Note: the linked article is an Ask Mr. Robot analysis of the effect of haste plateaus on hunter DPS — they’ve done this so far for SV and MM. Ask Mr. Robot is a site that automatically calculates potential upgrades for your character, including gems, enchants, and reforging.
Ask Mr. Robot: TLDR: Analysis shows that “Haste Plateaus” are not important for Hunter gear optimization. Hunters, like every other DPS spec in the game, will achieve optimal DPS executing their abilities based on a Priority Queue. Any “rotations” that they may fall into for certain portions of a fight are incidental and not important in and of themselves.
Hunter gear optimization theorycraft is based largely on the idea of “shot rotations” and reaching specific “haste plateaus” where you can fit an exact number of filler shots between longer cooldown instant shots. For a Marksmanship Hunter, the idea has been that you want to gear such that the cast time of Steady Shot allows you to execute a specific number of abilities between Chimera Shots, thus minimizing or eliminating the time that Chimera Shot is available and not being used. This is complicated by haste procs, bloodlust/heroism, different phases of the fight, etc. So, the benefit of the theoretical haste plateaus will not be realized for 100% of the fight, but, the idea is that they are still very meaningful.
Mr. Robot sat down and thought about this for a little while. If this strategy is viable for Hunters, shouldn’t it be viable for other classes as well? Take an Elemental Shaman, for example. They want to minimize or eliminate any time that Lava Burst is cooled down and not being used. Would it not then make sense for them to gear such that they can fit the perfect number of Lightning Bolts between Lava Bursts? The idea has been explored in the past and it turns out that doing this does not increase DPS. When Death Knights were first introduced into the game, the idea of ability rotations was very popular, to reduce or minimize any time where runes were available and not being used. But, it was soon discovered that using a priority queue approach to play resulted in better DPS. To Mr. Robot, it seems very odd that Hunter theorycraft is still focused on rotations – so much so that the recommendation is to tailor gear around particular haste values. So, Mr. Robot put this to the test! The idea looks logical on paper, but, it doesn’t work out for any other class in the game. There are some abilities in the game that benefit greatly from achieving certain values of Haste, abilities such as Improved Devouring Plague and Conflagrate, but, Hunters do not have any abilities with that mechanic.
Mr. Robot decided to test a number of gear sets. If being near a haste plateau is optimal, we should see DPS peak with gear sets optimized right at the plateau, and then it should decrease as we both add or reduce the total amount of haste.
Thanks to Garnath for the tip!
Loading ...
While I haven’t read the entire article yet, this quote seems like he missed the point of haste plateaus:
“If being near a haste plateau is optimal, we should see DPS peak with gear sets optimized right at the plateau, and then it should decrease as we both add or reduce the total amount of haste.”
Being near a haste plateau doesn’t mean that our DPS increases, and moving away from it decreases, it means that how much DPS we gain from a certain amount of haste changes based on where we are relative to those plateaus.
More haste is always more DPS. Always. But at certain points, it’s giving us more DPS than at other points, so the relative value of haste goes down, but it’s still always a positive value in terms of DPS.
The first time that I ever came across Mr Robot, I gave it a shot. I tried uploading my character info and I hit optimize. The results were frightening….Had me reforging crit into haste, crit into hit and seemed to favor haste over almost any single stat. I immediately dismissed the site as not optimized for hunters. I already know of the haste issues that can occur with Hunter DPS modeling so I somewhat wrote it off to that.
I did read the entire article here and while some interesting points are made, I think the above comment is dead on. While I am no Hunter mathematics expert by any means I really think Mr. Robot is extremely overvaluing Haste.
After reading his article, I would really recommend to pay it no attention. He isn’t understanding what a Haste Plateau is, and is trying to use data to find Haste Peaks (certain points that have higher DPS values than those around them).
A good theoretical analysis will have all things constant and measure the effects of haste across a wide range (very low numbers to very high numbers). Theoretically, it will show a graph (Haste as x-axis, DPS as y-axis) that looks like a set of stairs.
A good practical analysis will have a given gear set, and spend time reforging to as many different levels of haste as possible in between totally away from it and totally toward it.
