Wudijo 2.2 detailed guide to Demon Hunter

Wudijo created a very detailed guide to 2.2 dh.

Really amazing work that worths spreading.

Here is the full thread:
"[Introduction]

So you might remember my post back in the beginning of 2.1.2 where I outlined why the physical element is superior to cold for the M6 CA build, and which also helped to spark the whole Taeguk + Hexing Pants meta. Back then I used a spreadsheet that I had created to math out the differences between each build and compare different item and skill choices. Now, I have updated it for 2.2.0 and polished it, so that it is in a state that can give approximate results for almost any situations and rift layout, including runtime, buffs, cooldowns, number of enemies, number of elite packs, area damage procs etc., and I also decided to release it for public use this time.

I have also uploaded a video on my Youtube channel where I go over the how the spreadsheet works and how I concluded my results here if you want to check that out. Obviously it's a little bit difficult to model everything in a spreadsheet like this, and not all of the numbers are exactly accurate. Other than that, just reading this post should be more or less enough to give you an understanding of relative strength of the builds and item preferences.
When it comes to the actual damage values in the spreadsheet, I have tried to model fights versus different amounts of monsters and the fact that you will not hit everything all the time. This was done by assigning a share of your total time spent in the rift to each monster group size and then calculating approximate values for total damage done per skill use for every increment. I feel that the numbers I used are a little bit too generous towards large groups for your average rift but should be more or less adequate for very dense dream rifts (remember we're talking about "optimization" here). Also, I divided the sheet dps by 1 million to make the numbers more readable. No matter if my "rift calculation" is correct or not, you will still have the exact same stat weighs (other than area damage obviously) because those are not influenced by it.

General assumptions for the spreadsheet are a high-end character with the following stats:
Paragon 800
Rank 70 gems
11000 dexterity
Max rolls on everything
Dex, CC, CHD, socket on Hellfire and rings
Full glass cannon
I have also run the numbers on a lower geared character, but the relative strength of the builds / item choices won't really change if you just have lower dex / gems / weapon damage etc, just the total damage done decreases. If you have really crappy items (e.g. you miss 2-3 full RCR or CC rolls and have vitality instead), the missing stats will obviously become more desirable. For every other setup the following numbers are what you can expect to see in the game as well.

My general conclusion is that this time around, there are many "if's" because the numbers are fairly close, and a lot will depend on your rift composition. To give you a couple of examples, getting a power pylon will decrease the benefit of Calamity and Marked for Death but favors Balefire and crossbows. Similarly, a channeling pylon will favor the faster weapons. For Windforce, you want to have rifts with big monsters that can be knockbacked, for Leonine you want to have very dense rifts where you can get insane area damage procs.

[Convention of Elements + Hexing Pants vs. Focus + Restraint]

It's pretty clear that Focus + Restraint win by a large margin. For the record, other combinations than CoE + Mr Yan / Pride's / Cindercoat are not even worth considering if you don't want to use F+R (yes, Manald Heal is shit for UE). In the updated version of the spreadsheet, I added Convention of Elements (assuming a flat multiplicative 50% damage increase, yes you can time it, but you will also have a lot of wasted procs because you might just kill stuff in 1-2 seconds with nothing left on the screen) instead of a Stone of Jordan, which makes double Unity even worse (20-30% less than CoE). Depending on setup, F+R are about 25-40% stronger than any other combinations (50-80% more than double Unity + Hexing Pants). Since we have less resources to work with, F+R favor fast weapons with high dps, which makes crossbows seem to be about 5% worse than Bows and 10% worse than Balefire / Calamity if you have equally rolled weapons (meaning relative to the maximum possible roll, a 3000dps crossbow will still be more or less on par with a 3000dps Calamity). Because we don't use Pride's Fall / Hexing Pants anymore, we will have to regenerate a lot more, thus putting more emphasis on attack speed rather than passive reg, companion use or globes.

[Slow weapons vs. fast weapons]

So back in my old post I was talking about how strong bows and Calamity seem to be, however in practice they have never turned out to be better (or were just about even) and everyone still went with crossbows (also due to easy availability of near perfect rolls by crafting Arcane Barbs). Well, first of all I found and corrected something in the spreadsheet that was favoring faster weapons by a few %, but most importantly it comes down to how effectively you can spend your passively generated hatred. Imagine the following scenario: You don't have any hatred regen, no Templar, no health globes and no skills that regenerate hatred; what's left is the hatred you regenerate from shooting generators and nothing else. Let's say you need to cast two generators in order to cast one spender, so you can cycle 2 - 1 - 2 - 1 - 2 - 1 indefinitely. In this scenario, your attack speed will not have any influence whatsoever on the cycle, and the only metric that will define your damage is weapon DPS. Since this is not a realistic scenario, it is all about how you can spend the leftover hatred from all the things I have excluded before. The higher the share of this passive regen, the better the slower weapons become (since they have better damage per hatred spent ratios). You can see an example of this in the video.

