BM Hunter Opener Assistance (Mistress Mythic)

[Hygeia]

Hi Guys

I'm pretty new to hunter came from a healing perspective. I would love to get some insight and assistance.

Here is my profile: https://worldofwarcraft.com/en-gb/ch...aenor/nimerisa

And this is a person i know and just want to see what makes the big difference: https://worldofwarcraft.com/en-gb/ch...ak/sultangurde

Log comparison:

I'm looking at the big difference in opener, where i burst for like only 2Milj and he for 6milj.

My log: https://www.warcraftlogs.com/reports...done&source=27

His log:https://www.warcraftlogs.com/reports...22&view=events

Am i doing something wrong or playing wrong here?

Any help will be appreciated.

Regards
-N-

[Azortharion]

He saves his cd for fish and you don't.

[Hygeia]

Wow, didnt think that could make such a huge difference.
Ok great i will try that thanks for the quick response.

[syrupyroffle]

There's also a separate and bigger issue where your setup is a bit unoptimized.

I'm not sure if you being "fresh" to hunter means you lack the legendaries for it, but you typically don't run boots with belt + killer cobra talent. There are 3 ways to play BM right now (although highly dependent on your gear and legendaries).

1) Dire Frenzy (T2) + Killer Cobra w/ some combination of shoulder/belt/chest legendary (with heavy reliance on stacking haste + mastery)
2) Stomp (T2) + Killer Cobra w/ belt + class ring (no real definite stat priorities here, although you want a good amount of crit (~30+%) and mastery)
3) Stomp (T2) + Aspect of the Beast w/ boots + sephuz/prydaz/class ring (less emphasis on mastery, same crit as (2), and haste is more of a personal preference for fluidity of rotation although you don't want to drop below ~10-15%)

Both 2) and 3) gain a lot of damage from having good T19 2P tier, whereas 1) is full item ilevel (hence the option to use tier legendaries). 2) and 3) are tricky to gear out as well (since you need a good balance of crit/haste/mastery) - hence every BM hunter saying "sim it" for any drop.

Most people find 3) to be the go-to / FotM, but as you can see from logs, 2) is also quite competitive. 1) is pretty far behind. In your case, 2) "might" perform slightly better than 3) simply due to your lower ilevel and less overall secondary stats to work with as 3) is more reliant on secondary stats than 2).

As for the specific comparison you mentioned,

1) Saving for fish is huge, like Azor said. You can probably use BW on pull and have it up in time for when the fish spawns, but definitely save AotW.
2) Boots really doesn't synergize well with belt + Killer Cobra. Boots is basically a "half-reset" compared to Killer Cobra, and essentially gimps your rotation during BW (which where you do all your damage) since you have to choose between either waiting for KC to come off CD or "wasting" focus to finish resetting KC with CS. Lowers your burst potential compared to the other guy.
3) Owl is a much better ST burst trinket than Tome, although you can optimize Tome for that fight by using it on a low add to proc a longer crit buff. Potentially higher output than owl, but much worst burst if we're looking at just your on-pull damage.

[syrupyroffle]

There's also a separate and bigger issue where your setup is kind of unoptimized.

I'm not sure if you being "fresh" to hunter also means you lack the legendaries for it, but you typically don't run boots with belt + killer cobra talent. There are 3 ways to play BM right now (although highly dependent on your gear and legendaries).

1) Dire Frenzy (T2) + Killer Cobra w/ some combination of shoulder/belt/chest legendary (with heavy reliance on stacking haste + mastery)
2) Stomp (T2) + Killer Cobra w/ belt + class ring (no real definite stat priorities here, although you want a good amount of crit (~30+%) and mastery)
3) Stomp (T2) + Aspect of the Beast w/ boots + sephuz/prydaz/class ring (less emphasis on mastery, same crit as (2), and haste is more of a personal preference for fluidity of rotation although you don't want to drop below ~10-15%)

Both 2) and 3) gain a lot of damage from having good T19 2P tier, whereas 1) is full item ilevel (hence the option to use tier legendaries). 2) and 3) are tricky to gear out as well (since you need a good balance of crit/haste/mastery) - hence every BM hunter saying "sim it" for any drop. Also, legendaries are more flexible for 3) than 1) or 2) - you see players showing apex ring or wrists coming out on top as well.