If I had the time, I’d totally write a guest post. Hrmm… Anyone interested in helping with the data collection? (You know, the time consuming part?)
“A good practical analysis will have a given gear set, and spend time reforging to as many different levels of haste as possible in between totally away from it and totally toward it.”
Isn’t that exactly what Mr. Robot did?
Nope. He only looked at a very small range of haste right around a theoretical haste plateau, looking for haste peaks. To understand haste plateaus, you need to have data that encompasses almost a full spread of a haste plateau.
Unfortunately, with a given gear set, it’s hard to see beyond a small range of haste values, making it harder to see the full effect of haste alone.
That’s a double standard. You can’t expect Mr. Robot’s testing to encompass an enormous range of Haste and at the same time stick to reforging.
I honestly fail to see what you’re critizing that isn’t a necessary limitation in the mode of testing itself.
I agree with Fugitivelama that Mr. Robot seems to be over valuing haste. I plugged in Mr. Robot’s “optimized” numbers, all of which leaned heavily in favor of haste, into Femaledwarf.com and saw a very significant decrease in DPS.
In contrast, using Reforgenator’s (reforging recommender add-on) suggestions in Femaledwarf.com resulted in a much higher dps configuration than Mr. Robots’
For the time being I am going to continue with Reforgenator’s recommendations until I understand this better.
By the way, Im survival spec.
What you notice isn’t unexpected. Both Mr. Robot and Femaledwarf are theorycrafting based on their own custom algorithms.
They however use different algorithms and therefore the output/result will be different.
The only way to know who’s right is to put those numbers to the test, equip the exact same equipment you used to generate them and start shooting at things using the exact same rotation/shot sequence used to generate the theoretical results.
Femaledwarf assumes haste plateaus/caps are vitally important, thus their numbers take that into account.
Mr Robot assumes they’re far less relevant, so their numbers reflect that.
Only extensive field testing (not a single 5 minute session by a single player with a single set of equipment but a wide range of people with different gear in different situations) may show who’s right and who’s wrong (or more likely that they’re both somewhat right yet somewhat wrong).
Of course there’s always the point that the whole theorycrafting is nice and good but doesn’t reflect real world dps well because it doesn’t take fight mechanics into account (e.g. in a moving fight you’re not standing still long enough to use steady shot twice in a row to get the improved steady like clockwork).
So, maybe its just me, but what I love about this analysis is that it tells me something simple. Its about the player.
If you have Hunter A specced out so that he is in “Set 1S” and pulling just under 31k, and you have Hunter B in “Set 4S” and pulling just over 31k…Are you really complaining either way?
While I love our community for having such a detailed outlook on the data that drives our performance, I also have always felt perhaps we over analyze it at times to an extreme. Admittedly, I am not the data hound and min/max freak that many other hunters are. I do my best to ensure I am not under hit cap, or wasting hit and way over it. I try to ensure I am on target for my haste levels and then gain as much crit as I am able. I keep my rotation on point as best I can. etc etc etc.
In the end you can see if a Hunter is doing their job or not. And as an RL so long as my Hunters are not a liability, and are pulling comparable dps to other dps classes (read: They dont have to top the chart, but should be competitive)…then I am happy over all.
Definitely a good philosophy to have.
In reality, the DPS difference between properly reforging/gemming/enchanting and not reforging/using obvious gems & enchants is typically less than the margin of error due to RNG and a lot less than the difference between players of different skill levels using the same equipment.
So while it’s still an increase that many hunters would rather have than not have, it shouldn’t be thought of as something that will make up for a lack of skill.
Agreed whole-heartedly. I would also add that I have played alongside great hunters who really knew what they were doing but just had grating personalities….and along others who were barely average but such a great and nice person were totally worth the lesser performance!
All the reforge testing I’ve done confirms this as well. I’ve never found an extreme way of reforging that changed my dps more than the margin of error. This really relates well to Frost’s recent article about skill.
I think that I need to re-word some of my write-up of the analysis that I did. At Mr. Robot, we take a very practical approach to gearing. I did not mean to imply that Haste scales linearly, since it does not.