[M6 Cold vs. Phys]

For those of you who still want to run M6 for whatever reason, you should know that you are deliberately lowering your potential by 1-2 tiers, it just can't keep up with the other builds, and even for nonseason there is a stronger UE non-lightning build. Still, it has some better defensive value, mostly because your companions help you here and there and because you can kite enemies offscreen a lot easier if you can watch your sentries shooting at them from a distance, plus reflect is easier to deal with or you can hide behind a corner and still deal damage in dangerous situations.
Iceblink brings the cold Cluster build a lot closer to the physical build, to the point where they seem more or less the same. Cold has a higher luck range if you get only few and easy elites and Stonesinger, while physical should be more consistent. I guess that on 56+ the rift guardian will be a huge problem for cold, so physical might pull ahead regardless, also keep in mind that pylons used on the RG will be a lot more effective with physical. Enforcer, Gogok and BotP are still crap (7-12% worse) and if you want to play M6 on nonseason you should really stick to the Taeguk. With Iceblink, you should always use Spitfire Turret.

The best passives to use are Awareness, Ballistics, Custom Engineering, Cull the Weak, Night Stalker (5th). I strictly recommend using a Hellfire Amulet with this set because all of the passives are very strong and you don't want to miss any. Other options for the 5th passive are (in that order): Steady Aim, Single Out, Ambush, Archery.

For groups, Cluster Bombs might still have a spot if you have someone who can spawn many globes for you, but it will probably not outperform either UE fire or lightning.

My current estimate is that M6 (Cold or Phys) will clear tier 56-59 solo.

[Unhallowed Essence Fire]

All of the special bows (Windforce, Leonine, Odyssey) seem to be about on par with Calamity / Balefire and they all bring their own strengths. For Windforce, using Steady Strikers or Strongarm should be about equally strong. Leonine can shine in dense rifts with lots of slow monsters, however you probably want to use sentries to reliably proc the slow. Odyssey's End might shine at the top end (56+) because it can speed up your RG kill enormously, but will be a lot more difficult to use during the rift (since you want to use sentries for slow and use Mortal Enemy). Cluckeye only has a marginal damage bonus, but the added benefit from the stun is also nice to have.

Other than that, the 1h Crossbows seem to be the strongest options raw dmg wise. Given equal rolls, Calamity and Balefire are more or less the same, if you have fire% on your amulet Calamity wins out by a few %, also keep in mind that in groups you will buff others with Calamity as well. Also, a somewhat hidden champion might be Kridershot with Immolation Arrow + Mortal Enemy + Ess of Johan. It's behind numbers wise, but I feel that it is underrated in the spreadsheet and it could turn out to be insane if you can play it well.

In addition to Evasive Fire - Focus, Multishot - Arsenal, Preparation - Invigoration and Vault - Tumble, it seems preferable to use Companion - Bat + Vengeance - Seethe for crossbows, Bat + Marked for Death - Mortal Enemy for Bows and Caltrops - Bait the Trap + Mortal Enemy for 1h crossbows. All of the choices are pretty close (within 1-5%, except Seethe for fast weapons), so you can experiment a little. If you want to run fire in groups, Bait the Trap and Mortal Enemy should be the best for every weapon type.

For nonseason, Chakram with Spines of Seething Hatred presents a viable alternative (1-2 tiers lower than DML, and about the same level as the strongest M6 setups). I have only done the numbers for the fire Chakram, but the cold one should be more or less the same (plus you can use Iceblink instead of Polar Station). For crossbows, you can also use Chakram - Shuriken Cloud and macro it to cast automatically all the time. I have tried it manually and managed to cast it about 100 times per minute for an extra 400 hatred, in a full rift that's 6000 hatred and is more or less the same In general, for the generator-heavy builds (Chakram / Elemental Arrow), dropping all RCR in favor of better damage stats (IAS, Elite, AD) is preferred (up to 2-3% more dps per item). Chakram specifically profits a lot from the higher attack speed of one-handed crossbows, which makes it the only case where you prefer a regular 1h over a crossbow.

