The Blender: Dynamic Tank Gear Ranking

The Blender is a new feature that allows tanks to “blend” DPS-oriented gear with toughness-oriented gear.

Tank gearing advice has always been kind of… nebulous. You can figure out the best stats and items for DPS, no problem. And now with our tank simulator and its latest updates, you can very reliably figure out the best stats and items for staying alive – also not a problem.

But the hard thing is this: at some point I’m plenty tough for the raids that I’m doing. For example, most 915+ ilvl tanks feel pretty solid in Heroic Tomb of Sargeras. Getting even tougher… won’t really make much of a difference. So tanks will generally start shifting to some more DPS-oriented stats, items, or talents. But… how do I know when to make that shift?

That crossover point can be very personal – some tanks want to live on the edge and do more damage. Some are more conservative. There isn’t a one-size-fits-all answer to this question. Also, it’s not a “single” point – it’s a gradual shift.

It’s a hard problem, but this is the question that tanks really want answered by a gear optimizer: tell me how to be tough, and how to do damage, and how to balance the two per my preferences. So we have answered it! Enter…

#The Blender
This simple tool lets you define when to shift from toughness-oriented gear to DPS-oriented gear. To use it, click on the gearing strategy picker to the right of your character name at the top of the gear optimizer page. The Blender appears at the bottom of the popup.

We have provided several presets for you – just pick one from the list, and press Save at the top-right of the gearing strategy popup. The presets are:

  • All Toughness: gear entirely for toughness, completely ignore DPS. If you completely out-gear your current content (e.g. you have 925 ilvl gear and are doing heroic Tomb), this option is not recommended.

  • Conservative: this is the default, it will start mixing in DPS gear conservatively as you begin out-gearing content.

  • Balanced: smoothly transition from toughness gear to DPS gear, but a bit more aggressively than the Conservative approach.

  • Aggressive: go for all DPS unless you are horribly under-geared for the content you are trying to do.

IMPORTANT: The Blender will work pretty well with our current “default” gearing strategies, but it works really well with a custom gearing strategy, or our upcoming Adaptive gearing strategies (not quite available yet for tanks, but will be soon!)

ALSO IMPORTANT: There is no “right” choice. Pick the preset that fits best with your raid team and play style, and don’t let anyone give you shit.

###Custom Rule Sets
You can also define your own custom blend of toughness and DPS if one of our presets doesn’t quite fit your needs.

The way it works is that you define a list of “rules” that tell the optimizer to reduce the value of DPS as you get squishier. You can define toughness in terms of either NPS or raw Death Chance, whichever you prefer (NPS is almost directly proportional to Death Chance, so neither is better than the other, it’s just whatever makes more sense to you). This is best described by example:

Rule 1: At 90% chance to die, reduce the value of DPS gains by 90%.
Rule 2: At 10% chance to die, reduce the value of DPS gains by 0%.

Here’s what these two rules do: if my chance to die is really high (90%), make DPS worth very little. If my chance to die is pretty low (10%), make DPS worth full value. For the points in between, it does a linear interpolation, e.g. at 50% chance to die, DPS would be worth 45% with these rules. This gives a gradual increase in the value of DPS gains as you get tougher.

This example is pretty similar to the Conservative preset, and is a good starting point for tweaking your own rules.

###Pure DPS
If you want a pure-DPS strategy, you need just a single rule:

At 100% chance to die, reduce the value of DPS gains by 0%.

This rule makes DPS worth full value no matter how tough or squishy you are, and thus turns it into a straight DPS strategy. The Aggressive preset is pretty close to this.

As a long-time tank… this is finally the gear ranking tool that I’ve always wanted. Nobody wants to go pure toughness. Most people can’t pull off full-DPS without dying. Just blend man, and life is awesome. In the past… the problem was trying to figure out which toughness metric to use, and how much of it was “enough”, and blah blah blah. With NPS and Death Chance being calculated by a real simulation of a real fight, it’s pretty intuitive: lower my death chance to an acceptable level, then start getting more damage. That just makes sense. It’s quantifiable. Logs of simulations can be analyzed to figure out why I am dying. The Blender can be tweaked in just a few seconds to try out different balances in the optimizer very quickly.

