Optimizer does not appear to be giving the best simulated results for gearing

The optimizer does not appear to be giving the best-simulated results for gearing.

As the title suggests, I am seeing some strange things going on with AMR. I have been using it for a while now, and every time I mention it I get grumbles to it, and I’m told I should use RaidBots to simulate my gear and decide on the best equipment. I’ve ignored people just as much as they have ignored me for years. I like the features of AMR, and even if what I’m going to talk about isn’t fixable, I will probably still use it in some capacity. The gear manager itself is worth it to me. Now on to what I am seeing.

I broke down and started using RaidBots to sim my warrior. I am using their dungeon slice BETA tool/Top Gear against AMR’s Mythic Plus/Best in Bags.

After messing with RaidBots gear for hours (it takes seconds in AMR) I came up with the following as my best gear:

Raidbots report: https://www.raidbots.com/simbot/report/6AuGzeH7CoMsXkPYdRccgb

AMR Report: https://www.askmrrobot.com/wow/simulator/report/aeaaa139ed8443c1b3fa3ed72bf447ad

Note: I did not switch out the belt, to see the belt I am wearing click “Top Gear” in the Show Gear Difference Option under the character sheet section.

AMR’s sim shows this setup (Yes I went back to AMR and ran the RaidBots setup): 87,057 DPS

Raidbots shows: 86,858 (I did not change the belt, do pay attention to the larger number at the top look in the choices under the character sheet).

So, they were reasonably close. I figured you know what, AMR might be off a little, I’m okay with that, and I removed all of the restriction from my gear choice.

What I expected to happen was AMR would make some changes that would improve my DPS; after all, the results were similar. That is NOT what happened. AMR’s suggestion is here: https://www.askmrrobot.com/optimizer#us/illidan/solpraesidia.

This resulted in a simulated (on AMR) DPS of 80,094 DPS. 6,963 DPS lower than the AMR simulated DPS for the RaidBots setup.

I am baffled by this. The point of the optimizer from my view as a paying subscriber to AMR is to have the Best In Bags choice selected from the gear I have available. This is not the case based on the results I am seeing. I have not set up a lab to perform this comparison, so there may be mistakes in my results, but based on what I was trying to test and the method I went about it, I am, as I said, baffled and concerned. Please help!

For us to see what BiB picks for you, you need to use the “help” link above the gear table and post the snapshot id that you can generate with the “create support post” button. That lets us see what is in your bags and what settings you used.

I can look at this case in particular once I see that.

Keep in mind that you can’t compare total DPS numbers between our mythic+ script and the simc “dungeon slice” - they are not the same script. You can compare the relative values of sets of gear within each simulator - but comparing cross-simulator for this particular case is going to not work so well.

Correct. I am aware they aren’t the same script. I wouldn’t have known the correct terminology, but - what I did was sim from AMR.

The only thing I did with RaidBots was choose my highest DPS gearset based on thier utilities. I simmed here and got the numbers from here as well. To do that I hand selected the gear I wanted AMR to use (and traits, etc.).

I then removed all of the exclusions and locks I put into place to keep the Mythic + Gearset the same if I wanted to optimize tanking.

WIth the locks removed I simmed with AMR again. The second AMR sim was lower than the first (raidbots gear) sim. So AMR is simming higher with the gear sleection from raidbots than with the gear selection chosen by AMR.

90cca1ad04304e16b92d1358cdd9217c

That the number you are looking for I believe.

Here are the simulations I’m getting, just so you know what I’m talking about:

BiB using settings you had when I loaded the snapshot you provided:
86,333 DPS

BiB unlocking everything and turning off custom settings:
89,812 DPS

Manual set of gear you put together on raidbots:
91,433 DPS

The optimizer finds a set within about 1.75% of the max you found manually. That’s pretty good. We aim for a bit better than that when we can, but the number of complex interactions between systems (talents, azerite, corruption, essence) has gotten crazy at this point in BfA.

In this case I think that the estimate we use for Sk’shuul Vaz is the culprit - that item is a pain because it scales off armor. It introduces error into the optimizer that we can’t avoid completely. If I exclude that item from the optimizer and run BiB with no custom settings I get this:
91,893 DPS

So, I think the optimizer is working well in general, but that one particular effect is a bit over-valued for your available gear. I can see if there’s a way to tweak it that works in general, but it’s hard to determine if this is a specific case to you or a general case.

1 Like

Thank you. It is working a lot better now. It’s not the same as the RB setup, but I’m fine with the miniscule differnce in there now. Doesn’t even mean that one is real life higher than the other, but they are far closer now.

I get the same grumbles from people when I say I use AMR, I’ve run out of ideas on how to change the perception.
From what I’ve worked out it seems to be a case of peer pressure and ignorance.
The Raidbots crowd take the numbers as gospel without actually thinking about them, I notice the same thing with Pawn users.
You’ll hear things like, “That’s a 36% upgrade for me,” after a boss kill. I’ve given up trying to explain to them it isn’t, they say the same thing about items with procs or corruption.

I’m familiar enough with my class to notice when something seems wrong, I’ll try and work out what, run many sims and then post on this forum if I can’t work it out. Sometimes I’ve found a bug, other times it’s just within the error margin.

1 Like

If you look even deeper you can find more differences between both sims.

For example AMR use in most of situations very simply rotations, that anyone can learn and use and they have combos that everyone should know. Also they are very practical to use in any situations, but they are lack of very advanced combos.

SimC use very complicated rotations, that include advanced combos, but in practice i don’t know that you would be able to use them 1:1 without assistant addons or rotation helper, if you aren’t very talented player. Also, if you use them in practice, then they have continuity problems and often get stuck, because the creators often didn’t anticipate that something might interfere with their rotation, or they are simply bugged. This is especially visible when building combos during encounter with movement.
For example Demonology try avoid to use Soul Strike, when Unholy DK try avoid to use Soul Reaper.

Then simple rotations from AMR are more practical to use.

Currently i’m using AMR only, because my rotation is simply and im use “alternate build” for my class. Then i can faster sim everything on AMR for 4 different situations.
The only thing what i miss here are options to choose amount of targets and Fight Length to precise my results and avoid useless things. At the moment i must exclude them manually.