As for passives, the best to use are: Awareness, Cull the Weak, Ballistics, Night Stalker, Steady Aim. Ambush is also really close (about 2% behind with a 60% DML affix) and might become better because you will deal slightly more damage in reality, especially when enemies don't have much life, and you will also profit a lot more from power pylons. At the high end, it might be preferable to use Single Out because you will have to spend a lot of time on the RG.

My current estimate is that UE Fire will clear tier 57-60 solo.

[Unhallowed Essence Lightning]

It comes to no surprise that the UE Lightning build with Kridershot is a clear winner dps-wise. Due to the nature of the Ball Lightning skill it is very hard to model so the numbers are just a very rough estimation and don't take into account different monster sizes etc. I have done the numbers for a Hellfire Amulet and an Ess of Johan with equal rolls, funny enough they are basically exactly the same (0.1% difference). Since there are so many variables to it I didn't try to exactly reproduce its damage capabilities, but the numbers seem about right compared to M6 and UE fire (approximately 60-100% stronger). I have a feeling that the Ess of Johan is vastly underrated in the spreadsheet and will eventually pull ahead because you have a higher luck range and can get insane area damage procs if you manage to pull a lot of monsters. Also, this build is likely going to be one where you can see a really noticeable difference between season and nonseason because of the Iceblink gem (should be approx. 5-15% stronger). The increased potency of the slow is going to yield a lot of damage (which I calculated very conservatively) because of more ticks (on top of easier kiting), frees up another skill slot and requires you to cast a MS only once every 3.5 sec (Iceblink duration) instead of 3 sec (Taeguk duration). With Taeguk, you should use sentries to proc the slow (% chance to slow secondary on pants helps) and add Marked for Death - Contagion to it. With Iceblink, I think Rain of Vengeance - Flying Strike could be a good choice (especially with Ess of Johan), otherwise you can push the limits with Caltrops - Bait the Trap. Be aware that actually finishing a rift in time will require tremendous amounts of RNG. You will need to fish a lot of rifts to actually find one where all the stars align (good monster types, easy packs, good RG, good pylons). Since Blizzard has removed basically all of the defensive abilities this build relied on in the past, M6 and UE fire are a lot more consistent.

In high-end groups, a lot of the time will be spent on the rift guardian. Iceblink will have no more additional value because of a better slow, so what is left is solely the 10% crit chance which is a little bit underwhelming by itself. I suggest to use BotP, since you generally have a higher incentive to kill elites in group runs and will get more procs. At the high end you will spend a lot of the time on the RG, which is why you have to maximize your damage there, which makes Single Out pretty much mandatory. The 15% elite bonus from BotP is very close to a high Taeguk even without the proc. Also, you should not have higher than 55% sheet crit chance if you run with a Crusader because you will cap out at 100% (20% Judgment, 25% Single Out, 10% Bait the Trap). My numbers suggest that a high-end DH will deal approximately 15-25 billion dps on a typical RG.

The best passives to use are Awareness, Cull the Weak, Steady Aim, Single Out, Ambush (5th).

My current estimate is that UE Lightning will clear tier 60-62 solo.

[Natalya]

Unfortunately I couldn't include the Natalya set in the spreadsheet, the most important reason being that we don't even have a finalized build for it yet. In the last couple of days people have made huge progress (up to tier 55), so there are some good builds already, however I think there is still a lot to discover and that some of the set's potential is still unused. I'm very excited to try it out myself once I can start pushing solo, I have lots of ideas that want to be tested. For now, it's best to just keep all of your good Natalya pieces and anything that might go with it.

However, I can still state what there is to know about Natalya currently: F+R seem preferable. Strafe + a generator is very good to reduce the cooldown of Rain of Vengeance, you can hold down your force attack button, keep strafing and hit your generator button manually whenever you can. Cooldown is a very strong stat, a lot stronger than e.g. RCR for other builds. N6 works for any damage including gems and conduit pylon, both N4 and N6 are multiplicative bonuses and also work for your Crashing Rain belt. The belt summons a crashing beast (very short visual effect) with a small AoE effect that looks similar to one of the Stampede rune at the position of your cursor and uses the same element as your selected RoV rune. Both Crashing Rain and RoV (and Fan of Knives) use only your main hand (left weapon slot), so you don't need %damage (or any damage roll at all if you don't care about generator / Strafe damage) on your offhand.

