Doppelgangers Proc at Lower Rate While Engaged in Group and Raid Content

As a part of the Chaos Magic weapon specialty, Controlled Chaos, the player has a 33% chance to generate Extradimensional Doppelgangers once one has accumulated eight Paradoxes. Conducting extensive tests on the test dummies in Agartha has shown this to be true. And yet, further testing done in practice shows Doppelgangers are not being generated at the rate that I either calculate them to be or what repeated parsing has shown it to be—and I don’t know why. Regardless of the reason, it significantly reduces the outgoing damage capabilities of Chaos Magic, since it is from Doppelgangers that such a player must rely upon to deal the majority of one’s damage.

The simplest way to demonstrate that there is a problem is by comparing two series of long parses, one that I had done on an Agarthan Dummy and another on the Bunica Paduruii boss fought in Deep Agartha. First, here is the build I used in both tests:


  • Entropy

  • Tumultuous Whisper

  • Pandemonium

  • Breakdown

  • Deconstruct

  • Savagery


  • Dissolution

  • Blessing of Octed

  • Body Double

  • Butterfly Effect

  • Resonance Cascade


  • Warped Visage of Havoc Mk II

  • Bladed Gauntlets of Havoc Mk III


  • Effigy O35 Turret

Waist Talisman

  • Tumultuous Sash

The build is designed to deal optimal single-target Chaos Magic DPS by maximizing my time spent using a Chaos Focus in order to generate as many Doppelganger procs as possible. Tumultuous Whisper, three of my passives, and Tumultuous Sash all work to generate Paradoxes much faster than the base rate, and my Chaos Focus, Warped Visage, generates one additional Doppelganger every time I had reached eight Paradoxes. I begin combat by activating Entropy, Tumultuous Whisper (which triggers Dissolution immediately), Savagery, and Pandemonium, and then proceed to expend my remaining Energy on Breakdown until Dissolution expires, making sure that my Energy never reaches or exceeds 15 before my Elite and Special abilities are off cooldown again to repeat the rotation.

And now, for my test results:

Bunica Padurii Results

  • Time: 1,616 seconds

  • Cacophony Hits: 125

  • Dissipation Hits: 640

  • Dissonance Hits: 112

  • Total Doppelganger Procs: 125 + (640 / 5) + 112 = 365

  • Doppelganger Procs Per Minute: 60 / (1616 / 365) = 13.552

Note: Because Cacophony deals damage in a range around its target, and Bunica Padurii periodically spawns forest spirits, I had to remove 20 hits from my log to arrive at the precise number of Cacophony procs.

Agarthan Dummy Results

  • Time: 1,526 seconds

  • Cacophony Hits: 187

  • Dissipation Hits: 860

  • Dissonance Hits: 178

  • Total Doppelganger Procs: 187 + (860 / 5) + 178 = 537

  • Doppelganger Procs Per Minute: 60 / (1526 / 537) = 21.114

The stats are clear: Doppelgangers procced an average of 35.82% fewer times per minute in the Bunica Padurii parses than in the Agarthan Dummy parses. I witnessed this same phenomenon countless times fighting Hel in Deep Agartha during the Winter event, as well as fighting Regional bosses with 9 – 12 other players–and I suspect this is true also in Elite dungeons and raids.

It doesn’t make much sense for high or inconsistent latency to be a factor. In the Bunica Padurii parses, Deconstruct hit every 1.5 seconds, Breakdown every 5.15 seconds, and Pandemonium every 20.46 seconds; whereas, for the Agarthan Dummy parses, Deconstruct hit every 1.46 seconds, Breakdown every 5.04 seconds, and Pandemonium every 20.35 seconds. Using the average hit rates of the Agarthan Dummy parses as a baseline, I might have gotten an extra 30 Deconstruct casts, an extra 8 Breakdowns, and no extra Pandemonium with the same millisecond delay in the Bunica Padurii parses, or, put another way, one extra cast every 40.16 seconds, which in no way explains 35.82% fewer Doppelganger procs.

I don’t know what could be the cause, or if there are perhaps multiple causes. But it cannot be denied that Doppelgangers aren’t operating in the same way when I am alone parsing against a test dummy and when engaging an enemy with a group of other people. Is it a difficulty in properly generating hits divisible by eight in large-scale fights? Or might Doppelgangers have trouble finding room to spawn in areas crowded with a large boss and many players? Or could Doppelgangers fail to render as a means to reduce latency, thereby never dealing the damage they are supposed to? I wish I could tell you, but I haven’t a clue.


