on par with SimulationCraft for the classes that it supported (feral cat, moonkin, elemental, enhancement, dps dks).

You guys seem pretty hardcore about your theory -- I think that you'll like this tool and the subsequent ease with which you can turn your theory into meaningful gear suggestions.

[Arielle]

Another thing that I plan to put into the Mr. Robot site soon is a minimum threshold for stamina, so that people can tweak sets of gear to have at least X amount of total health.

The only normal mode encounter with a health threshold is Chimaeron, and only if you're tanking double-attacks.

There just aren't that many choices in the stamina department.

There are far, far more Stamina choices than Agility. Most of the Agility options are terrible.

You guys seem pretty hardcore about your theory -- I think that you'll like this tool and the subsequent ease with which you can turn your theory into meaningful gear suggestions.

If it's accurate, sure.

[Yellowfive]

By "not many choices" in the stamina department I mean that the lion's share of your stamina is 'fixed' based on the item level of your gear. The significant wiggle room that you have is your two trinkets and your gems/enchants. So in practical terms, when you get down to it, the decision is "do I want stamina trinkets or not?" And, "do I want stamina gems and enchants or not?"

Making a tool like Mr. Robot spit something out for you that allows you to provide a custom answer to those two questions isn't too difficult. A stamina cut-off is really all that you need, similar to how we allow healers to set a cut-off for spirit beyond which they say "i don't need any more mana regen." It's hard to quantify in code, so we leave it up to the user to tweak.

This simulator's main purpose is to find a good relationship between all of the non-stamina stats in terms of actual damage reduction over time (i.e. minimize total damage taken over the course of a fight). That seems the best way to relate the non-stamina stats. (It's possible that a portion of stamina's weight could matter if you are vengeance-capped on a typical boss fight. This simulator will also determine if you are vengeance-capped or not based on how hard/how often you are getting hit.)

And of course underlying all of this is that threat is completely irrelevant, and thus should not factor into weights for bear druids. This has been my experience in-game as well.

[Myrrar]

If you aren't trying to add something to the discussion by helping them fix their theorycrafting, don't post. And no, coming in here claiming they are wrong and not posting on why is not constructive.

They aren't saying their theorycrafting is perfect, kind of the whole reason they are asking for help.

[Arielle]

By "not many choices" in the stamina department I mean that the lion's share of your stamina is 'fixed' based on the item level of your gear. The significant wiggle room that you have is your two trinkets and your gems/enchants. So in practical terms, when you get down to it, the decision is "do I want stamina trinkets or not?" And, "do I want stamina gems and enchants or not?"

I agree with this.

Making a tool like Mr. Robot spit something out for you that allows you to provide a custom answer to those two questions isn't too difficult. A stamina cut-off is really all that you need, similar to how we allow healers to set a cut-off for spirit beyond which they say "i don't need any more mana regen." It's hard to quantify in code, so we leave it up to the user to tweak.

Fair enough.

And of course underlying all of this is that threat is completely irrelevant, and thus should not factor into weights for bear druids. This has been my experience in-game as well.

At this time, yes. However DPS is scaling faster than our threat is. It's quite likely it'll become an issue in Cata T2. But I suppose you can always just alter it when that point arrives.

I'm very interested to see the results. However given the amount of misinformation I've had to quell over the past 4 months, I'm a little apprehensive.

[Zoopercat]

Katarn, I totally understand the skepticism, especially with so many tools out there that only 'kinda' do the things you want. Mr. Robot was built to solve all of those 'kinda's' when it comes to gear choices. We have the basic stat weights gathered from EJ, but now are doing a few deep dives into the classes and specs that aren't very straight forward. And Mr. Robot is very handsome, so I think some people are skeptical and think that if he's got beauty, he must have no brains How can something so easy (pushing one button), be so smart on the backend? Just kidding around with ya

Anyway, I wanted to post to point out one of the coolest things about the Mr. Robot tool. It's been built to be completely flexible. At any point we just have to update a few formulas or stat weights or whatever else changes, and then BOOM, Mr. Robot is up to date. None of the numbers are hard coded, or so intertwined that to change 1 thing means rewriting all of the code. Nope, none of that. Mr. Robot is built to have different things 'plug and play' if you will (I'm the non-engineer of the group, so forgive my non-tech speak).

Now the hard part, getting the right formulas to start with Which is why we're reaching out to communities for input. I suppose I could have waited until we were 99% positive we had all of the right logic and then come in here claiming we're the best thing ever. But, I know that the community is smart, and thought, hey, why not get input from some other smart people and make this go faster And I knew I'd get haters hating, and people posting with no logic, but I figured a few GOOD pieces of feedback was worth wading through the pointless posts. Mr. Robot will take a beating if it means getting the right formulas... Robots are tough.