My current estimate is that Natalya will clear tier 57-62 solo.

[Reroll preferences]

Here is some general information about what you look for on your items (if it hasn't been specified before already).
Generally, you get RCR on every item you can, excluding jewelry, and add area damage on shoulders.
On rings, Avg Dmg instead of CC or dexterity is about 4-5% weaker on crossbows and up to 1-4% better on 1h crossbows, and more or less the same for bows. Keep in mind that in parties you will probably cap crit chance on the RG with Single Out, Bait the Trap and Judgment. RCR is also an option for crossbows (1-2% weaker).
For the UE builds you can basically drop any CDR because it is almost useless. Only take it if you want to go full glass cannon (and even then having some more toughness might be helpful for Plagued and Desecrator). Removing all of the CDR (gem, paragon, shoulders) is a 1-2% dmg loss with Bat Companion, 3-4% if you use both Bat Companion and Seethe. I recommend to drop it completely if you use MfD and Caltrops.
Elemental damage is better than dexterity or CC on amulet, up to 5-10%. Probably even a 15% roll is preferable. Ancient dex value might be close.
IAS instead of RCR is an option for 1h crossbows (1-2% weaker) and preferred for Kridershot and Spines of Seething Hatred.
Dropping Multishot damage on quiver for hatred / elite / area damage is a 3-6% damage loss. Dropping RCR for hatred is a 0.5-2% damage loss (keep in mind you can vault more with RCR).
Steady Strikers / Lacuni Prowlers can be better than Reaper's Wraps if you don't get many globes.
For UE lightning, I recommend elite bonus as your last primary, especially for groups. Area damage is fine as well.

[FAQ]

Q: Dafaq is this sh*t? You can't math everything out.
A: Take it or leave it.

Q: What are additive and multiplicative damage bonuses, and how do they work?
A: Additive means that stacking more of the same damage bonus (e.g. "Damage Increased By Skills" in the extended sheet tab) will have diminishing returns. If you have 0% bonus and add 20% to that, you will deal 20% more damage. If you add another 20%, you will only deal 140/120 = 16,67% more damage relative to that. There are different damage categories that are additive within itself but multiplicative as a whole. For DHs, the relevant calculations go like this (exclude those that you are not using):

((Weapon Min Dmg + Avg Min Dmg) + (((Weapon Max Dmg + Avg Max Dmg) – (Weapon Min Dmg + Avg Min Dmg)) / 2)) [Average Weapon Damage] x ((100 + Dexterity) / 100) x ((CC / 100) x ((100 + CHD) / 100) + (1 – CC / 100)) x (1 + Elemental Bonus / 100) x ((Steady Aim (20) + Skill bonuses (generally 45) + Taeguk (((10 + (Rank / 2)) / 100) x expected uptime) + Calamity (20) + Hexing Pants (25) + Strongarms (20-30) + Bane of the Powerful Proc (20 x expected uptime) + Marked for Death (15 or 20) + Archery [Bow] ( + Power Pylon (300) + almost any party buff) / 100) [Visible and hidden "Damage Increased By Skills"] x Cull the Weak (1,2) x Ambush ((1+(1/((0,25/140)+(0,75/100))-100)/100)) x (1 + Bane of the Trapped / 100) x (1 + Zei's / 100) x (1 + (Elite Damage on gear + Bane of the Powerful Passive) / 100) [If fighting Elites] x Focus (1,5) x Restraint (1,5) x (1 + Convention of Elements Bonus / 400) x (Skill Weapon Damage / 100) = Average Damage per Hit

Bait the Trap, Iceblink, Archery, Sharpshooter, Single Out, Judgment are just added to the CC / CHD numbers. CC can exceed the 75% cap with those skills and vanilla SoJ bonuses.

For skills that are not your main spender, you have to divide by (1 + your full "Damage Increased By Skills"), subtract (Skill bonuses / 100), and then multiply it by the remainder.

For skills that are not your main element, you have to divide by (1 + Elemental Bonus / 100).

Attack speed is Weapon Speed x (IAS on Weapon / 100) x ((Combined IAS bonuses from gear and paragon + Gogok + Law + Voodoo) / 100).
For M6, add "x 6" for personal damage, and "x 5 x (1 + Sentry Bonus / 100) / (1 + Elemental Bonuses / 100) x (1 + (Elemental Bonuses x Enforcer Bonus) / 100)" for sentries.

