[X] My Assistant

Honestly, I don't understand why you people approach with average damage.

In the real cases, it's how the combination of damage and CR affects turns. If not in terms of turns, the method you determine better styles in terms of average damage could be misleading.
For my part, I believe willow(oak for holy) is better for all cases. The more damage you do, the more resist hinders your speed in terms of turns.

Lestion's method of mage analysis seems to be very similar to what I do with melee: take sssss2's experiment data and calculate further interpreted results from it. As well as calculate interpreted results from the game formulas directly when possible. Then try to mathematically compare and logically explain any differences.

Since I can actually understand some mage stuff nowdays I'd like to cross-check his calculations but unfortunately with me having zero mage experience, I'm not worthy. I can only hide under the blankets and watch. Maybe Sapo84 will be able to fix any mistakes. Edit: guess he did

The question of how much Resist monsters have mirrors the question of how much Parry monsters have, which I've discussed with Sapo84 before. I plan to return to this topic eventually. Some new Overpower data would be welcome, but it won't be easy to gather because someone has to temporarily make Overpower 5 weapons and it wastes a lot of credits. Let's wait before doing it.

And, bear in mind that we do play imperil against SG's. All mages. There is no non-imperil play against SG.

Dark DOT through ripened soul: 3.7% of all damage inflicted. That is better than upgrading DD2 to DD3!

Thanks, this all makes sense to me. I recently calculated that School Girls actually have less PMit (and probably MMit) than typical monsters but the difference doesn't matter here; it's still enough. Because they have way more HP they can't be blown away in a few turns, so all styles of maging benefit by casting Imperil on them first to lower the MMit.

This post has been edited by BlueWaterSplash: Oct 23 2019, 00:34

Honestly, I don't understand why you people approach with average damage.

In the real cases, it's how the combination of damage and CR affects turns. If not in terms of turns, the method you determine better styles in terms of average damage could be misleading.
For my part, I believe willow(oak for holy) is better for all cases. The more damage you do, the more resist hinders your speed in terms of turns.

My post has now been corrected and you will find that your belief is correct even in the case of average damage.

But if you would have done me the courtesy of fully reading my post, it did say on MANY occasions that this is the case only for raw damage output, and NOT for the actual effect this has on turn counts. I did explain that there are other factors that weigh into that because of the nature of mage's damage delivery (AoE) and that if you would've overkilled everything with either staff, then one resisting monster can cost you a turn to cast an extra spell.

sssss2 added new elemental comparison data, thanks!

CODE

- Setup B: Attacks with 3 strikes
======================================================================================================================
      Setup B |   total |       damage |  average ||     hit |  average ||   crit |  average |  crit rate |     bonus
----------------------------------------------------------------------------------------------------------------------
          You |  11,979 |  315,356,105 |   26,326 ||   5,415 |   19,006 ||  6,564 |   32,364 |    54.80 % |  170.28 %
  Elec Strike |  11,979 |   55,752,039 |    4,654 ||  11,979 |    4,654 ||        |          |            |          
  Cold Strike |  11,979 |   55,113,393 |    4,601 ||  11,979 |    4,601 ||        |          |            |          
  Void Strike |  11,979 |  110,848,085 |    9,254 ||  11,979 |    9,254 ||        |          |            |          
======================================================================================================================

- Setup C: Attacks with 3 strikes
======================================================================================================================
      Setup C |   total |       damage |  average ||     hit |  average ||   crit |  average |  crit rate |     bonus
----------------------------------------------------------------------------------------------------------------------
          You |  12,027 |  315,848,858 |   26,262 ||   5,538 |   19,361 ||  6,489 |   32,151 |    53.95 % |  166.06 %
  Dark Strike |  12,027 |   62,400,600 |    5,188 ||  12,027 |    5,188 ||        |          |            |          
  Cold Strike |  12,027 |   55,748,816 |    4,635 ||  12,027 |    4,635 ||        |          |            |          
  Void Strike |  12,027 |  111,966,181 |    9,310 ||  12,027 |    9,310 ||        |          |            |          
======================================================================================================================

Elec Strike 55,752,039 total 4,654 average
Cold Strike 55,113,393 total 4,601 average

Dark Strike 62,400,600 total 5,188 average
Cold Strike 55,748,816 total 4,635 average

What happened? These results are way different from before! And Cold Strike has Flame Shield and it's about the same as Elec Strike? Edit: I found the reason. Because this is tested on Electric Day and Dark Day, the result is difficult to interpret.

