The Team Robot Simulator and Gear Comparison Tool druid section is now live!

Both feral and moonkin are supported.

What is it?
The Team Robot site combines two theorycrafting tools that many of you may already be familiar with: a simulator, and a gear ranking tool. Simulators attempt to simulate the actual game, and then use statistics to draw conclusions. Gear ranking tools use these conclusions, or conclusions that have been worked out manually, to rank gear by how much it will increase your DPS.

By combining a simulator and a gear comparer into one tool, you get the best of both worlds: easy gear rankings backed up by a simulator. Some items are very difficult to estimate, such as certain trinkets. If an estimate looks off, all you have to do with the Team Robot tool is press the Simulate button -- no back-and-forth with the developers, no relying on some guy's estimate of how the proc should be averaged out: a powerful and accurate tool for answering hard questions is always just one click away.

Features

• Completely web-based: no downloading, installing, or updating! (Requires Microsoft Silverlight. Tested and works with IE, FF, Chrome on Windows, and FF, Safari on Mac.)
• Load your characters from the armory
• Easy, visual interface for customizing talents, gear, gems, enchants, glyphs, and buffs
• Powerful rotation editor that allows you to create your own priority-based DPS rotations
• Save versions of your entire character, sets of gear, talent specs, and rotations
• Easily share any of your saved data with a simple yet effective import/export feature
• Detailed, visual simulation summary and searchable combat log
• Ranks gear, gems, and enchants using a DPS estimation technique that extrapolates the value of items from millions of cached simulation runs.

Yeah... I should probably make that more obvious: you start by loading a character from the armory and tweaking it.

Maybe I am reading the combatlog wrong, but Savage Roar doesn't seem to be consuming combo points.

Edit: A request. Would be nice if the summary listed uptime on buffs. (rake, rip, savage roar)

The UI for this thing is really nice...

- Is there any way I can snoop around code, or is it closed-source?
- Default profile never Berserks.
- Default profile has 42% white damage, which is unrealistically high. Probably due to low yellow damage than high auto-attack damage.
- Looking at the default profile, the last "shred" conditions never let you get over 60 energy except when TF is up, which is why berserk never gets triggered. Look at the new SC feral profile for the kind of conditions you'd want to examine.
- It would be nice to have quick access to your alternate talents.
- Hitting the "Sim" button gives us a +-50 DPS range, and multiple attempts often have a larger spread than that. Any chance you can up the iterations and narrow the gaps? Most upgrades are in the range of 50 DPS upgrades. If you don't want to change the quick response time, you might want to store cumulative sim results so if I click the button 5 times without changing gear, it stores everything.
- Recent characters aren't only my characters-- "Swol of Spinebreaker" shows up on my recent character list.

I'm a co-developer of the website, and I can respond to a few of your items (Yellow will have to get the rest):

The code is closed-source. We will be exposing all of the formulas and ability/gear/proc information that we use to run the simulation, via a search-able html format.

When you say "quick access" to alternative talents, do you mean that you would rather not go through the load -> saved talents?

The length of the simulation when you hit the "Simulate" button can be adjusted to increase the accuracy if desired. The dps values in the item lists are just estimates based on cached simulations, so the actual values may be off. But, the relative values are 99% accurate. Rarely will you find an item listed as "better" than another item if it does not actually increase your dps. If you do find an example, please let us know. You will have to actually hit Simulate to get an accurate DPS result.

We have a cached DK, Swol of Spinebreaker, so that people can always have a starting point for the tool, even if the armory is down and they have no characters saved. We should also add a default feral druid.

Thanks for checking it out!

If you click on "self-buffed" and them click on the "custom" option for buffs, it will show you which buffs are being used. Same procedure for full buffs.

I didn't notice the functionality of saving/loading talents. But I'd like to be able to access someone's alternate talent spec from the armory, not just their active one. This should be possible, as I know SimulationCraft does it.

While it might be true in the general-item case to not see one item being better than another with those large ranges (or the actual difference not mattering in the end), you will almost certainly see the case where gem evaluations can be out-of-order. Also, since we're dealing with hard caps/soft caps so often in T10 gear, rarely is the decision between a single item and another, but a switch of multiple gear pieces.