Please try it out and give feedback! Did it work well for you? Which preset did you like the best? Did you have trouble striking the balance you wanted or was it easy? Did anything weird happen in the optimizer that you can’t explain?

For those interested, here are the rules being used under the hood by the presets:

All Toughness
Blender is disabled, based purely on NPS.

Rule 1: At 90% chance to die, reduce the value of DPS gains by 90%.
Rule 2: At 10% chance to die, reduce the value of DPS gains by 0%.

Rule 1: At 90% chance to die, reduce the value of DPS gains by 75%.
Rule 2: At 50% chance to die, reduce the value of DPS gains by 25%.
Rule 3: At 10% chance to die, reduce the value of DPS gains by 0%.

Rule 1: At 100% chance to die, reduce the value of DPS gains by 25%.
Rule 2: At 75% chance to die, reduce the value of DPS gains by 0%.


Yeah ! You ruinning every things ! :stuck_out_tongue:
Haha i was doing exactly that but without the tool … well doing to try it ! =D

Could do the same with healers… :3 ? Like Disci maybe ?

Yeah in theory it could be done with discipline – something we could visit once we polish off the new healer stuff.

Let me know how The Blender works out compared to what you were doing before with your tanks – does it work well? Or is it not quite hitting the balance you want?

1 Like

Can’t run sim yet, but next week i will try to run it =D

(Because from my personnal computer i got : “Windows cannot access the specified device, path, or file etc…” for the client 332, since forever, i will execute sims from my works pc)

Were you perhaps trying to run it from a network drive or a symbolic link or something? That would probably cause issues.

“Nop, i’m running it on the C or the D. I’m admin, the client have all windows right. Can’t find a solution online about it :/. I thinks it’s Win10. I even tried to shutdown my malware protection but not more information. Don’t worry, i will do it Tuesday.”

I found a solution ! =D
It’s Windows defender

I’ve run the Blend during this night but …
"The chosen boss script requires a fight length between 110 and 480 seconds."
I will redo it with 300, i wonder why this time limit ?

Currently running for 300s

We put a restriction on some of the boss scripts so that the fight length stays within what it was designed to handle. Going outside the bounds could cause the script to stop functioning properly.

1 Like

So, the gearing V2 is now finish, i’m going to show you my result. The objective is to find the best compromise between “Tanking” and “DPS”. My idear is during progress, going with a full “Tanking” than when you killed this boss more than 4 time going with the best compromise and finnally go with pure DPS.
That suppose those 3 points are far (in terms of NPS & DPS) from each other.

First let’s begin with the result from the old gearing. I will use this result as a reference.
Reference Tanking : 384159 NPS : 357607 DPS
Reference DPS : 383936 NPS : 389468 DPS

Honnestly after that i would stop right here, improving my DPS by almost 9% and an insignificant chance to Tanking. I would go for a fully DPS.

Let’s do a fully tank gearing (see BUG 1) and a DPS gearing with the “Blender”.
Tank Pure V2 : 388947 NPS : 364713 DPS
DPS Pure V2 : 375064 NPS : 395892 DPS

Is that bad ? Can’t say. Let’s compare V1 & V2

From the tanking perspective (first tab): this gearing combinaisons is better, i’m 1,25% better than the previous one (and i also improve my DPS).
From the DPS perspective (second tab) : this gearing is also an improvement with 1,65% better DPS but i lose -2,31% in NPS.
On this point it’s good but (there is always a but :stuck_out_tongue: )
When you look the impact on the Tanking when your gear is DPS, you lose -3,57% when for the last version you only move for -0,06%.
More than that, the full tank gearing is a better because i don’t lose anythings.