I think that on the -10% mitigation Element Day the damage is roughly increased by 20%. So in today's monster population Cold Strike with Fire Shield is 8% better than Dark Strike? That's a huge change from Research for 1H.

Can sssss2 repeat this test on element neutral days? For example use Elec+Cold and Dark+Cold on the Fire, Wind, or Holy day. Maybe also use Haste so that it's the same as Research for 1H those years ago. (I don't know if you used Shadow Veil before).



I notice Void Strike still does only 9254/19006 = 48.69% or 9310/19361 = 48.09% of the main hit. This is better than before, because we fixed the 3 strikes problem. But it's still not 50% so maybe there is another problem that I don't understand...

In your original Research for 1H data, the elemental strikes only do 96% damage when they crit. I don't know why. Can you separate the elemental strike crits in this new data?

Maybe everybody these days made weak PL monsters, or everyone made Celestials? There is an unanswered question of how the game chooses monsters. Can your script check how many monsters of each species you met? Did you meet more Celestials than Giant? Or did all monster species appear equally?

This post has been edited by BlueWaterSplash: Oct 23 2019, 07:48

The conclusions here invalidate my earlier findings and my hypothesis altogether: this is a very clear ruling in favour of the status quo - that counter-resist has a significant benefit on damage which is not outweighed by the EDB differences between staffs.

My previous post now reflects more correct results, which are in favour of Willow definitively.

Since it's not quite as good on Willow, what about Arctic? Is Peerless Arctic Willow Staff of Destruction better or worse now than Peerless/Legendary Arctic Redwood Staff of Destruction/Niflheim?

Edit - this post has been superceded by newer, better models. Please see my post here for more data about staffs.

This post has been edited by lestion: Nov 13 2019, 03:35

So in today's monster population Cold Strike with Fire Shield is 8% better than Dark Strike? That's a huge change from Research for 1H.

Now that I think about it and did more secret calculations, I think this is a believable result with a stronger PL monster population, but such a huge change should still be confirmed if possible.

CODE

Cold Strike |  18.74% | * Flame Spike
Dark Strike |  18.60% |
Holy Strike |  18.51% |
Fire Strike |  17.93% | * Shock Spike
Elec Strike |  17.17% |
Wind Strike |  17.12% |
Cold Strike |  16.90% |
Fire Strike |  15.47% |

If the change is this great, then Fire Strike + Shock Shield would also become better than Dark Strike and Holy Strike, 17.93% x 1.08 = 19.36%

Cold Strike would be 18.74% x 1.08 = 20.24%

And my belief is that it's also worth it to use Cold Shield and Wind Shield if you have a Tempestuous/Shocking rapier. By my calculation there will be a penalty of 6% due to the counter attacks problem. So Elec and Wind Strike would be 19.00% (estimated since never tested with shield) x 1.02 = 19.38%

So it may already be true that Dark and Holy Strike are the weakest elements against regular monsters! In the future as typical monster PL continues to grow, Dark and Holy Strike will continue to get weaker.

Maybe I can test it also, or anyone could test it. I don't think I have the same script so I would have to test in a less accurate way, but it should still be good enough to verify if Dark Strike and Holy Strike became the weakest elements now.

-snip-

That's interesting.

I also want to remind about dark and holy imperil in case of calculations about imperil style.

-snip-

Added some data to the original post.

CODE

Void Strike: almost 50 % (9,511 - 9,665)
Dark Strike (day bonus): 27.87 % (5,396 / 19,361)
Cold Strike (day bonus): 26.61 % (5,097 / 19,157)
Elec Strike (day bonus): 25.17 % (4,783 / 19,006)
Cold Strike (spike shield): 24.73 % (4,701 / 19,006), 0.2488 (4,817 / 19,361)
Dark Strike (lacks shield): 22.67 % (4,342 / 19,157)

Void Strike is 50% now, great! I guess the elemental crit mystery caused the problem.