[Arielle]

We have the basic stat weights gathered from EJ

Ack! No! Bad! Stop! Do not use EJ for any Bear information. None. Most of it is bad or exceptionally behind. Not to say that other classes or specs may be in similar situations, but Bear information is just terrible.

[Zoopercat]

Katarn, funny you say that We launched with EJ stats and quickly realized that was NOT the resource for bear tanks. That's one the main reasons I brought this question to MMO-champ instead of EJ, because we wanted a better set of theorycrafters for bears

Totally agree

[tangedyn]

If you aren't trying to add something to the discussion by helping them fix their theorycrafting, don't post. And no, coming in here claiming they are wrong and not posting on why is not constructive.

They aren't saying their theorycrafting is perfect, kind of the whole reason they are asking for help.

That's what I did originally, I posted links to prior work done on calculating SD uptime, my own spreadsheet, and an article I've written a while back about dodge diminishing returns. Unfortunately that post was rejected because I'm apparently not allowed to post links.

The correct formulas are out there. They were pretty much collaborated on and have been extensively peer-reviewed, and Katarn has pointed them out to the team. Whether they choose to use them is up to them.

[Myrrar]

That's what I did originally, I posted links to prior work done on calculating SD uptime, my own spreadsheet, and an article I've written a while back about dodge diminishing returns. Unfortunately that post was rejected because I'm apparently not allowed to post links.

The correct formulas are out there. They were pretty much collaborated on and have been extensively peer-reviewed, and Katarn has pointed them out to the team. Whether they choose to use them is up to them.

Did it file it under spam? I just checked my folders and can't see it under there.

If you still wanted to post it you can PM me and we can work out a way that I can add all the links for you.

[Deleted]

I'm sorry that this may be qualified as un-constructive posting, but I'll say it again, because I feel it is vastly relevant to these guys' project. Myrrar if you disagree, then I'm sorry, and maybe I'm the one not getting it, so feel free to delete my post.

If you are trying to model a system that will measure the damage mitigation over time to determine what stat you should stack, you are not contributing to anything else that helping, maybe, some guys gear for normal modes. As these gradually become less and less relevant to the raiding population, why bother?

Your tool should be focusing ilmho on what is the best gear possible to tank hard modes. And this, as I'm sure you understand, is not achieved by a simple "mitigation" calculation, but also a mitigation vs. time-to-death comparison. Now as you may suspect from my name, I'm not a math person, but I'm pretty sure there is a way to model that based on boss damage output, your HP, your "never fails" mitigation (IE armor + talents), you can calculate an optimum situation where tank A can survive for X second with a typical healing throughput of Y.

This is what the community needs imho. If you can tell HM players how they should gear, you're making a useful tool. If you're telling us that agility gives more mitigation than stam in a calculator... well, we already know that. But it's just too far off the pulse of what happens in an actual fight.

Thanks for reading, again, sorry if you think it's irrelevant to the thread.

[Arielle]

You need 180-190k HP for most hardmodes.

[Yellowfive]

Nodata -- you would be surprised how many people want a tool that can help them gear up for everything from 5-mans to the cutting-edge raid content. Mr. Robot has had over 300,000 visitors in February alone. We definitely want to reach the hardcore community, but not at the expense of the raiding community at large.

As far as stamina... it's really easy to gear for stamina if/when you need it. All you have to do is slap some stam gems in your gear, a couple enchants, and get a trinket or two. There really aren't any more choices than that. Mr. Robot will easily do this for you if you just slide up the weight on stamina.

Probably the most difficult question is "how much stamina do I need for boss XYZ?" Those questions are very specific, and it would be very difficult to make a tool like Mr. Robot give you an answer for individual boss encounters.

Regardless, I have to believe that reducing total damage taken over the course of a fight is relevant for pretty much any encounter, though your mileage may vary depending on your healers.

But to the main point: as promised, I have completed a bear druid simulator that helps to illustrate how the different stats affect total damage taken in a fight in a much more concrete way than a spreadsheet is able. I don't think that I'm able to post links in these forums yet -- I'll ask zoopercat to do it in the morning. In the meantime, you can go to the Mr. Robot site and load up a bear druid: there is a link to the RoboMaul simulator at the top.