Most people find 3) to be the go-to / FotM, but as you can see from logs, 2) is also quite competitive. 1) is pretty far behind. In your case, 2) "might" perform slightly better than 3) simply due to your lower ilevel and less overall secondary stats to work with as 3) is more reliant on secondary stats than 2).

As for the specific comparison you mentioned,

1) Saving for fish is huge, like Azor said. You can probably use BW on pull and have it up in time for when the fish spawns, but definitely save AotW.
2) Boots really doesn't synergize well with belt + Killer Cobra. Boots is basically a "half-reset" compared to Killer Cobra, and essentially gimps your rotation during BW (which where you do all your damage) since you have to choose between either waiting for KC to come off CD or "wasting" focus to finish resetting KC with CS. Lowers your burst potential compared to the other guy.
3) Owl is a much better ST burst trinket than Tome, although you can optimize Tome for that fight by using it on a low add to proc a longer crit buff. Potentially higher output than owl, but much worst burst if we're looking at just your on-pull damage.

[Hygeia]

@syrupyroffle

WoW thanks alot for this advice!

Yeah unfortunately i lack legendaries, also i dont have any T19 pieces Only pieces i have is, Head, shoulders and legs and then the boots and belt)
So playing (2) without T19 will be worth it? I have owl so i can swap that for the trinket im using atm.

I would have tried MM but i have no legendaries for it..I feel really gimped without having T19, but i see some hunters do well without that, so if you can recommend pieces i should equip and maybe what stats to aim for will be great.

[Tharkkun]

Best thing to do is go to http://raidbots.com and use the Top Gear button. Then grab the Simpermut addon from the link in that page and export all your available gear. It will figure out what the highest DPS combination should be.

Keep in mind you may need to limit it to 1-2 pieces per slot or number of iterations will be too high to run. It takes a bit to narrow the gear down and you may need to run it multiple times if you can't fit all your gear into 1 sim.

Or you can use the Gear Compare option in Raidbots and just add pieces from your bags and see if they are upgrades over what your currently equipped gear is. It's too hard to make recommendations on individual pieces because ilevel and sockets change the value of each piece so you must sim for every upgrade.

[Hygeia]

@Tharkkun
Thanks for the information, do these sims like Mr Robot and Raidbots take into consideration the type of build you play? E.g. BM Stomp or BM Frenzy based on ur talents.
Since for example i simmed both spec's with various gear pieces and my stomp build gave me a 1.19mil dps on a fight like mistress and my BM Frenzy build with gear chosen gave me 1.59mil dps, but i do like 400k dps more with stomp than frenzy...
But thanks ill try raidbots as i havent used that before.

[Tharkkun]

@Tharkkun
Thanks for the information, do these sims like Mr Robot and Raidbots take into consideration the type of build you play? E.g. BM Stomp or BM Frenzy based on ur talents.
Since for example i simmed both spec's with various gear pieces and my stomp build gave me a 1.19mil dps on a fight like mistress and my BM Frenzy build with gear chosen gave me 1.59mil dps, but i do like 400k dps more with stomp than frenzy...
But thanks ill try raidbots as i havent used that before.

They do. There's a talent sim comparison on raidbots as well. I would use that over AMR even though the AMR owner will probably come here and bitch me out.
Just make sure you do roughly 10k iterations, patchwerk, 1 boss, 5 minutes or you'll see weird DPS results.

[Swol]

@Tharkkun
Thanks for the information, do these sims like Mr Robot and Raidbots take into consideration the type of build you play? E.g. BM Stomp or BM Frenzy based on ur talents.
Since for example i simmed both spec's with various gear pieces and my stomp build gave me a 1.19mil dps on a fight like mistress and my BM Frenzy build with gear chosen gave me 1.59mil dps, but i do like 400k dps more with stomp than frenzy...

To answer your question more directly:
There are two simulators out there: Ask Mr. Robot and SimC. Raidbots is just a web front-end for SimC that sets up sims for you. Both AMR and SimC have well-maintained Rotations(simc calls them APLs) that tell the simulator how to play the game for any given set of gear/talents. So, in that sense, your build is taken into account. I think that might have been what you were asking about?

As far as figuring out which combination of talents/gear is best for you, that is a complicated problem in Legion.