I think i replied to the original post on reddit as well. Without having a combat log to support it, it seems to me that there exsist “optimal conditions” depending on your characters current gearing, when it comes to generating paradoxes. While my chaos alt was going through story mode dungeons, i noticed that she generated paradoxes a lot more in DW than in Ankh or HR. Later on she generated more paradoxes in Ankh in comparison to DW and HF. Did you save any data on the other chaos effects? It seems relevant to check if it is just the clones, or if it is the paradox gen.

It can be difficult to gather empiric evidence for Paradox-generated events, as it requires a sufficient sample size to rule out a natural deviation from the mean and a way to reproduce experiments without conditions that could skew results. How else can we distinguish between what is broken and what is bad luck?–or what are abnormal or optimal conditions? There’s no easy way to gather data on Singularities and Chaotic Enigma. But, if Doppelgangers fail to generate at the rate they are expected to, it’s possible Singularities and Chaotic Enigmas are also affected.

True, while Singularities do damage, they can very easily miss the target, and enigmas wouldn’t show in ACT i suppose, although maybe in the in game combat log? I noticed it because i run a build that force at least one clone to spawn every time i hit 8 paradoxes. When i felt like i wasn’t seeing as many clones as usual, i started watching my paradoxes. Sometimes i generate them fast enough to bring out a second group, while the previous one is still going. Other times i keep hitting and hitting, but no paradoxes generate for a long time. I dunno if this could be used to somehow measure it?

You can track your own Chaotic Enigmas by using the Advanced Combat Tracker, right-clicking on a parse, selecting “View Log,” and inputting “You gain buff” followed by the buff’s name in the search box. But this won’t track the Wellness Enigma that procs while you are at full health, and those Wellness Enigmas that do heal you may be logged under a different name. Alternatively, you can do without Resonance Cascade and track all Enigmas that proc on your target.

I’m working on a method to determine whether Chaotic Enigmas are also affected in the same way Doppelgangers are in group and raid content. I’ve conducted parses on a test dummy to measure the number of times each Enigma has procced and their up-time. But now I must wait to gather data on mega-bosses. This could take some time.

More data points is always great. If my feeling of slower paradox generation is right, and it depend on the stats of the opponent i am fighting, i suppose it should apply on different test dummies? I’m not entirely sure if you tested on different dummies, or if you did several tests on the same type of dummy?

All of my parsing on test dummies has been on Anima - Sim: Dungeon. I tested your hypothesis that targets with higher stats might effect the generation of Paradoxes by parsing for 25 minutes on the Anima - Sim: Nightmare dummy. Here are the results:

Anima - Sim: Nightmare Parse

  • Time: 1,508 seconds
  • Cacophony Hits: 168
  • Dissipation Hits: 815
  • Dissonance Hits: 154
  • Total Doppelganger Procs: 168 + (815 / 5) + 154 = 485
  • Doppelganger Procs Per Minute: 60 / (1508 / 485) = 19.3

The Doppelganger procs per minute in this sample are 8.59% lower than those I had done on Anima - Sim: Dungeon. We can perhaps come closer to a conclusion on the significance of this discrepancy by comparing the rate in which a Doppelganger was generated for every Chaos Magic ability activation, and then looking to see if these rates fall within the standard deviation of either of these experiments with a confidence interval of 95%.

Anima - Sim: Dungeon

  • Ability Activations: 1,498
  • Total Doppelganger Procs: 537
  • Doppelganger Proc Rate: 537 / 1498 = 35.85%
  • Standard Deviation: 1.96*\sqrt(0.3585 * 0.6415 / 1498) = ±2.43%
  • Standard Deviation Range: 33.42% - 38.28%

Anima - Sim: Nightmare

  • Ability Activations: 1,473
  • Total Doppelganger Procs: 485
  • Doppelganger Proc Rate: 32.93%
  • Standard Deviation: 1.96*\sqrt(0.3293 * 0.6707 / 1473) = ±2.4%
  • Standard Deviation Range: 30.53% - 35.33%

What can be inferred from the standard deviations is that, if I were to repeat these experiments, 95% of the time the results would fall within 33.42% - 38.28% for the Anima - Sim Dungeon parse and 30.53% - 35.33% for the Anima - Sim: Nightmare parse, making it possible that, because their ranges intersect at 33.42% - 35.33%, there’s not a statistical difference to be found. Whatever we can make of the Nightmare parse data, we cannot say that it mirrors the Bunica Padurii parse, which had 1,548 ability activations and 365 Doppelganger procs, giving Doppelgangers a proc rate of 23.58%. There’s just no reason to believe a target’s higher stats have an effect on Paradox generation based on data I’ve collected.

