I have been running a few sims on my disc priest and have noticed that when using the Best in Bags tool AMR shows stat values as vers > crit > haste=mastery. Information elsewhere seems to show pretty much the exact opposite of this. Best in Bags is also telling me to use a 171 ilvl ring, Entwined Gorger Tendril with +12 mastery enchant over a 187 ilvl ring, Bloodoath Signet with +16 haste enchant.
So since everywhere else would most certainly recommend the latter, I ran two separate simulations; one where I am wearing Entwined Gorger Tendril, and one where I am wearing the Bloodoath Signet. With the 171ilvl ring the sim shows just over 4k HPS with just over 4.5 ally deaths and with the Bloodoath Signet itâs 4.3k HPS with just under 4 deaths. This looks like a giant upgrade for just 1 ring, so I am completely lost as to why the Best in Bags tool is telling me to use 171 ilvl Entwined Gorger Tendril?
I have been gearing purely using AMR so far and ignoring all of the information available elsewhere, but it really looks like Iâve been doing it wrong.
Can you explain whatâs going on in the sims please, and why it seems to be so off? Iâm also curious if it is just a disc priest thing?
I would need you to link the simulations you did and some more information about your character. It shouldnât be possible for one ring to swing the results that much - in most cases this is usually caused by inadvertently changing some other settings as well.
As far as general stat priorities - it is normal for ours to be significantly different advice from other sources. We are the only people using simulation to do our theory - so we offer a totally different perspective. Versatility in general gets ranked higher by us because we take into account the actual damage reduction you get from it - which is effectively more free healing.
Haste is a stat that cannot be given a real calculated value without simulation, in my opinion. That is one of the main reasons I took on the healing simulator project. It isnât possible to determine the overall value of haste with respect to your mana pool and total healing ability over the course of a fight with anything simpler than a simulation of the fight. You could calculate a way to get the most burst healing in a short window without a simulator, maybe - but once you look at a whole fight and are mana-limited⌠you really need something more than a spreadsheet.
Healing is also a very âfuzzyâ thing to optimize for. You never have the exact number of healers necessary to heal the damage being done in a fight and not one bit more. You have more healing than you need. It becomes very ambiguous what it even means to have the best gear. Our method with the simulator gets good results that anyone can use to be successful. We would never say it is the only good way - just âAâ good way.
I do concur that one ring should not result in such a large increase, I wouldâve perhaps not even expected such an increase to even go from no ring to either of these rings. Both this result and the BiB results confuse me, hence the post!
I didnât save the sims, but Iâve re-run them. As far as I can tell Iâve kept everything the same apart from the ring.
I do get what youâre saying re stats, but does make me question the AMR stat weights when all the top parses on warcraftlogs of disc priests are stacking haste first and that also seems to be the general consensus best secondary stat.
There are no âAMR stat weightsâ that isnât how the BiB selection works.
As for the sims you initially did theyâll be in your simulation history. On the simulator page thereâs a white bar on the left with a menu icon up the top, if you click on that you can see the ones youâve saved and the history tab.
I had a look at the results of the ones you reran, Iâm not an expert but the stand out for me was Mindgames, for some reason it has a count five times greater with the extra haste. Sure haste will give you âmoreâ but five times seems like a bug!
We will need to wait for Swol and see what he thinks.
Yeah, there is something wrong with the heal on Mindgames. Iâll work on a fix for it and I think that will clear this up.
Parsing high with a healer has very little to do with effectively keeping your raid team alive. It is a different optimization problem entirely. If you want to get a high parse as a discipline priest, getting higher haste to maximize those spirit shell or evangelism windows is effective, especially when coordinate with your team.
Our simulator/optimization will find the best way to get maximum healing from your available mana. Haste is rarely the way to do this until you donât have to worry so much about mana. This is a common difference between our theory and what is popular.
Once I have some time in the coming weeks after the holidays, Iâm going to be experimenting with some different healing scripts/models - maybe we can make a script and rotation that is focused more on high parses.
When I heal on my raid team, I coordinate everyoneâs cooldowns with the fight and so I donât have to battle the other healers to get the heals. It doesnât matter if I have more haste, I can go for more theoretically optimal builds. Not everyone has the luxury of a guild team that knows they are a sandbox for my theory
Cheers Swol. I look forward to the update and re-running the sims.
I get where your theory is coming from and obviously agree that if mana is an issue haste decreases in value. Fwiw, I am not looking to hit big parses per se, but believe that overall bigger HPS will largely result in fewer deaths⌠at least until the mana pool depletes and you become near useless. I also believe efficiency and not over-healing too much is undervalued by many, but think that the HPS that results in the fewest deaths and 0 mana at the end of the fight is what we should be striving for. If this is what AMR is trying to simulate then I am totally on board.
Just another couple of questions if you donât mind;
Are all DPS classes/specs modelled to maximise a DPS parse? Seems thereâs not really much else to consider other than not dying but I imagine this is modelled into the rotation when moving?
How exactly does BiB work? Obviously it isnât doing a separate sim for every combination since you can have thousands of different combos! Is it working by picking gear that most closely matches your stats to the âoptimal stat distributionâ as shown from the graph?
Yes, DPS is a much simpler case. DPS is the only metric that matters. Moving will cause downtime - the rotation/simulator knows what can be used while moving and plays around it as best possible. If one were so inclined, they could make a script with the specific timings of a fight and create a rotation to specifically time up cooldowns with that fightâs mechanics. No one has actually done that yet⌠but it could be done.
In broad strokes, we simulate a few hundred thousand or a million different setups to get some data. A setup included talent, covenant, legendary, stats, etc. With that data sample we create a mathematical model. The model can predict the result of a simulation very closely. When you run BiB, we use that model as the scoring function along with some tailored search algorithms so that we can examine huge numbers of possible gear sets and find one that is highly likely to simulate highest in our simulator.
The model smooths out a lot of the noise inherent in simulation data. We believe this is a superior method to using raw simulation data because simulations themselves arenât perfect - it is best to get lots and lots of data and fit a model to the overall trends instead of reading too hard into any one result/data point.
Nice one. Cheers for your answers and time. I do enjoy the functionality and interface of AMR fwiw, hope none of what (or how) I said came across as harsh or critical, just want to understand how everything works!
You didnât come off as harsh or critical. You should see what some other people have to say about how we do stuff⌠We have developed extremely thick skin over the years.
Whether people agree with or like what we do, we think that we are a valuable part of the community because frankly⌠who else is there to provide a second opinion or methodology at this point?
I play poker for a living where sims have become the norm for learning/improving so Iâm always searching for sims when I play WoW and have since using rawr back in WoTLK days, though a sim is only as useful is itâs inputs; shit input = shit output. I think there seems to be a good methodology in place here though from what I can tell and agree you should be seen as a valuable part of the community.