One way is to simulate the talent/gear combinations you are interested in.
Pros: you will know for sure, with very little error, which setup you tested does the most damage.
Cons: You have to figure out exactly what to test. How do you know that you tested the best setups? You need to have a really good idea beforehand of what might be the best combinations of talents, legendary items, set bonuses, trinkets, and stats. Simulating all relevant setups is not feasible, since you'd have to simulate for days or months. Testing just 2 items in each slot would take a month of simulation time.

What if, like this case, you just don't know already which talent builds are best? You could do a batch of simulations that compare different talent builds for your current gear. Both AMR and raidbots will set that up for you. The problem you run into here is that different talent builds will prefer different gear (stats/legendary items/set bonuses), so, doing that test against one set of gear is usually not sufficient. Really you will need to run that talent batch sim against many different sets of gear that might be viable.

This is the rabbit hole that is simulation. The combinations of talent/gear/relic/legendary/set bonus/trinket are so high now, that there is no way to actually test them all explicitly.

Raidbots helps you set up batch simulations for the SimC simulator. The AMR simulator page helps you set up batch sims for the AMR simulator.

People telling you that you have to use raidbots and not AMR are usually biased against our gear optimizer, and haven't even used our simulator or figured out how to set it up. If all you want to do is simulate some sets of gear, both websites do the same thing and you can use whichever one you like more. The advantage of AMR is that you don't have to wait in a queue or pay money for doing huge simulations - you can use your own computer to chug on them for free. Another advantage is that we model some real boss fights and the AMR simulator acts more like a highly skilled human than a robot, resulting in simulations that are much closer to real combat log parses.

The AMR gear optimizer is a totally separate tool that can help you pick optimal gear. You can pick a set of talents, and then the gear optimizer will rank items for you. We do this by doing millions of simulations ahead of time and then building a predictive model that can predict your simulated DPS for any gear setup instantly, completely bypassing the simulation step, which allows us to search the billions of gear combinations in a few seconds. We do this at the cost of a small margin of error on the result compared to explicitly simulating each and every set of gear you are interested in.

[Azortharion]

1. The AMR APL's for all Hunter specs are pretty far off the mark (the mark being actual human play, and ALSO correct play, it doesn't achieve either).
2. You can run SimCraft on your own PC too, and just use RaidBots to create the batches. Same thing. Yes, for free.

The models used for real boss fights are, at best, a bit closer to the actual fights than a Patchwerk sim of the same rough fight length. But most of the time, just really arbitrary and doesn't really get closer to being a useful "baseline" for sims compared to a Patchwerk sim.

Given how previous interactions have worked out for me and AMR, I won't pursue the discussion any further, and I won't write OP/whomever a novel about it, I am sure he or she is smart enough to work it out on their own. It's good to see the AMR PR machine is as efficient as always.

[Swol]

1. The AMR APL's for all Hunter specs are pretty far off the mark (the mark being actual human play, and ALSO correct play, it doesn't achieve either).
2. You can run SimCraft on your own PC too, and just use RaidBots to create the batches. Same thing. Yes, for free.

The models used for real boss fights are, at best, a bit closer to the actual fights than a Patchwerk sim of the same rough fight length. But most of the time, just really arbitrary and doesn't really get closer to being a useful "baseline" for sims compared to a Patchwerk sim.

Given how previous interactions have worked out for me and AMR, I won't pursue the discussion any further, and I won't write OP/whomever a novel about it, I am sure he or she is smart enough to work it out on their own. It's good to see the AMR PR machine is as efficient as always.

This isn't an AMR vs SimC thread, a person just asked a question and I answered. Good to see the Azortharion PR machine is as efficient as always!

The AMR Rotations are very good for all hunter specs. Rotations are always an evolving work in progress and I'd never say any rotation has zero room for tweaks, but I know for sure that none of the Hunter specs are "far off the mark". We test our rotations vs logs on a regular basis, and try to do a write-up near the beginning of raid tiers. The one for the start of ToS can be found here: http://blog.askmrrobot.com/simulations-match-real-logs/

The rotations have been tweaked and updated since then to follow trends as they develop.

I'm always happy to engage in theorycraft discussion if you would like to actually have one. I know from our past interactions that you much prefer to spread negativity instead of have real discussion, though.

