Why Remove Custom Stat Weights?

I have loved using AMR for a long time. Just returned for WoW after a decent break and reactivated my AMR. The feature I loved most was being able to utilize my custom stat weights to search for gear I like. I am returning an realize that functionality is gone. I though perhaps the simmed numbers match what I would get and that is not the case (Atleast for Destro Lock).

I think a great part of AMR is if I disagree with the stat weights being used I could just change them myself but still use features like upgrade finder and where to use bonus rolls. Since I cannot use custom weights and I disagree with the current stat weights for my class I see no benefit for using AMR like I have in the past. Please bring back fully functional stat weights and not just the secondary stat adjustment.

You can still do the same thing, we just don’t have input boxes with “stat weights” like before. If you use the “customize” feature, you can choose to customize your stats and use the sliders in there to specify how you want your secondary stats split up.

This update actually gives you even more control to customize the entire optimizer to do exactly what you want, even more than what we had before!

I know this isn’t what you want to hear but it needs to be brought to your attention.
All of the gear recommendations from AMR are based on simulated data, not from stat weights based on feelings. So what you’re disagreeing with is the simulated result.

Why do you think the results are wrong?
Are you getting your stat weights from another source?
Are you testing on a target dummy with no mechanics to deal with?

Have you read any of the posts, made from various sources, stating that stat weights are a bad way of dealing with things and that uneducated players put too much value in them?
Here’s one such post about them to save you looking.

From reading the OP, I think that he found the secondary stat customization option already.

The reason that we use that approach now instead of stat weights is that gear has become significantly more complex since the days of stat weights. The secondary stat adjustment actually gives you more control over ranking those stats than stat weights – you can reach a specific ratio/balance of secondaries.

Using weights to e.g. adjust how Intellect is values compared to secondary stats isn’t all that useful anymore. There just aren’t that many choices you have to make as a player that trade off primary stats for secondary stats. There are some trinkets with secondary instead of primary static stats, but usually the equip/use effect will be the more difficult choice to make there. So we give you the trinket customization options to more easily and precisely customize the optimizer there. We also let you adjust how gems are chosen – e.g. how much you value primary stat gems over secondary stat gems or vice versa.

Minor stats (move speed, avoidance, leech) appear almost exclusively as titanforged bonuses, so we give you an explicit way to control those in the context of titanforging.

After stats and trinkets, that leaves azerite – we give you a very flexible rule system to completely customize that. Stat weights would be of almost zero value in determining which and how many of most azerite traits you would want.

Give it a shot – so far people have told us that this new system lets them do just about anything they would want with their gear.

Hey,
The stat weights I was utilizing were from SimC based on my current gear. My BIS gear list based off those did not agree with AMR. It seems from the current AMR system (I could be wrong here) that all secondary stats (Crit, Haste, Etc…) are equal to each other and Primary Stats are always preferred but in my destro research Int is actually WORSE for me than secondary stats (This may change as I get more gear because secondary stats have a steeper diminishing return vs primary. I utilized the stat customization and the piece I am concerned with is I cant make Int less valuable. I just don understand why AMR would limit the type if gear I could go after based off my research and sims.

Why would we take away a feature especially when there can be different play styles? Destro lock this is less impacting but for example when I played a H Pally last expansion there was a divide among top H Pallies on how to gear (One route was more Crit based and the other was more Vers based) with the current option it seems to take away on exact weights I could utilize based off my research/preference and forced to do things the AMR method but I never have agreed with AMR stat weights and it seems that is still true but the other tools on where to find upgrades or what bosses to roll on was helpful based off my current weights. Perhaps I prefer to see the “weights” the current AMR sims and see if my findings match and adjust as needed but that seems impossible in the current methodology.

Edit: I understand weights will not include things like azerite or benthic but I think as a paying member for a product I should have the ability to adjust weights for primary and secondary stats based off of my Sims.

So a couple things:

  1. When you go to edit the custom secondary stat balance, it starts at a default on the UI of all stats equal. Our gearing strategies do not treat all stats as equal by default. We use a large number of simulations to create a statistical model that determines the relative value of different combinations of stats. The “stat suggestion” next to a gear solution will give you a rough idea of the type of stat combo it tends to go for.

  2. We do not recommend using stat weights generated by simc. Most people don’t use that method anymore – not even the people who use and write simc. They just don’t seem to work that well, and have been out of favor in the theorycraft community for several years.

  3. We didn’t take away a feature (the ability to customize how gear is ranked) – we just changed it! The game changed, so we changed along with it.

  4. Point for point, secondary stats do tend to be more valuable than primary stats. Our default gearing strategies reflect this. The game itself reflects this by giving you more primary stats than secondary stats for an equivalent budget of the item (e.g. an ilvl400 trinket with int would give 359, but crit would give 186 – blizzard has scaled things to roughly estimate that crit is worth a lot more point-for-point than int). We have you covered on that already – our base gearing strategy handles that trade-off with high accuracy. (And as mentioned before – minor tweaks to this have almost no impact on gear choices, because the game doesn’t give you much opportunity to change your gear based on this. Maybe when to use a JC gem or not… but we let you control that explicitly.)

