Shadowlands Gear Optimization Roadmap

when is shadowlands ready… still have battleofazeroth as gear guide??

Shadowlands optimizer is already up there! The guide section of the site we’re working on, will be available this week sometime.

When are you going to fix Fury Warriors optimizing? It’s still wanting me to enchant and gem Mastery. It should be Haste. I finally customized the wheel to 640 Haste> 520Mas> 460Crit > 380Ver. Which gave me a lot better DPS but it’s not perfect. Thanks.

We would need more information to respond to this. In general the optimizer is looking to get both mastery and haste on Fury warriors, as those are the two strongest stats and optimal builds will have both in roughly equal amounts. The gear you have available to you will determine what the optimizer does in any particular case.

Wow and Icy Veins both say:
The stat priority for Fury Warrior is:

  1. Strength;
  2. Haste;
  3. Mastery;
  4. Critical Strike;
  5. Versatility.

On their Graphs it looks like Haste is just barely above Mastery. Just about the same amount that Mastery is above Crit and Crit is above Vers. That doesn’t make them the same. When I ran the opt. the first time and it wanted me to Gem & Enchant Mastery. I Did…Waisted my gold because my DPS wasn’t right. So I replaced all the gems and what enchants I could with haste and it went up. The I played with your customizing wheel for quite a while and settled on, like I said above. Settled on 640 Haste> 520Mas> 460Crit > 380Ver. and my dps was much better. My gear level is only 132.
Not really sure what more info you need. Hope this helps. Thanks for the reply.

Our gearing strategies are centered around character that are higher item level.

Regardless, you won’t see much of a swing in damage if you play optimally whether you get more haste or more mastery.

We don’t use stat weights anymore - they are very limited in their usefulness. We create a model of the game that predicts simulated DPS. The optimizer picks the gear that will most likely simulate highest, which will most likely give you the best damage in-game. If you want me to look at your specific case, use the help link above the gear table and create a snapshot id. Post that and I can show you how the optimizer arrives at the suggestions that it makes.

I’m confident right now that the optimizer is working correctly for Fury Warriors.

Our gearing strategies are centered around character that are higher item level.???

What do you guess that minimum will be when we can trust it to be correct? 158ish? And if it tells me to gem and enchant everything Mastery at that level, then…

Sorry. I don’t see any help link above.
Thanks!

First: I love “Ask Mr. Robot”! You are doing a fantastic job. I am a huge fan and promote you whenever I can.

My question: How confident are you with the current Legendary recommendations/Simulations overall?`

I am a Balance Druid and my question concerns mostly “Balance of all Things” and “Circle of Life and Death”.

You seem to prefer “Circle” over “BOAT” when I use “Best In Slot”. Which is pretty different from everything else I read on the Internet (Blood Mallet / SimulationCraft / WowHead / Icy Veins…). Is this just not updated or do you consider more “details”?

I like to trust you more, but do you maybe have some additional Information for me to understand the difference?

Same with covenants. Your default Simulation Balance-Druid is Venthyr. Is this by design or not updated? Do you know more? Do you test better/differently?

If you have time for a response, it would really help me out.

btw: I am a programmer, you can talk sharp if you like xD

AMR’s team did a blog post if you want talking about “covenant choice” that you may be interested in :
Picking a Covenant: Mostly Good Choices – Ask Mr. Robot
I have no clue about your legendary tho. Swol is probably the one with the more insight about that.

1 Like

There were pretty major bugs with Convoke the Spirits (Druid NF ability) until last week, so it wouldn’t be surprising if the covenant rankings were biased against NF in one way or another.

In general, the data quality on a site like AMR is driven by two factors:

  • The quality of the simulator
  • The quality of the rotation being simulated

AMR’s simulator is generally OK, though it definitely doesn’t have as much manpower behind it as SimC, and it’s not always up to date with the research and data mining (e.g. trinkets missing implementation, more obscure mechanics like Convoke not correctly implemented). Some features I find are really cool, for example I find it better at simulating movement and human latency than SimC is.

The rotations however I’ve found to be very variable in quality. Usually you need rotation adjustments to make the best out of some trinkets or legendaries, and since most of the theorycrafters work with SimC the SimC rotations usually take these things into account while AMR rotations don’t. BOAT on Balance is a great example of that, where there is a 5-10% performance difference between BOAT with the “normal” rotation and BOAT with a special rotation to make best use of the proc. Since AMR doesn’t have that special rotation, it’s likely undervaluing the legendary effect by quite a lot. Unlike BOAT, Circle doesn’t really need any rotation change to make full use of, so that one would be valued “correctly” by AMR.

Unfortunately, the only way to really figure this out right now is to dig into the SimC and AMR rotations and understand where the differences are coming from.