So there is an improvement for the tanking pure witch is good. There is also an upgrade for the DPS.
The use of rules is a bit confusing, i not sure if i know to use it properly. For me, it’s should be different:
3 points are insteressting:

  • Tanking Pure
  • Best compromise
  • DPS Pure

The Tanking Pure should be the default one, DPS is here only if you really need a push on some bosses or want to play for the recount. The best compromise should be create by minimizing the lose of Tanking for the best DPS.
I have an idear: from the best in bag, you have to order your spec. It’s the same idear between DPS / Tanking. You order witch one you care about, and you choose “max” (DPS or Tanking) or compromise.
Like this:

Well it’s nice to see some changes, but i’m still not sure if NPS and DPS is relevante in compare to live serv.
There is a log versus a sims.
Log : 296,736.2 DPS
Sims : 334813 DPS
Still far away (it possibly an human factor)

BUG 1 : So when i wanted to do a “pure Tank” i add a rules with:
At 0% chance to die, reduce the value of DPS gains by 100%
Got an “Invalid Rule”, Threshold must be between 0 and 100, and Reduction must be between 0 and 95%.
Why 95% ? And how can i do an 100% tanking ?

BUG 2 : The previous message comes after “saving”, it’s shouldn’t be that way, it’s should be right after a rules with a button “submit” or automatically when updated a value.

1 Like

That’s not a bug. If you want a pure tanking strategy, just disable The Blender entirely. If you are using it, you can never reduce the value of DPS to zero.

We do plan to make a different UI for it once we’re done testing it. But it will use this technique under the hood. The problem with just saying “compromise” is that it gives you no control over when to start compromising, or how “steep” the drop-off is. That all might be advanced for the typical user, and we will definitely provide some easy options, but we want that power to be available to the savvy user.

Your NPS values show that you probably auto-configured the strategy to up the difficulty level to something challenging. At that level, The Blender won’t really kick in unless you force it almost all the way to DPS. Our goal is to provide default tank gearing strategies targeted at ToS heroic, and ToS mythic difficulty. Then you can pick which one you are doing. As you get tougher, Heroic won’t be threatening anymore, and our default blender rules will start moving you towards DPS. Then if you switch to Mythic, it will want you to go all toughness again.

We’ll probably offer two M+ levels as well, something around what is appropriate for people with Heroic raid gear, and another for Mythic.

Ok ! So it’s two gearing strategy one for full Tank and the “other” like every point beetween 95% tank and full DPS, if i’m understanding right.

About the compromise, from my point of view i had the idear to go for the gear where your lose of tanking is minimize, and the DPS maximize. It should be only one point (I’m probably wrong but i’m curious)? If you run a simulation with the derivate you should target only one set of stats, no ? I’m seeing the comprise as a wave with local maximum/minimum for NPS, and one other for DPS, so it’s make a matrix 2x2 to optimise.

Blablabla I’m going too mutch technical here, you probably already know that and know what you want to do :p.

So the strategy up the difficulty, i did know that but i thought it will do it in MM+ difficulties. Maybe re-do the sims gearing with that off ? But how can I turn it off ? I really like the idear to give your current state on the raid and AMR “choose” between thoughness and DPS. Hope you will do something awesome (you are going to anyway ;-))

If you check “manual tank options” when running a gearing strategy, it will not override your difficulty settings.

One of the reasons we want to provide more thorough defaults is that it can be difficult to configure tank simulations to get the information that you really want for ranking gear.

The best way to configure it is the following: say you are a Heroic raider – that’s all your team does. Run a strategy with your character/gear at a level that is actually challenging for Heroic. For Nighthold, that was probably around ilvl875-880 or so, without many (or any) of the old paragon traits. Set a good but conservative “blend” on that strategy as described in the example at the top of this post. It will heavily favor gear that makes you tougher.

Then, as you get more gear and reach where we are now, with 50+ artifact traits and ilvl900+ gear, that same strategy and blend is going to heavily favor DPS gear. It’s probably going to start wanting to swap in DPS trinkets even, or DPS enchants, or different stats. That is exactly what you want.

OK ! (I thinks my first try of tank gearing i use the “manual tank options” but not on this one)
I will do it and see what’s change.

i really like this idea, to blend over the artifact + iLvl, i have only one problem in mind.
The progress is over, i’m at 920 iLvl but my raid is changing and some are taking rerols. So the difficulties will be heavier but the gearing will go for DPS gear when i would probably need tougher gear. I can’t run a Gearing strategies to change it, it would take too mutch time. You have an idear how you want to manage it ?
In simpler version, how can we change fast between a Tank gearing and a DPS gearing ? I thinks it’s the point where you need to simulate and have feedback right ?

