Has exclude item been added?

Is it still not possible to exclude items? There was a post in July saying it wasn’t, just wondering if the situation has changed.

Clarification of why: AMR best-in-bags really really likes https://www.wowhead.com/item=134542/jeweled-signet-of-melandrus&bonus=3573:1562 (increases auto-attack damage by 10%)

Auto-attack is only about 3% of my damage in a best case scenario so in-game combatlogs & simulationcraft both strongly suggest AMR is over-weighing this affix.

And given the range of random affixes & ilvls possible via relinquished gear, there are always going to be a few items like that.

I would like to be able to manually exclude such items without having to pre-emptively guesstimate which combo of rings I should be locking in place. If I was already certain on the gearset I wanted to lock-in then I’d hardly be using best-in-bags, no? =)

Do you have an Addon export of this case so I could take a look?

Export is below but I’m using a custom gearing strategy (created via simulation, using machine learning, versus Goroth with default rotation).

The default Goroth strategy isn’t as keen on the Signet - instead it’s terribly keen on re-gemming/chanting for crit (also a bad idea but can be readily solved via locking in rather than needing exclusion).



You can link your custom gearing strat, Swol will need that link to better answer your question.
Just click the view button in the strat picking section, then copy and paste the link here.

When I import that string and hit the Simulate button I get:
1252220 DPS
When I press the Simulate button after doing BiB I get:
1306590 DPS

That’s almost a 5% gain!

That’s a pretty big gain there, not sure why you think it’s wrong as you haven’t provided any reasoning or links to simulation reports showing it’s wrong.
I’m guessing you’re weighting someone’s opinion from a guide or Discord higher than you should. If they don’t provide you all the data with how they come up with their recommendations you’ve got no way of verifying it, if you’ve got no way of verifying it it’s not something you should give much value.

This is getting away from my original question - ‘has exclude item been added?’ - the answer to which seems to be no.

What I’d like to know now is whether the functionality is going to be added or whether it’s explicitly off the table.

It’s possible to force it via vendoring a (non-legendary) item, exporting AMR data & then buying back the item, but I’m not keen on that solution for obvious reasons.

Anyway, regarding the gearing/crit/opinion thing…

On the reasons why I dismissed crit out of hand despite the simulated 5% gain (and ignoring the fact the initial string wasn’t actually my primary gearset at the time - notice that the prydaz isn’t even gemmed/chanted)…

No guides or discords involved. I’ve had a low opinion of hunter theorycrafting for years, although it does seem to have improved a bit lately. There are three elements that artificially boost crit in that sim:

  1. It’s a gearset that’s comparatively light on crit
  2. The 2pc bonus BiB equips strongly favours crit
  3. It’s simulating a pure single target fight

If you swap to a conventional mastery gems/chants setup, using AMR’s suggested gear, and run a Sassz’ine or Mythic+ sim then mastery will pull ahead. And on single target that sort of mastery setup is not concerningly behind. So the AMR BiB sim is ‘correct’… it’s just not good advice, and that’s due to the limitations of the information it’s been given as an input.

I’d also argue that AMR over values the legendary cloak, which I personally feel disrupts the flow of your rotation via its haste proc (much like running non-expertise capped used to be a theoretical boost to some melee but many top end players capped for ease of use). Interestingly enough the Simulationcraft runs I’ve done actually regard the gloves as the higher DPS legendary of the two, but I’d personally be making the argument largely based on the utility the gloves provide. The cloak provides either leeway for sloppy play, or possibly gives you the opportunity to add a shot into every windburst-generated vulnerability window…

Infact, now I look again, I’m honestly not sure which reason is causing AMR to boost the cloak for that gearset. When I tried swapping in the non-legendary cloak option & the legendary gloves, leaving AMR’s gear choices otherwise unchanged, the gloves option simmed higher on default Goroth in AMR too (albeit with higher error %).

And therein lies the root of my problem.