Edit: While the length of the simulation can be adjusted, it won't let me save a number higher than 120. Consider that the SimulationCraft standard is 1000 or 10000 if we're being thorough.

I cant see the info for the proc on the Ashen Verdict ring when I run the simulation. I know it is possibly on a to do list already but just in case.

Otherwise it looks really good, the UI is functional, swapping gear is straight forward, and the rotation builder seems to work well.

It would be nice to have the ability to "fix" a gear set as a control or starting point if only so you can refer back and see what changes you made, having the option to stack iterations of gear would be even better. If this is already present then I apologize, just wanted to give my first impressions.

@Tazarch: we originally had a lot of ideas around comparing sets of gear, but didn't have time to implement them. I'll definitely put it on the to-do list and give it some more thought.

Yes -- the armory gives info on both talent specs. I should be able to add that in.

Right now I put a hard cap on the total amount of time that the simulation can run (length of simulation * number of iterations). I did this mainly so that people wouldn't run really long simulations and then say waaaaaa it freezes. This program was entered into a competition, which was the main reason I put that safety check in. The competition is over though, so I can probably lift that restriction and let people run the simulation for as long as they wish. It runs entirely on your own computer, not our server, so it isn't a scaling issue either.

About it being open source: as rev said, we plan to make all of the formulas and theory used available in an easy format. It isn't exactly closed source -- I just haven't taken the time to clean up the code and make it available.

I'll add the Ashen Verdict ring proc, it is indeed missing.

As for the rotation never using Berserk... when I run it with Yellowsix, it does indeed use Berserk. There must be certain combinations of stats such that energy is consumed efficiently enough that the conditions for Berserk are never met. I'll play with it and tweak it a bit. While I was raiding last night, a few other thoughts on the rotation occurred to me as well.

In general, feel free to play with the rotation editor and let me know if you come up with a better default rotation. You can save the rotations and export it as xml and send it to people too.

I should have also mentioned: to toggle between a mangle and no-mangle rotation, check/uncheck the Bleed Damage Percent raid debuff in the Buffs section.

The UI is gorgeous (love the Robot graphics/font theme). Particularly impressed with the combat log.... I am supremely jealous!

A couple recommendations I always throw out to sim efforts:

(1) Latency (or "gap" or whatever) is very different depending upon whether the preceding spell was gcd-limited,
channeled, or cast-time-greater-than-gcd. From a UI perspective you could still start with just one number and the extrapolate it out for those three categories.

(2) Consider changing to target-death for sim-completion. Target health is then inferred from the dps and refined through each iteration. (Settles down in <10 iterations if you dampen it.) This makes it much easier to use abilities/talents that are target-health dependent.

Publishing results:

Do you have a link-to-this-profile or link-to-this-result feature? If you did not want to store the profile, you could support fully encoded links:

h t t p : //askmrrobot/profile?n=Name;c=Druid;s1=id,id,id,id,id;s2=...;g=id,id,id,id,id,id;t=10101010101010. ..

The wowhead profiles are nice for communication, but they require a user id. Having it fully encoded in the link (auto actions, ofc) would be better.

@Allev -- definitely, we want to get the data out there so that users can verify the simulator's accuracy. We're working on a strategy for doing this... hopefully we'll have it up soon.

@dedmonwakeen -- the current implementation of latency is admittedly naive... I wasn't quite sure how to do it. Right now, whenever you use a GCD-limited ability, it tacks the latency on. e.g. if I have a 200ms latency, if I use an instant at time 1 second, the next time I can use an instant is time 2.2 seconds. We haven't done classes with channeled or cast-time abilities yet... I welcome any details on how I should implement this.

And yes, a change to make it more target-health than duration centric may be in order, particularly when doing classes like warlocks and warriors. Perhaps instead of a fight duration, doing something like:

Target Health: [2,400,000] (300 second fight @ 8000 DPS)

The part in parentheses would be calculated and displayed on the UI as you vary the target health (perhaps via a slider). This might work out better anyway, as varying the fight durations slightly is a good idea to avoid weirdness with particular values of haste.