It seems my previous estimate that Cold Strike is 8% stronger than Dark Strike was correct. Your new data shows 24.73% / 22.67% = 9%.

So it's probably really true that in today's world, Dark and Holy Strike are the weakest elements against normal monsters. I'm pretty sure even Fire, Wind, and Elec are stronger if you use the correct shield. I estimate they are 4~5% stronger with their shield.

Of course without spike shield Fire, Wind, Elec, and Cold are about 10% weaker than Dark and Holy. This should not change much from Research for 1H many years ago.

The reason for the big change with spike shield is that when monsters' PL rises their average mitigation rises. In Research for 1H the average Dark not-mitigated was 52% today it has become 22.67% x 2 ≈ 45%. In the future if most monsters are PL 2250 it will reach 40%. This means that after 20 more years Dark and Holy Strike will receive the same weakening a second time.

The spike shield -25% mitigation is additive, therefore its effect becomes much larger after this happened. The day bonus -10% is also additive. And you found an interesting result: Dark strike on Dark day is stronger than Cold Strike on Cold day. I expected this because if you stack the day and spike shield together their effect won't improve as much.

Most days won't match your bonus so that's not the main important result. I think Dark and Holy melee prices should drop after this.

Dark and Holy are still way better against School Girls, this will never change. They are 4x damage (+300%) compared to elements. With Imperil the difference is about 1.5x damage (+50%).

And you found an interesting result: Dark strike on Dark day is stronger than Cold Strike on Cold day.

I turned off a flame spike shield when testing a cold strike on Wednesday.

So, Dark strike on Dark day without Spike shield is stronger than Cold Strike on Cold day without Spike shield.

This is an obvious result.

Most days won't match your bonus so that's not the main important result. I think Dark and Holy melee prices should drop after this.

I think Ethereal is the best prefix for 1H.
SG Arenas: Holy and Dark strike
GF, IW: Cold strike with Flame spike shield and other strikes which match their day bonus.

And I still think Hallowed/Demonic is better than Arctic or other elemental prefixes.
Most low-mid level melee players spend most of their time clearing SG Arenas.
Even if they sometimes try GF or IW with Hallowed/Demonic weapons, the penalty will be very small.

A player who spends hours clearing GF?
Go mage.

I think it is better to bring following thread (for research about riddlemaste)
https://forums.e-hentai.org/index.php?showtopic=229352

Oh you turned off flame spike shield? But I feel your results are what I would expect if you did use flame spike shield on Wednesday. Maybe next week you can turn flame shield on and compare.

I feel it is strange that Cold Strike with day bonus is so much stronger than Elec Strike with day bonus, without spike shield. 26.61% / 25.17% ≈ 6%. In your old Research for 1H, Elec and Cold are about the same without spike shield.

It might still be possible to argue that Dark and Hallowed could be as good as Arctic in regular arenas; my idea from long ago is to infuse the strike of the day bonus, then switch your spike shield to match this element. Only Dark/Hallowed weapon has the freedom to choose this. Arctic weapon must always keep the flame shield.

I am planning to discuss that idea more later. Right now I'd still like to study the new results and check for mistakes.

This post has been edited by BlueWaterSplash: Oct 24 2019, 16:08

Average Resist of Monsters

A couple of years ago I did lots of tests, and calculated that the base resist of monsters are
- Weakest: 13~ %
- Average: 20~ %
- Strongest: 27.1 % (maximum)

So I inferred that
- Weakest: WIS 4 % * TOKEN 0 % * PFUDOR 10 % = 13.6 %
- Average: WIS 7 % * TOKEN 5 % * PFUDOR 10 % = 20.48 %
- Strongest: WIS 10 % * TOKEN 10 % * PFUDOR 10% = 27.1 %

[attachembed=135277]



Recently I did 20 x PFUDOR grindfest as a cold mage.