1 Like

I did some work on this last week. I disagree that you could get 5-10% more damage with balance of all things compared to a “normal” rotation - it’s not that strong an item by itself. I was able to squeeze out another 1% or so. We’ll crunch the numbers and re-rank accordingly. I know the rotations I make look simpler and less “researched” but that is generally just a perception. The rotations I use are how they are intentionally, not because of a lack of manpower. Sometimes there is a clever optimization for a particular build that other folks come up with first, and I incorporate those trends as they emerge. Sometimes I find things first, and those eventually make their way into the world too - usually when someone else comes up with it independently :wink:

In my experience the simc APL authors include a lot of extremely complicated conditions that often only specifically yield DPS increases in patchwerk simc sims. Often I cannot recreate the theoretical gains even if I use the exact same rotations in our simulator. A lot of this gets down to our two different models of the game and how we allow the simulator to make decisions. Simc allows for perfect reaction time and a few impossible types of choices to be made. APL authors can mitigate some of that by using things like .react in the simc APLs, but the onus is on the APL author. The AMR simulator has that built in to the simulator settings instead of being rotation-specific. Also, our simulator was built from the ground up to be more like a human as far as reaction time goes. It’s not as popular as simc, of course, but we built it after using simc heavily for many years. Our simulator is an evolution of simulation beyond what simc does, in our opinion. Otherwise it would have been a waste of time to make it.

For balance in particular, we went over the simc model vs the amr model with a fine-toothed comb during BfA. Neither I nor simc folks could find any particular outstanding bug with either model after a little back and forth. But, simulation results were measurably different for some cases. Extra conditions in their APL that gave DPS increases in simc gave no corresponding increase in AMR. To me, this was a good indication that the underlying models being used were influencing the results. Maybe that still exists in the shadowlands version. I haven’t had the time yet to do a deep dive into simc. If big differences persist, I can do so once things settle down.

I’d say I’m pretty confident there won’t be huge swings. I am still actively doing QA (I never stop really), but I also expect a balance pass by blizz that hits some of these items. I can’t totally rule out some clever build I haven’t noticed increasing the potential power of a legendary item. I also can’t rule out certain items having more practical power than a simulator gives to them. After the raid comes out, I do plan to make some more scripts for the simulator that simulate a few fights in current tier content. That could be useful for people as another data point to help make their decisions.

2 Likes

Just to follow up on this: I think that the people referencing Balance of All Things as way stronger than other legendary items are doing it mostly in the context of Night Fae/Convoke the Spirits. Using Convoke right at the start of Celestial Alignment to benefit from Balance of All Things is pretty huge. Our rotation is already doing this - I put in a little tweak that makes sure not to delay it which adds a tad more damage.

As far as I can tell, the rotation we have is making good use of Balance of All Things at this point. I think we need to recalculate some of the rankings, which we can do in the next day or so.

1 Like

Thank you Swol for your feedback, very appreciated.

Can you tell please me at what gear level the optimizer starts working best for Fury and where the help link above is? All i see is Gear, Simulator, Addon, Blog and Forum. Thanks in advance. Fozzgate.

It should work well as soon as you get to heroic dungeon level, even lower it will be fine. so like 160?

Castle Nathria Update

We have just released a round of tweaks based on data and feedback gathered in the last two weeks in preparation for Castle Nathria. We fully expect Blizzard to tune things a bit over the first couple weeks of raiding, so we’ll keep an eye on that and do another update when appopriate.

Here are the highlights:

Guides!

The guide pages are back up and running! We will of course be tweaking things over the next few weeks, but they should be in a pretty good spot for all specs. You can view Covenant recommendations, Talent recommendations, and use our interactive Rotation viewer to get a detailed yet easy-to-read flow chart describing how to play your spec. The rotation adapts to talents, covenant, legendaries, and conduits that you choose – be sure to customize it!

Updated legendary and trinket rankings

We have updated the rankings for several legendary items and trinkets. A few of them were more difficult cases to test on the beta, such as the Inscrutable Quantum Device and Mistcaller’s Ocarina – those should be ranked for all specs now. The Inscrutable Quantum Device is one of those trinkets that you may have to use personal judgment – the value that you get out of it will depend a lot on your specific situation.

A few legendaries have been updated, e.g. Reanimated Shambler is filtered to Unholy DKs only in the game data, but can be used by other specs. We have manually overridden that spec requirement and added rankings for Blood and Frost.

In the upgrade finder legendary ranking, some legendaries where the special effect doesn’t have any value for a particular build were getting small non-zero values (e.g. a 0.5% upgrade). This wasn’t an error – it was actually pretty clever. That legendary ranking doesn’t look at the legendary effect in a vacuum: it predicts how well that legendary will fit into a near-BiS set of gear. Thus other gear options in that slot and others can influence the value of a legendary (as it should).

Therefore, simply being able to make a high item level item with perfect stats for a slot has a small value, even if the legendary effect does not. That said, we feel that such legendaries should show up as zero in the list anyway – since you can only use one legendary, there’s no point making one just for the stats. We are now ranking legendaries at zero if the effect is of no value for your talents+covenant.