[Azortharion]

I serve the wider community, you serve your employer, and that's fine, doesn't mean we can't talk like adults and not fling insults whenever an opportunity presents itself.

I had written up a little tidbit comparing the simulator to a parse of mine, but then I realized that the simulator had gotten 4 Trueshots in 328 seconds, with a fixed non-CDR Trueshot cooldown of 142 seconds.

Because for one reason or the other, AMR believes that Trueshot has a ~1m40s cooldown with no cooldown reduction equipped.



This is the gearset:



This is my gear in the actual log:


And here's a screenshot of me using Trueshot with this gearset equipped (bit large): https://puu.sh/xRLca/010e6e455a.png

Now anyone who has an MM Hunter can tell that there are no CDR items, set bonuses, trinkets, legendaries in this gearset, yet AMR seems to think so.

So, it got 1 more Trueshot than what was possible on this fight length, yet it still had:

4 less Aimed Shots (despite getting 2 more Lock and Load procs)
3 less Marked Shots
15 more Arcane Shots (and thus 500 wasted focus)
1 more Windburst

And overall, about ~5% less damage.

All of this despite getting 33% more Trueshot uptime than what it really should have gotten.

[Swol]

Looks like when I was refactoring code a couple days ago, I messed up that formula. Sorry about that, it's updated on live now.

As for the comparison - you didn't post any links for me to verify the setup or compare myself. So I found one of your recent Heroic Goroth kills. I know you raid mythic, but I wanted to use a heroic log since I know our boss script for that is really solid.

I am using this log, 1.63m DPS: https://www.warcraftlogs.com/reports...done&source=17
Here is the simulation report 1.68m DPS: http://www.askmrrobot.com/wow/simula...626daa5eaf3508

Here's a quick comparison.

• Aimed shot: 66 casts in the log, 64.4 in the sim. 192m total dmg vs 208m total dmg.
• Marked shot: 18 casts in the log, 17,9 in the sim. 28m dmg in the log vs 27.8m in the sim.
• Arcane shot: 67 casts in the log, 70.1 in the sim. 25m dmg in the log with a 43% crit rate vs 24.8 in the sim with a 38.5% crit rate.
• Windburst: 9 casts in the log vs 8.9 in the sim. 14.3m dmg in the log vs 14.6 in the sim. 33% crit vs 38%.
• A murder of crows: 3 casts in the log vs 3.7 in the sim. 10m dmg in the log vs 15m in the sim. 20% vs 38% crit rate.
• Auto shot: 89 casts in the log vs 91.6 in the sim. 8.36m dmg in the log vs 8.5 in the sim.
• Crucible traits: the usage and dmg match closely as well.

That matches fairly well, wouldn't you agree?

Maybe something was going on in the other report you were referencing. If you share a link to the log you used and the sim report I can compare those as well or see if something odd is happening.

Also, if you didn't do it before, make sure to manually select the crucible traits when you setup the simulation, since those aren't imported from WCL or the Armory.

[Azortharion]

Your AMR sim there has a 15% variation on the fight length which is a bit sketchy when you're comparing 1:1 with a log anyway..

The AMR sim also had 2 more Lock and Load procs, yet it had fewer Aimed Shots (when each Lock and Load proc is 2 free Aimeds), and with Mythic Goroth naturally having more mechanics to do than the Heroic one, we're looking at an almost 10% discrepancy in Aimed Shot count alone (which far trumps anything else in terms of importance, but yes, everything else looks almost completely similar).

Here is the report (log enabled): https://www.askmrrobot.com/wow/simul...555b7878fc83f4

Here is the Sisters parse I compared it to: https://www.warcraftlogs.com/reports...pe=damage-done
Here's a timestamped video of it, just to show that it was not what you'd call a target dummy parse: https://youtu.be/O1iYEBlGm6A?t=291

Here's a SimCraft run of that exact fight length: https://www.raidbots.com/reports/eTU...4ci/index.html

Myself and SimCraft are roughly equal in terms of output. AMR is 5% behind despite the additional Trueshot (surprising as they are closer on Goroth).

Why it's so close on Goroth but a huge discrepancy on Goroth HC vs my Sisters parse (where the Trueshot thing + the fact that Sisters is much more heavy movement are both big players "against" AMR), I am not sure, but as far as I can tell the reason SimC beats AMR more often than not is that it optimizes T20 2pc handling better, something my Hunter guide puts big emphasis on when teaching the rotation and one of the cornerstones of playing MM well. I don't know if it was intentionally programmed to be a bit worse than 100% optimal to be closer to "highly skilled human" than ""robot"".