CODE

===============================================================
        - |    cast |   target |    gains |  resists |    rate
---------------------------------------------------------------
  Imperil |  68,570 |  192,266 |  179,919 |   12,347 |  6.42 %
===============================================================

===================================================================================================================================
           - |    cast |          damage |  resisted |   target ||     hit |  average ||   blast |  average |     rate |     bonus
-----------------------------------------------------------------------------------------------------------------------------------
      Freeze |   2,830 |     502,814,269 |    9.64 % |    3,047 ||   1,404 |  120,783 ||   1,643 |  202,821 |  53.92 % |  167.92 %
    Blizzard |  15,690 |  10,234,004,969 |    9.06 % |   41,875 ||  19,052 |  177,658 ||  22,823 |  300,104 |  54.50 % |  168.92 %
  Fimbulvetr |  19,940 |  48,241,533,735 |    8.91 % |  153,362 ||  69,538 |  228,653 ||  83,824 |  385,825 |  54.66 % |  168.74 %
===================================================================================================================================

It seems that the average resist has increased.

Average Deprecating Resist
- Estimated: 6.09 %
- Actual value: 6.42 %

Average Damage Reduction
- Estimated: 8.39 %
- Actual value: 8.91 % ~ 9.64 %

Average Resist of Monsters

A couple of years ago I did lots of tests, and calculated that the base resist of monsters are
- Weakest: 13~ %
- Average: 20~ %
- Strongest: 27.1 % (maximum)

So I inferred that
- Weakest: WIS 4 % * TOKEN 0 % * PFUDOR 10 % = 13.6 %
- Average: WIS 7 % * TOKEN 5 % * PFUDOR 10 % = 20.48 %
- Strongest: WIS 10 % * TOKEN 10 % * PFUDOR 10% = 27.1 %

[attachembed=135277]
Recently I did 20 x PFUDOR grindfest as a cold mage.

CODE

===============================================================
        - |    cast |   target |    gains |  resists |    rate
---------------------------------------------------------------
  Imperil |  68,570 |  192,266 |  179,919 |   12,347 |  6.42 %
===============================================================

===================================================================================================================================
           - |    cast |          damage |  resisted |   target ||     hit |  average ||   blast |  average |     rate |     bonus
-----------------------------------------------------------------------------------------------------------------------------------
      Freeze |   2,830 |     502,814,269 |    9.64 % |    3,047 ||   1,404 |  120,783 ||   1,643 |  202,821 |  53.92 % |  167.92 %
    Blizzard |  15,690 |  10,234,004,969 |    9.06 % |   41,875 ||  19,052 |  177,658 ||  22,823 |  300,104 |  54.50 % |  168.92 %
  Fimbulvetr |  19,940 |  48,241,533,735 |    8.91 % |  153,362 ||  69,538 |  228,653 ||  83,824 |  385,825 |  54.66 % |  168.74 %
===================================================================================================================================

It seems that the average resist has increased.

Average Deprecating Resist
- Estimated: 6.09 %
- Actual value: 6.42 %

Average Damage Reduction
- Estimated: 8.39 %
- Actual value: 8.91 % ~ 9.64 %

Could you please update HV Utils with this information?

Recently I did 20 x PFUDOR grindfest as a cold mage.
...

Thanks for separating your data into casts, and targets. This is really interesting and useful.

Looking at the imperil data in particular: 68,570 casts over 20 fests is an average of ~3429 casts per run, which in turn is ~3.4 casts per round. That's enough for me to assume that these tests assume you recast until everything is affected by imperil.

Thanks to this data, we can assume the actual average monster resist is currently somewhere around 21.6% (6.42/0.2972 = ~21.6).

This post has been edited by lestion: Oct 24 2019, 16:44

Monsters Statistics

- 2,513 monsters / 167,318 times for 20 x PFUDOR grindfest

CODE

========================================================
      Class |  Count |  Average |  Encounter |    Ratio
--------------------------------------------------------
      Giant |    362 |       60 |     21,575 |  12.89 %
  Dragonkin |    248 |       74 |     18,369 |  10.98 %
  Celestial |    249 |       73 |     18,112 |  10.82 %
     Undead |    220 |       75 |     16,457 |   9.84 %
  Arthropod |    260 |       63 |     16,426 |   9.82 %
  Mechanoid |    188 |       68 |     12,692 |   7.59 %
  Elemental |    152 |       68 |     10,322 |   6.17 %
      Beast |    154 |       66 |     10,125 |   6.05 %
     Daimon |    142 |       71 |     10,069 |   6.02 %
   Humanoid |    128 |       75 |      9,648 |   5.77 %
     Sprite |    124 |       68 |      8,449 |   5.05 %
  Reptilian |    152 |       55 |      8,299 |   4.96 %
      Avion |    130 |       52 |      6,741 |   4.03 %
       Rare |      4 |        9 |         34 |   0.02 %
========================================================