The point of the analysis was that, for gear optimization purposes, we can ignore the non-linear scaling of haste and value haste at its highest value. This will safely result in a DPS increase almost all the time.
For those of you who report that you have used the new stat weights on Mr. Robot and seen a DPS decrease after optimization when you use FemaleDwarf or SimC – Mr. Robot would very much appreciate it if you would share your before and after gear with him on our forums at forums.askmrrobot.com.
That’d be great. I think it came off a lot different than it was meant to (and my previous comments were based off of how I initially interpreted it, so my apologies there for possibly commenting a bit reactively)
Since, I’ve read Frostheim’s post about actually being able to sit down and talk with you about it for a while, and it sounds a lot more clear than from reading the initial write-up. So I’d love to see the write-up after the rewording.
I think Mr. Robot is missing the point. Shot priority? Rotation? Hunter DPS isn’t about any of those. It’s about our main spell (CS/ES/KC) and focus. Our goal is to have enough focus whenever our main spell comes off cooldown without overcapping. That’s it.
Well, maybe it’s something like a shot priority with variable priorities (the only constant being the main spell)…
So are you saying if you have the focus for a Kill Command available, but Kill shot is up…you don’t go with the shot priority of KS > all?
Or better yet, if you have just enough focus for an ES OR a KC, but ES is 1 sec away from cooldown and KC is available now…you’d hit KC and not wait that 1 sec for an ES?
Personally, I think the issue where haste muddies everything is in the focus regen. At what point does haste become a greater boon based off of it providing us more focus faster?
No, I’m not saying this. Maybe I shouldn’t write comments when I’m totally tired… :)
Afaik, KS is always better than KC due to its higher damage without eating focus away that’s needed for KC. And of course, we act slightly differently when we know that we’ve got 2 GCDs to passively build up focus ahead of us. But that’s only 20% of a fight at most, thus situational and not general.
What I meant with “variable priorities” is that sometimes, we favor CS and sometimes we favor AS, depending on how much focus we’ve got, despite AS being the higher damaging shot. Because we want to cast that KC as soon as it’s off CD. That’s not a shot priority as I understand it, because priorities shouldn’t change then. It’s a dynamic focus-and-signature shot-dependent decision every time we fire something.
Frost made an important note along the lines that it’s important to know haste plateaus in order to react accordingly to them (= know when to push off the signature shot slightly) and that simulators can’t do that as well as humans do. I guess that’s what I tried to say too. Thinking a few seconds ahead and planning the next shots accordingly isn’t something FD&co. will ever achieve, but it is heavily impacted by haste.
The entire analysis is off-balanced… It plays with haste, but it treat haste as is all that we have.
The entire idea of haste plateaus is to find the points where haste is beter than the other secondary stats and the points when it becomes less efficient. Of course, haste will have always some eficiency, hastening some of our abilities, but the main question for any secondary stat is: cant we get more if we reforge it?! All stats will give us some gain, but some are better than others, hence the priorities of each specc and class.
Now, beside the regen purpose, what is a SV gain from haste? Faster CS casting, yes?! But that becomes efficient only when we gain an extra CS, between ExSh CD. Before that is just marginal profit: 1,68 or 1,72 CS cast is kinda the same, because you are still forced to wait for ExSh CD. So, only when we reach a plateau, we gain maximum efficiency from haste, as 3 CS will be always more dps than 2 CS.
Now, where is the mistake of this analisys… Between plateaus, what is happening with the haste if we dont wanna let it there as it comes? Is simply washed away or stored in the bank? Nope! Is reforged, usually into crit, if hit cap is reached.
We agree then that haste is king at plateaus, as it give us a tangible bonus (an extra CS). So, the only question that remains is if we gain more dps from simply stacking haste until next plateau or if we reforge it into crit?
Given all these, there is no point in asking if ”haste beneficial from us all the time”, because it will allways be some gain from it. A serious analisys should question ”how is haste better used between plateaus: as it is or reforged into crit?”
Agree and I’ll also add that I have been playing with a big hunter who really knows what they are doing