AMR has always had some black box tendencies… but with the increased number of special procs on items, the emphasis on simulated targets, the increased difficulty of double-checking via Simulationcraft, and the non-human readable stat weighting system, it’s starting to suffer from the problems I’ve seen time and again with algorithmic trading systems. More processes get included & so it’s both more vulnerable and harder to check, and even if it works 99% of the time that leaves 1% of problems (e.g. Long-Term Capital Management). Any system that requires blind faith is a bad system.

Algorithms only work well when you clearly understand the weaknesses in the data going in, and the potential flaws in the data coming out, and intelligently filter to ensure there isn’t garbage clogging up the works.

This requires the ability to both force include & force exclude items (imo), especially in the case of all special proc items.

PS. The recent addition of alternate legendary suggestions is a good start, but does not go far enough imo.

To answer your question, we may still add back in the ability to exclude specific items. We are plugging away at our feature list as fast as we can, trying to bring the site to our complete vision for it.

I’m not sure I completely follow your reasoning for the rest of your post, or maybe I just disagree with it a little.

If you are using the single target adaptive strategy, why would you think the advice is bad if it tells you the best way to to single target damage? We have a multi-target adaptive strategy based on the mythic+ script as well. If you switch to that one, it favors crit less and mastery more, as you would expect.

The idea that you can “merge” gearing strategies for multiple fight types into one generic overall gearing strategy is an idea that we disagree with. We think that gear should be optimized on a per fight basis, since that is the only way to optimize based on real data. We are going to be updating the addon to support multiple gear sets from multiple gearing strategies, which will then let you quickly swap gear on the fly in raids.

I’d argue that we have a deep understanding of the strengths and weaknesses of the data we are working with. This is why we are trying our best to educate our users about the margin of error that is involved. The legendary customization feature is another step in trying to help people see that gear optimization is not a black and white problem with just “one” solution. There is a cloud of near-optimal solutions that we pick one from. Personal preference may push you towards a different one than we would pick by default.

You might want to check out my recent article about the MM hunter rotation:

The optimize picks the gloves for you right now for single target, and not the back. It favors the gloves in all the sims that I try. The back isn’t bad - there are combos where it is very close in value, but generally not quite as good. I understand from a play perspective that the gloves offer great utility as well, which may allow the player to perform an optimal rotation for more of the fight. There is really no way for us to assign a “value” to that though, which is why we added the feature to make it easier to customize the results.

Our “stat-weighting” system is human readable, imo. The problem is that the gear optimization problem has become way more complex in legion with so many special effects on items. Stat weights themselves are of extremely limited use when it comes to optimizing gear now. We have to use simulation data to optimize gear, otherwise we’d basically just be making things up based on anecdotal evidence.

As far as picking stats on gear around the special effects, we have known for quite a while that simple stat priorities were never great at picking gear which matches the simulation data that created the stat weights. Stat weights are a linear predictive model. People have used stat caps to try to shoehorn stat weights into the simulation data a bit better, but, they fall far short of the mark. In reality, there are combinations of stats all over the place that can be optimal. This is hard to show/explain to the user compared to simple stat priorities, but we’re giving it a shot because we don’t want to just tell everyone that they HAVE to stack a particular stat when there are other stat combinations that would also be good.

In the case of the export string you provided, your gear already has a lot more mastery than crit. The main loss of mastery is from putting on lower item level tier items. The main gain of crit is from replacing the poor haste/vers ring with a high crit ring. Those are both reasonable optimizations for a MM hunter, imo. The 2T20 is very strong.

1 Like

Hi ! I would too appreciate very much to be able to exclude some gear items when using the best in bags functionality.
I play a tank dk and I like to switch between 2 gear sets, one for maximized DPS and another one for max NPS, based on the gearing strategy settings when I use Best in Bags.
But there are some of my gear items that AMR tells me I should equip for both of these gear sets.
And I don’t want to gem and enchant them each time I switch…
I would like to build 2 sets of totally different gear items and it would be great, when I use Best in Bags for maximized dps gearing strategy, to be able to exclude the gear items that are already part of my max NPS gear set.