Don't get me wrong, being within 5% of SimCraft which is one of the stronger modules is not bad at all, but when we're talking such fine margins between gear pieces already it's not something I could recommend on the basis that "well AMR is closer to what real humans do". SimCraft just has a better APL than AMR, and maybe "highly skilled human players" are somewhere in between, or closer to SimCraft, or closer to AMR depending on the colour of their socks and what they had for breakfast that day.

One thing I can say is that I don't think SimCraft, for all of its "chaos" with almost anyone being allowed to contribute or change things, has had such a gamebreaking issue as Trueshot having a 40s lower CD than it is supposed to for a few days. Partly because so many people use it that they'd notice pretty quickly, partly because the SimCraft module for Hunters has been stable for a long time now.

[Swol]

Hey, I'm out with friends for the evening - I will take a look tomorrow. Thanks for posting all of those links.

One quick note off the top of my head - just to make sure it was clear, the True Shot bug is fixed.

Lastly, here's a link to the sim I did earlier, but with almost no variance. I was rushing out the door and forgot to set that. My bad! http://www.askmrrobot.com/wow/simula...736fc58574cffb

[Azortharion]

Nah I know your recent sim had fixed it, I am talking about the Sisters-length 328s one.

[Mavick]

Please don't use AMR sims. They have quite often been found to be wrong by many in the community.

[EDIT] Original post was a bit over-the-top :>

[Swol]

Your AMR sim there has a 15% variation on the fight length which is a bit sketchy when you're comparing 1:1 with a log anyway..

The AMR sim also had 2 more Lock and Load procs, yet it had fewer Aimed Shots (when each Lock and Load proc is 2 free Aimeds), and with Mythic Goroth naturally having more mechanics to do than the Heroic one, we're looking at an almost 10% discrepancy in Aimed Shot count alone (which far trumps anything else in terms of importance, but yes, everything else looks almost completely similar).

Here is the report (log enabled): https://www.askmrrobot.com/wow/simul...555b7878fc83f4

Here is the Sisters parse I compared it to: https://www.warcraftlogs.com/reports...pe=damage-done
Here's a timestamped video of it, just to show that it was not what you'd call a target dummy parse: https://youtu.be/O1iYEBlGm6A?t=291

Here's a SimCraft run of that exact fight length: https://www.raidbots.com/reports/eTU...4ci/index.html

Myself and SimCraft are roughly equal in terms of output. AMR is 5% behind despite the additional Trueshot (surprising as they are closer on Goroth).

Why it's so close on Goroth but a huge discrepancy on Goroth HC vs my Sisters parse (where the Trueshot thing + the fact that Sisters is much more heavy movement are both big players "against" AMR), I am not sure, but as far as I can tell the reason SimC beats AMR more often than not is that it optimizes T20 2pc handling better, something my Hunter guide puts big emphasis on when teaching the rotation and one of the cornerstones of playing MM well. I don't know if it was intentionally programmed to be a bit worse than 100% optimal to be closer to "highly skilled human" than ""robot"".

Don't get me wrong, being within 5% of SimCraft which is one of the stronger modules is not bad at all, but when we're talking such fine margins between gear pieces already it's not something I could recommend on the basis that "well AMR is closer to what real humans do". SimCraft just has a better APL than AMR, and maybe "highly skilled human players" are somewhere in between, or closer to SimCraft, or closer to AMR depending on the colour of their socks and what they had for breakfast that day.

You need to set up an apples to apples comparison if you want to compare the two simulators. You can't compare a patchwerk fight in simc to a goroth fight in AMR and expect the ability use to be the same.