==================================================================
    PL |  Count |  Average |  Encounter |    Ratio |  Accumulated
------------------------------------------------------------------
  2250 |     46 |      479 |     22,049 |  13.18 % |      13.18 %
  2150 |      2 |      479 |        957 |   0.57 % |      13.75 %
  2100 |      3 |      459 |      1,378 |   0.82 % |      14.57 %
  2050 |      1 |      410 |        410 |   0.25 % |      14.82 %
  2000 |      2 |      401 |        802 |   0.48 % |      15.30 %
  1950 |      5 |      361 |      1,806 |   1.08 % |      16.38 %
  1900 |      1 |      309 |        309 |   0.18 % |      16.56 %
  1850 |      4 |      302 |      1,209 |   0.72 % |      17.28 %
  1800 |     89 |      232 |     20,685 |  12.36 % |      29.65 %
  1750 |      5 |      261 |      1,304 |   0.78 % |      30.43 %
  1700 |     29 |      162 |      4,699 |   2.81 % |      33.23 %
  1650 |    221 |      129 |     28,558 |  17.07 % |      50.30 %
  1600 |    156 |      100 |     15,675 |   9.37 % |      59.67 %
  1550 |    323 |       70 |     22,611 |  13.51 % |      73.19 %
  1500 |    564 |       50 |     28,480 |  17.02 % |      90.21 %
  1450 |    327 |       24 |      8,010 |   4.79 % |      94.99 %
  1400 |    301 |       16 |      4,939 |   2.95 % |      97.95 %
  1350 |    117 |       11 |      1,329 |   0.79 % |      98.74 %
  1300 |    291 |        4 |      1,152 |   0.69 % |      99.43 %
  1250 |      6 |        1 |          6 |   0.00 % |      99.43 %
------------------------------------------------------------------
  1200 |      5 |       59 |        295 |   0.18 % |      99.61 % # MONSTER DATA IS OUT OF DATE
  1150 |      1 |       14 |         14 |   0.01 % |      99.62 % # MONSTER DATA IS OUT OF DATE
  1100 |      1 |      112 |        112 |   0.07 % |      99.68 % # MONSTER DATA IS OUT OF DATE
  1050 |      2 |       33 |         65 |   0.04 % |      99.72 % # MONSTER DATA IS OUT OF DATE
  1000 |      2 |       61 |        121 |   0.07 % |      99.80 % # MONSTER DATA IS OUT OF DATE
   950 |      1 |      100 |        100 |   0.06 % |      99.85 % # MONSTER DATA IS OUT OF DATE
   900 |      4 |       52 |        209 |   0.12 % |      99.98 % # MONSTER DATA IS OUT OF DATE
------------------------------------------------------------------
     0 |      4 |        9 |         34 |   0.02 % |     100.00 % # BOSS
==================================================================

=====================================
  Mitigation |  Average |  Imperiled
-------------------------------------
        fire |  59.91 % |    21.83 %
        cold |  56.01 % |    19.01 %
        elec |  59.11 % |    20.64 %
        wind |  57.38 % |    18.75 %
        holy |  53.35 % |    28.66 %
        dark |  51.51 % |    26.83 %
=====================================