It has been asked whether Chaotic Enigmas are similarly affected as Doppelgangers are in group and raid content. I’ve done parsing using the build I have noted above, except this time, I replaced the Warped Visage Chaos Focus with an Enigmatic Apparatus and the Body Double passive for Long Term Chaos, to ensure I generate as many Chaotic Enigmas as I can for a long as possible. My first series of parsing began with an Agarthan test dummy:

Anima - Sim: Dungeon

  • Total Combat Time: 1,517 seconds
  • Potency Enigma Procs: 39
  • Potency Up-Time: 257 seconds (16.94%)
  • Avoidance Enigma Procs: 37
  • Avoidance Up-Time: 241 seconds (15.89%)
  • Reciprocity Enigmas: 47
  • Reciprocity Up-Time: 320 seconds (21.09%)
  • Average Number of Procs: (39 + 37 + 47) / 3 = 41
  • Enigma Procs per Minute: 60 / (1517 / 123) = 4.865
  • Average Up-Time Per Enigma: (257 + 241 + 320) / 3 = 272.67 seconds (17.97%)

And now I conducted the second series of parsing on Kenememti the Fear Eater and Bunica Padurii until I have accumulated about as much combat time as on the test dummy:

Kenememti and Bunica

  • Total Combat Time: 1,625 seconds
  • Potency Enigma Procs: 29
  • Potency Up-Time: 190 seconds (11.69%)
  • Avoidance Enigma Procs: 30
  • Avoidance Up-Time: 209 seconds (12.86%)
  • Reciprocity Enigmas: 33
  • Reciprocity Up-Time: 223 seconds (13.72%)
  • Average Number of Procs: (29 + 30 + 33) / 3 = 30.67
  • Enigma Procs per Minute: 60 / (1625 / 92) = 3.397
  • Average Up-Time Per Enigma: (190 + 209 + 223) / 3 = 207.33 seconds (12.76%)

Do note that in neither of these parses could Wellness Enigma procs be tallied, since these are not logged unless you are below full health when it is triggered. But what data I have gathered demonstrates that Chaotic Enigmas are similarly affected in group and raid content as Doppelgangers. Chaotic Enigmas were triggered 30.17% fewer times per minute and their up-time 28.99% lower in the Mega-Boss parse than in the Agarthan dummy parse.

Because Doppelgangers and Chaotic Enigmas both demonstrate lower proc frequencies under the same conditions, I think it’s safe to conclude that Singularities must also proc at a lower rate, and that the problem isn’t specific to Doppelgangers but to Paradox generation. Again, it is still a mystery as to the cause, but at least we have evidence to believe that the problem must rest in the rate Paradoxes are generated in group and raid content, and not some peculiar problem for Chaos Magic damage dealers.

1 Like

Math isn’t my strong suit but could it be something to do with the higher defences reducing the chance of damage being divisible by 8? A rounding error or some such.

On the Anima - Sim: Nightmare parse, my Chaos Magic abilities glanced 109 times out of 4,411 hits. On all other parses, I did not glance any hits. The Mega Bosses do not have increased Protection compared to any of the Anima - Sim dummies as far as I can tell.

Yeah i dunno what would makes one enemy “optimal” conditins and another not. According to the mechanic description, the chance of dealing damage divisible by 8 has been modified, compared to normal statistics. In either case not sure what would make it fall outside the expected statistic. For my chaos user it might be glancing/evades, but nothing i have noted. She is only in the low Elites however so maybe. It does seem to me that the optimal conditions move, along with your character getting better stats, from my experience with the different Story Modes.

I noticed my fire shotgun shell stacks dropping off weirdly in world bosses, haven’t got a clue how to parse for that normally though. Just that on world bosses, my build that worked ok on dummies stopped keeping up. (shell salvage + passive for offhand shotgun, no combat reload - just using salvage at 3 shells remaining, spread out enough to maintain stacks)

As in no damage in ACT or just not visible on the boss bar?
Because there is a limit to the number of effects it will show on the world boss.

Instead of staying level, they’d drop to 0-1 in about a minute so parse about 1/6 as strong. The build can only increase stacks using the initial 15 energy at the start of a fight, after that it’s just maintenance alternating 5-6 stacks (rebuffing every 7s or so, which refreshes up to 6 and then drops to 5)

I believe we found a case where Paradoxes in general weren’t being generated consistently when an enemy has Exposed applied to them, or when a Chaos attack critically hits.

We’ll have a potential fix patched out as soon as we’re able. Keep an eye out for the patch notes!


How do you know this? As Jimmy said, if you’re looking at buffs, your own DB stacks might not be displayed after a while. If you’re tracking them with EffectsUI, you cannot track only your own stacks and you’ll always in fact see the most recently applied / refreshed stacks by anyone. You can see you own stacks briefly after you refresh them.

Might be a good idea to create a separate thread if you find out this is a real issue.

Sweet so there was something there =) thanks