So one thing to keep in mind – we’re making The Blender part of your options right on the gear optimizer. So you can switch between a more toughness-oriented and a more DPS-oriented build instantly, without re-running a new gearing strategy.

We also plan to provide a Heroic and Mythic set of default strategies for you to use. So if you are doing Heroic right now and getting overgeared for it, but plan to move into Mythic, you would probably want to switch to using the Mythic strategies, which will tell you to keep getting tougher to a much higher ilvl.

And you could of course use this same approach with your own custom strategies: create one at Heroic difficulty, and another at Mythic (for a raid boss script, you can get close by increasing the M+ level to +6 on one of our Heroic scripts… for ToS we may actually implement separate Mythic scripts with the extra mechanics, we’ll see).

1 Like

Where is this tool at so I can play around with it?

It shows up on a version 2 tank gearing strategy after you finish running it, at the top of the report. You can mess with it and then save and use that strategy.

I like The Blender, and it could be very intersting going forward. I have a couple of questions about it, though:

  1. Am I correct that the more rules you apply, the more accurate it gets? In your example you only have two rules, is that enough?
  2. Krossus and Spellblade are decent fights from a strategy standpoint, but with Tomb of Sargaras the fights will get more difficult if only because of bosses hitting harder. Having heroic versions of Nighthold bosses to test Tomb bosses might not be accurate.

What are your thoughts about these potential issues?

We will be adding Tomb of Sargeras bosses very soon – probably during the first week of raiding, since I really need to get in there and see it first hand to make a good script. So that will take care of that – going forward our entire gearing philosophy is that it should be based on the actual encounters of the current raid tier.

As for applying more rules, it’s not necessarily more “accurate” to have more rules. It just changes the “shape” of the curve. For example, the default rules in the original post:

Rule 1: At 90% chance to die, reduce the value of DPS gains by 90%.
Rule 2: At 10% chance to die, reduce the value of DPS gains by 0%.

So if you had a 50% chance to die, the value of DPS would be reduced by 45% – it does a linear interpolation.
At 30% chance to die, DPS value would be reduced by 22.5%

Now add another rule:

Rule 1: At 90% chance to die, reduce the value of DPS gains by 90%.
Rule 2: At 50% chance to die, reduce the value of DPS gains by 80%.
Rule 3: At 10% chance to die, reduce the value of DPS gains by 0%.

At 50% chance to die, the value of DPS is reduced by 80% now.
At 30% chance to die, the value of DPS is reduced by 40% now.

So you can add more or less rules to create different “shapes” that suit your needs, e.g. a really gradual drop-off, or a really abrupt drop-off. Whatever floats your boat.

I have found that I get exactly the results I’ve been looking for with one rule, but now I realize that I have probably completely misunderstood how the logic works. Can you better explain what happens with only one rule, what happens outside of the range contained within two or more rules, and how the logic handles rules based on NPS?

There are always two “hidden” rules at each extreme:

  • At 100% chance to die (or zero NPS), DPS is reduced by 95% (we never let you reduce it to zero, to avoid weirdness).
  • At 0% chance to die (or 500k NPS), DPS is reduced by 0%, i.e. it is full value.

Your own rules will fall somewhere between those two. So let’s say you enter a single rule:

  • At 50% chance to die, reduce the value of DPS gains by 70%.

So to figure out how much to reduce DPS gains at any point in between, we do a linear interpolation. For example, at 75% chance to die, you would be half-way between the built-in 100% extreme, and your 50% rule. So a linear interpolation would put halfway from a 95% reduction to a 70% reduction, or 82.5%.

If you create your own rule for 100% chance to die, it replaces the built-in one. Same idea for 0% chance to die.

Using NPS instead of Death Chance is just personal preference. In practice, the two are almost directly proportional – use whichever is more intuitive to you. If you have become used to aiming for particular NPS results when you simulate your character, you might like to define your “cutoffs” in those terms. Other people might prefer to think about death chance directly.

The NPS value at which you reach 100% and 0% death chance can vary a little from spec to spec or boss script to boss script. That’s why I tend to set up my Blender rules using death chance, it generalizes a bit better.