Updated low-level tank optimizations

We create all of our “toughness” data for tanking around max-level content for the current tier. In this case, that will be Mythic+ and normal or higher raids, where people will quickly be at ilvl200+. When you are significantly lower than that item level, the toughness data generated for those higher item levels doesn’t really apply. It’s also not that beneficial to really worry about toughness until you get a bit higher anyway.

That said, our new system can handle toughness rankings down to a much lower item level now, so we have enabled that. If you leave the toughness slider at “All TUF”, we’ll rank gear based on toughness no matter your item level now. As soon as you move it away from “All TUF”, we’ll rank your gear based on DPS if you are below around ilvl180, and we’ll do your desired “blend” of the two if you are above about ilvl180.

We can’t do the “blend” below around item level 180 because the toughness data isn’t calibrated for it, and as mentioned above, it doesn’t really matter that much. Blending becomes more interesting at higher item levels.

Pawn export

For people who like the pawn export, we have tweaked it for tanks and healers. If you have your toughness/healing slider set below the half-way mark, we’ll output DPS weights. If above half, we’ll output TUF/HPS weights. We don’t generate “blended” weights because they don’t make a whole lot of sense. As always, take Pawn weights with a massive grain of salt – we really don’t recommend it as a way to make serious gear choices, it is more of a casual convenience while leveling or something.

2 Likes

Thanks for all your help.

Everyone should really look at this for their own specs because you’ll be shocked how dumb some of the rotations are and how unreasonable they are for a regular human to recreate. If you need a visual representation of them load up Hekili sometime and try doing their rotation on a raid target dummy and see if it even makes sense.

Over many expansions, at least for the specs that I’ve personally played and checked, the AMR ones are just more realistic out of the box or they are quickly updated when new usable tricks are discovered.

I found what seems to me an anomaly in the ranking of Resto Shaman conduits for Mythic+.

Obviously healing simulation presents many issues beyond those for DPS.

For my item-level 182 Resto Shaman, with sim-optimal talents, Mr.Robot thinks the best Potency conduit is Heavy Rainfall (“Using Healing Tide Totem increases the healing of your Healing Rain by 90% for 20 seconds”). In fact, this conduit is the only one which gives results statistically different from using no conduit at all!

Based on my experience in M+, I think it’s very bad advice, though.

HTT has a 3-minute c/d, and Healing Rain has a 10s c/d. So the maximum bonus is easy to calculate: You get a 90% bonus on at most two HR casts every 3 minutes. If you are using HR on c/d, it is a 90% bonus to 1/9 of the casts, or a 10% increase in HR healing.

But HR is rarely useful in M+ fights. HR heals for 29.5% of spellpower per target initially, and then ticks for 5/(1+haste) times for the same amount. According to Mr. Robot, Haste is the worst stat for Resto Shaman, so let’s ignore it for simplicity (my Resto Shaman has only 3% Haste). Best case heal for HR in M+ is then 177% of spellpower per target on five targets, so 885% of spellpower total, absolute best case.

The best case is very optimistic in M+ though, as situations where all five people can remain stacked inside a 10-yard circle for 10 seconds during a heavy damage period are extremely rare. And a heavy damage period is precisely when you would use HTT, but if you want to prepare, you would cast HR first (just before the damage comes) and HTT second (as the damage starts). Realistically, you would then spot heal the people in greatest danger and be unable to cast HR again (to benefit from the conduit buff) for almost ten seconds.

tldr; HR is a preparatory/maintenance spell, and HTT is a reactive spell, and the conduit buff requires using them in the “wrong” order. While it may make sense in a raid, as the only healer (in M+), you do not cast HR while people are in danger of dying.

Baseline Chain Heal, by comparison, heals for up to 510% of spellpower total (if it hits four targets), it does it right now instead of over 10 seconds, and it allows you to prioritize the target who needs healing most in many cases.

I tried to understand the logic behind this questionable simulation result. It appears the mythic healing simulator assumes HTT will be cast at least once in every single fight. If it is not triggered by the “panic button” of many heavily-injured allies, it is always triggered at the start of a fight longer than 3m30s, and in the last 30s. So the Heavy Rainfall buff will get close to theoretically maximum uptime.

The normal logic for HR is to use it when 3+ allies are in a 7.5 yard radius. But if the conduit is present and the buff is up, it uses it unconditionally. It’s unclear whether there is any additional internal logic to discount the raw healing amount based on the number of allies it will actually hit over the next 10 seconds, in that case.

You’re right that the mythic+ simulation will get fairly optimistic use out of healing rain in general. We’ll see how it changes as I implement a new mythic+ healing script in the weeks to come.

If you find you can’t get practical use out of the heavy rainfall conduit, go ahead and exclude it. The way I look at it right now: the numbers are showing you that if you are using healing tide totem in a tough fight and you can heal 3 or 4 of your group members with healing rain afterwards… it’s probably the best conduit (the script heals 4 of the 5 group members with healing rain). Conduits are weak and fairly invisible in general, though - so it’ll be hard to tell a difference in practice.

1 Like