For UE, add "x (1 + (current discipline x 15 / 100)) x (1 + 20 / 100)", for Multishot, add "x (1 / ((1 – DML Affix / 100) + (DML Affix / 200))" to Multishot damage, "x 2" to the rockets of Arsenal, for Ball Lightning, add "x (100 / Meticulous Affix)" to Elemental Arrow damage.

For N6, add "x 5", for RoV, add "x 2"

Q: For UE, is it ever worth to reroll anything other than +max discipline if the item doesn't have it?
A: No.

Q: I have a 3200 dps Bow and a 3100 dps Crossbow with almost equal stats, which should I use?
A: The crossbow. Generally, your Bow has to have about 5% more dps to break even, Calamity and Balefire are pretty close if they only have about the same dps as your crossbow. To be safe, I'd use them if they have 50-100 dps more.

Q: What is the Iceblink duration, and how does it work?
A: Duration is 3.5 seconds and it works multiplicatively (60% slow x (1 + Iceblink Bonus / 100)) for up to 75% slow on rank 50. OP stuff. Purely damage-wise, it is a few % behind Taeguk, however this doesn't include the fact that you don't have to waste some hatred here and there to keep up the buff and that it will be a lot easier to pull more monsters, so I added a flat 5% damage increase on top of the 10% CC and this made Iceblink come out on top.

Q: Which gems should I use?
A: Bane of the Trapped and Zei's cannot be debated. For solo, add Iceblink (season) or Taeguk (nonseason). As a rule of thumb, you can drop Taeguk once per elite pack, otherwise use Bane of the Powerful. For groups, adding Bane of the Powerful is generally the best choice. For T6, Boon of the Hoarder will outrun Bane of the Trapped. For speed runs, Bane of the Powerful is the best third gem.

Q: Why Vault over Smoke Screen?
A: Smoke Screen was crap before and now has become even crappier. It's easier to use for most affixes, but you have very little mobility and can neither skip enemies nor rush ahead and pull more, effectively slowing down your progression. Especially with the Unhallowed Essence set we will be skipping a lot more than with M6 because we have so much more discipline to spend, which further favors Vault. In addition to that, given correct timing, you can evade every attack in the game save for Vortex (about half of the time) and Molten (ground effect) if you don't use Tactical Advantage.

Q: How will I get high keys for solo runs?
A: Last season you could get up to tier 52 keys in solo trials, I guess now it should be up to 56-58 (but it will be very difficult and inefficient). If you have trouble with that, just join a community (e.g. "High Level Trials" on EU).

[TL;DW] - The spreadsheet is only a spreadsheet and no dynamic simulation after all, so results are a little bit skewed. I did my best to eliminate any errors and tried out many different combinations to find out what is best. The number displayed at the end is an estimate of the total damage done, not dps. If you want to use the spreadsheet for your own character, never edit any of the greyed out numbers, and keep in mind that the standard settings are for solo GRs. Many things are automatically included, however some things are only rough approximations (most importantly area damage) and don't take breakpoints or pylons into account. If you want to model specific situations (boss fights, longer / shorter rifts, group buffs etc.), you should edit the parameters accordingly and just manually add the buffs to your stats.

[TL;DR] - UE lightning > UE fire > M6 (Natalya probably somewhere in between) with perfect items and RNG. Focus & Restraint cannot be beaten by anything. Given equal rolls, new meta favors faster weapons because a higher share of your total hatred is produced by your generator casts. In solo rifts, Iceblink is the best third gem to use along with BotT and Zei's, for nonseason Taeguk can produce more or less similar results but is harder to use (for UE lightning, Iceblink is strictly better and a Ess of Johan will likely outperform a Hellfire). For UE fire, your weapon preferences are Calamity > Balefire > Kridershot = Windforce = Leonine = Odyssey's = Cluckeye > Regular Bow > Regular Crossbow. The 1h crossbows and special bows all are very close numbers-wise and you will see difference mostly depending on how well you can use their affixes to your advantage. Kridershot + Ess of Johan + Immolation Arrow might be underrated in the spreadsheet and could turn out to be the strongest UE fire setup. For M6, always use spitfire turret if you have Iceblink. The new meta has brought cold and physical a lot closer together, Etrayu has become the go to weapon for cold, however RG will be a huge issue in high GRs so physical might still win out at the top end (with a Calamity). Natalya might be added to the spreadsheet in the future, I left it out because there is still no finalized build yet.

Feel free to add anything I might have missed in the comments, or feedback me about my spreadsheet.
Cheers, wudijo"