Give the secondary stat balance another try in conjunction with the other options – I think that you’ll find it works quite a bit better than stat weights, and gives you a lot more flexibility.

I suspected you were wanting to use SimC weights.
Read the links about stat weights, even the guy running Raidbots has a disclaimer about how they’re bad and links to a few different discussions on them, one of which I provided earlier.

If you’re just reading a guide and using the weights they provide they’ll be from a fight against a target dummy with zero movement and using the best gear.
That isn’t a useful place to start as the gear you have to get you to that point is not the same, also you don’t have the option to just stand still and cast ignoring all mechanics.

You can sim the gear you’ve got vs the BiB result vs what other sources are telling you and see the difference, it’s probably close enough that it won’t really matter.
If you’re unsure how to run those sims paste your addon export as a reply and we can see what you’re looking at and set the sims up for you.

Occasionally BiB gives me unexpected results so I run some sims to see, it also chooses a trait I’m not fond of so I’ve used the customisation option to give me the best, from what I’ve got available, without the traits I don’t want to use.
In my case that’s Undulating Tides & Iron Jaws, while they’ll do more damage - in an ideal situation like a sim fighting a target dummy - in a real fight the uptime on Undulating Tides is not worth it for the fights I’m progressing on. Iron Jaws is only slightly better than the second trait and isn’t useful in M+ and rather than reforge for content I choose the tiny decrease in potential, if all the stars aline, single target damage.

Thanks this makes me feel better.

Where is this? I am not finding.

I have spoken to several high end raiders and they all utilize SimC and Raidbots. I was a minority in my group who utilizes AMR but like what was stated I utilized AMR for a gear finding tool based off the weights I generated from Sims. What makes AMR better for simming vs that of Raidbots/SimC and why would US top 100 players be “bad” for using this method still?

It is a change but it limits my methodology of finding upgrades how I have in the past. I just don’t see “proof” that I should trust AMR sims vs others. If I am wrong that is fine, just I need proof to get I guess a “warm and fuzzy”.

Thank you guys for the support BTW!

Here’s a screenshot:


The circled area is a description of what the gearing strategy is trying to do, and you can click the “view optimal stat graph” to see a visualization of stat combinations that tend to rank well. Note that the actual gearing strategy has a lot more data than is depicted here – it is meant as a quick way to visualize the general trends in the strategy.

Using simc and raidbots is a popular approach – but “stat weights” are not involved anywhere in that process. That uses a brute-force approach of just simulating combinations of gear and picking one from the results. Very few people use the stat weight feature of simc for anything serious anymore, but many people use the simulator itself.

If you want to brute force it using the AMR simulator, you can do that as well. We have a powerful tool for setting up combinations of gear to simulate. You’ll get very similar results 9 times out of 10. The main difference is that you run the simulations on your own computer instead of on our servers – the advantage of that approach is that it is free and unlimited (well, only limited by the power of your own computer).

Our optimizer (which is different than our simulator) uses a slightly different approach: we run millions of carefully sampled simulations for you beforehand, and then do a statistical analysis of the results to produce a model that then ranks gear. This sometimes produces slightly different optimizations than using raw simulation data, but at that point it’s kind of splitting hairs. It is difficult (impossible?) to prove which approach would hold up better in a real game situation. There’s just not enough hours in the day to perform a controlled enough in-game test of very similar sets of gear.

The customization features are designed precisely to allow you to use a different source or tool to optimize your gear. The design principle is this: if you look at our default advice compared to the advice of pretty much any other tool or website, it is very similar for almost everything. The differences are almost always tiny, or limited to a couple of specific items that are more popular in some circles, or a particular play style that is trending and favors a particular combinations of azerite traits or items, or a slightly different stat emphasis.

Instead of requiring a person to completely describe to us how to rank every possible piece of gear or combination of gear, we have you override just the major things of interest to you. Then we fall back on our default strategy to fill in the rest for you.

To help give you that “warm and fuzzy” feeling: all of our gear rankings are based on our simulator, the details of which are completely open to anyone to browse. You can see how every spell is implemented in detail (https://www.askmrrobot.com/wow/theory?spec=WarlockDestruction&version=live). You can see exactly how we assume you’ll play your spec (https://www.askmrrobot.com/wow/guide/class/warlock/destruction#generic). We test this data regularly and compare it to logs of real fights by high-end players.

If those things look good to you, then we guarantee that our default gearing strategies will allow you to be competitive at the highest levels – they are based directly on that information.

That said – we don’t think that our defaults are the only way you could gear. Quite the contrary – we know for a fact that many different variations will work just as well or nearly as well. Don’t be hesitant to customize, or look at popular guides or players and tweak the optimizer to follow their styles.