I think that you'll find it to be a good basis for an extremely powerful theorycrafting platform: a simulator that benefits from the ease-of-use of the Mr. Robot site: simple UI, seamless integration with our top-notch armory loading, direct access to the gear optimizer to run comparisons between different gear sets, easily add/remove stats, and more. The current functionality is only a taste of what is possible, and done all in 2 days' work.

We have tried to include enough output and documentation such that users can verify the simulator's correctness. We will of course promptly fix any bugs that are found.

We would very much like feedback on this tool, as well as suggestions for future features that you all would find the most useful in such a tool.

[MatsT]

Yellowfive: Your understanding of diminishing returns regarding dodge is flawed. The relative percentage of damage reduction gained at different levels is what is relevant. Going from 50% dodge to 55% dodge is exactly the same effective damage reduction as going from 0% to 10%.

Your arguments regarding armor is also quite flawed. Not sure exactly what your point is but it seems as if you are trying to say that it doesn't matter which weight you put since you should always get the higher ilvl item anyway. That is exactly the point of having a high armor value. Setting a too low armor value could result in the optimizer recommending a lower ilvl piece with mastery+dodge as secondary stats instead of a higher ilvl item with crit and haste.

Regarding your simulator, I'm yet again not sure exactly what the point is. If you only want to calculate damage reduced, Tangedyn's spreadsheet provides VERY accurate calculations for this that does not rely on simulation. Calculation is always superior when possible, simulation is an alternative when it is not feasible to do the calculations. At this point the best solution is probably to rely on a combination of his Mitigation scores and TTL scores weighted manually. I'm not sure exactly what your plans are for the future of the RoboBear simulator, is it going to include healing too? It would be somewhat interesting to simulate tank damage combined with some semi-random healing over long periods of time to see which setup dies the least.

EDIT: Just want to add that I think MrRobot is an excellent and very useful tool, but it's not feasible for you to generate the stat weights for all speccs yourself. Stick to roughly correct values from different forums and spreadsheets for the masses and leave the option in to set your own weights for the more advanced users.

[Arielle]

It's still undervaluing Dodge for some reason.

Why is it going for 4pc T11?

Missing the VP Ring, and Fluid Death.

[Zoopercat]

As promised, here is the link to Robomaul. http://www.askmrrobot.com/wow/tools/robomaul

[Dendrek]

I used Mr. Robot to see what advice it would give, and when I hit Optimize it told me to reforge everything to Mastery. I was surprised by that so I checked the stat weights. Dodge isn't even one of the considered stats - it's not on there at all.

[Arielle]

It actually turned out pretty good. I'm impressed with Robomaul. However you might want to up the default unmit hit damage on Robomaul. It's ridiculously low. 110000 is panzy. I don't think there is a raid boss that hits that weak.

Also, add some way to edit the boss settings on the gear optimizer. I'm 99% certain it's using the "panzy boss" setting for its default recommendations.

[Yellowfive]

I just used the same default that was in use on Tangedyn's spreadsheet (110,000 dmg, 2 sec attack speed). I'll check a combat log on our next raid and come up with a better default that matches how hard the bosses are hitting more closely, and yes, I base the stat weights off of the default settings for RoboMaul, so I'll update that as well.

Set bonuses are kind of difficult to deal with... we just assign a static value to them that is in rough relation to another stat. I may need to adjust it downward... I don't even remember what I used for 4-piece for bears. Probably an equivalence to armor averaged over a fight assuming 2 uses per fight. I take it that many people feel the set bonus isn't all that interesting?

I think that fluid death isn't showing up as BiS because the weight on mastery is high enough that items like unsolvable riddle and tia's grace are being ranked more highly (and because hit is ranked pretty low). I'll double check it though.

Dendrek -- dodge is most definitely on the list. Perhaps you were in cat spec?

MatsT -- we definitely do not plan to come up with our own stat weights for every class; we get most of them from other people's work, or tools like SimulationCraft. I just happen to play a feral druid, so I decided to make an extra tool for feral I've always liked simulators -- it keeps things simple. I can put in complicated things like dealing with ramp-up time on vengeance, or re-casting demo roar if it misses, or heroism, or whatever -- and it doesn't complicate the math: just average the results at the end. And, I was looking for an excuse to integrate a simulation tool with our armory loader and gear optimizer anyway!

Also, with regards to DR on dodge... I'm thinking about everything in terms of total damage taken over the course of a fight. It seems that no matter how you slice it... 10% dodge is going to reduce your total damage taken by more than 5% dodge over a 5-minute fight (on average)?

If you guys think of anything interesting that you'd like to see in the simulator, let me know -- there are a lot of possibilities since it can easily call into the reforger, BiS gear finder, etc.