Here is an equivalent simulation from AMR to the log you posted, using a "patchwerk" style:
https://www.askmrrobot.com/wow/simul...9ad04e3633ace8

Also, when I was testing these examples with the nightly simc, I noticed a different result than what you posted. I copy and pasted the exact config file from your previous link and ran it against the nightly simc build and get this result:
https://www.raidbots.com/reports/8q7...tCP/index.html

Using the SimC APL in the AMR simulator, this is the result:
https://www.askmrrobot.com/wow/simul...90b2c623432cc8

So, using the exact same APL, we see 1740k DPS in simc vs 1686k DPS in AMR. The differences can be explained by the following:
1.) Tarnished Sentinel Medallion use. The SimC report doesn't explicitly show how this is being used in the APL section. To find that, you have to go into the code. I updated the AMR version of the APL to hold the last use until the end of the fight and align it with Trueshot:
https://www.askmrrobot.com/wow/simul...62d82b7151cdc1

Now the gap is much smaller (1740k DPS vs 1712k DPS). This takes me to the second reason for differences:
2.)SimC is using a couple more abilities. This is a general theme when comparing the two simulators, and this is what I'm talking about when I say SimC is more like a Robot and AMR is more like a human. SimC does things that actually aren't possible for a player to do in-game. Very small things, but they add up incrementally to account for about a 1% difference in DPS between the two simulators, on average.

The main culprit is reacting to the current state of Power(Focus) in the simulation. If you are deciding between using an Aimed Shot (50 focus) or falling through to a different action, and your Focus is right around 50 focus... it is not possible in-game to cast Aimed Shot until you have that 50 focus. So, you would need to have 50 focus a couple hundred milliseconds before your current cast/GCD is finished in order to queue up that Aimed Shot. Otherwise you'd have to delay and incur some reaction time lag, or use something else instead. AMR models this. SimC uses a much more basic implementation of decision logic and does not take this into account. SimC will check how much power you have at exactly the time your current GCD/Cast is finished and let you act immediately. This causes the "robotic" behavior that I talk about in resource-limited specs like rogues, hunters, feral, etc. And, this is why AMR simulations match logs a bit better than SimC, in addition to more robust boss scripting. This effect is a bit more amplified for MM, which is attempting to play around with fitting max aimed shots into vulnerable windows.

We can see some of this difference go away in AMR if we turn off the anticipation logic, but we can't turn off ALL the logic AMR has:
https://www.askmrrobot.com/wow/simul...971958359c8ad1 (1723k DPS)

As you see, given the same input, both simulators will give us very similar and actionable outputs with only very minor differences.

Now lets take a look at the differences between the old AMR default rotation and the SimC APL. The main differences are that the AMR default wasn't getting quite as much use out of the 2T20 bonus and didn't have a custom condition for Tarnished Sentinel Medallion. Lets first edit the old rotation to account for the medallion:
https://www.askmrrobot.com/wow/simul...122401f4e91efa

Now we are at 1712k DPS vs 1673k DPS, a difference of 2.3%.

So, the added logic to game the vulnerable window and 2T20 bonus in the SimC APL is worth about a 2-2.5% DPS gain. I took a look at the SimC APL and worked out how it was going about making more use of the set bonus. There is a variable in there that tries to guess how many Aimed Shots can fit in the current vulnerable window. With that piece of information, the APL can then insert a smart number of Arcane Shots early on in the vulnerable buff.

I noticed that the formula being used doesn't correctly calculate the number of Aimed Shots that could be cast, though, since it doesn't take into account the Focus you will gain by using the Arcane Shots. It only looks at passive regen. I was able to modify the AMR default APL to also use a variable to track this, but I updated the formula to also guess how many arcane shots you would use:
https://www.askmrrobot.com/wow/simul...273c66550dad8c

So, now we see about a 1-1.5% DPS increase in the AMR rotation compared to the SimC APL! (1741k DPS vs 1712k DPS)

Now that the AMR simulator is doing more damage than SimC with a better rotation/APL, are you going to tell everyone to use AMR instead of SimC?

[Azortharion]

I meant that for the purpose of comparing an AMR log to my log, you didn't match up the fight lengths and so the comparison seemed meaningless.

Good job on the updates, definitely brings AMR closer to optimal play. I don't think I'll "tell everyone to use AMR instead of SimC", because your improvements to the AMR APL are about half a raid-tier late (still, better late than never), and up until this point, SimC had the better "prediction" logic for how to lay out each Vulnerable window, and I don't imagine that trend changing (AMR being behind in terms of raw update speed) for Antorus, and the strange Trueshot thing raises questionmarks too, because given that most of AMR's users are on the casual side and don't enable the Log and spot the Trueshot thing, how long would that error have lasted if I didn't have this argument with you? ;p