================================================
  Rank |          Trainer |  Count |  Encounter
------------------------------------------------
     1 |         Nero-Arc |    200 |     16,332
     2 |          serorin |    127 |     14,135
     3 |         morineko |    182 |     13,296
     4 |           sssss2 |    200 |     12,230
     5 |          gc00018 |     97 |     11,336
     6 |      FreeloaderV |    101 |      9,213
     7 |     MidNightPass |     78 |      8,951
     8 |         RoadShoe |    200 |      6,409
     9 |         NerfThis |    200 |      6,384
    10 |            Koaen |      7 |      3,298
    11 |          danixxx |     17 |      2,953
    12 |        qdjseh001 |     66 |      2,874
    13 |         Mantra64 |     27 |      2,250
    14 |     Sushilicious |    116 |      1,980
    15 |       Hina Amano |    175 |      1,969
    16 |           DJNoni |     33 |      1,943
    17 |          Godde?? |     32 |      1,901
    18 |       Petal_Kiss |      7 |      1,744
    19 |        .@_@.@_@. |     49 |      1,663
    20 |         cylinnia |     28 |      1,490
    21 |         tempasdf |     14 |      1,428
    22 |         StonyCat |      4 |      1,423
    23 |         Honeycat |      5 |      1,345
    24 |           ddwiki |      4 |      1,255
    25 |   Kira.Yoshikage |     22 |      1,180
    26 |          lestion |     10 |      1,133
    27 |      Void Domain |      6 |      1,030
    28 |        sabregimp |     12 |      1,025
================================================

This post has been edited by sssss2: Oct 24 2019, 23:04

With sssss2's latest data, I was able to build an imperil simulator. I will remain vague about how this was achieved (parts of it could be adapted for unscrupulous means, if I did).

I simulated 10,000 fests for various levels of counter-resist, and compared how many imperil casts this actually saves.

I recommend skipping straight to the tables if you don't care about disclaimers for what parts may not be accurately modelled.

First, I want to talk about the part with the most questionable accuracy: with no data for round size distribution, I made some guesses.
- Using sssss2's T3 target count (which, judging by his previous data on page 3, is cast almost exactly once per round - probably excluding the very early rounds very the monster count is low enough that imperil coverage doesn't give enough time for it to come off cooldown) we can assume an average monsters per round of ~7.67.
- From this, I used some guesses about the distribution of monsters per round in order to give a similarly matching average. In this case, it was ~10% 5 monsters, 12% 6 monsters, 16% 7 monsters, 25% 8, 35% 9 and 2% 10.
- Obviously, we cannot have much confidence in these values. If anyone has actual data, I can easily re-run these tests to give more accurate results.
- I do think they're reasonable accurate, though, and should not make enormous changes to the results.

There is also another small problem: monster resist values may average out to 21.6%, but obviously this is not the case for every monster, and max resist count monsters may have unexpected effects on how many imperils have to be recast. (If you can't visualise this problem, just take my word for it!) To combat this, I used two methods for resist distribution.
- In one set of data, I populated all monsters with a static 21.6% resist.
- In another, I used random distribution equidistant from the 21.6% mark - up to the 27.9% max, and down the same distance to 15.3%. This is also unrealistic in practice because of plvl skewing, but it essentially offers a 'chaos' mode that should properly represent whether outliers with different resist chances have a strong effect on actual casts required.

In actual practice, it turned out these averaged out nicely over the large data set anyway, so they probably don't need urgent fixing.

Anyway - the most relevant parts of these tables are the CR and the average casts per fest. They're actually slightly off from sssss2's real data, but this could be put down to a difference in imperil behaviour or just variance in a smaller data set (I did not model it on a per-fest basis, so I do not know how much difference there could be between runs).

CODE

==========================================================================================================
|All monsters have 21.6% resist                                                                          |
---------------------------------------------------------------------------------------------------------|
|CR  | Casts    | Targets  | Hits     | Resists  | TargRes% | CastRes% | P.Loss   | P.Loss% | Avg Casts  |
|0%  | 43381929 | 98221939 | 76901036 | 21320903 | 21.71%   | 49.15%   | 15380941 | 54.93%  | 4339       |
|40% | 36739443 | 88364557 | 76903901 | 11460656 | 12.97%   | 31.19%   | 8738228  | 31.21%  | 3674       |
|70% | 32275652 | 82242668 | 76904786 | 5337882  | 6.49%    | 16.54%   | 4274489  | 15.27%  | 3228       |
|84% | 30263292 | 79654004 | 76898862 | 2755142  | 3.46%    | 9.1%     | 2264861  | 8.09%   | 3026       |
---------------------------------------------------------------------------------------------------------|
Monsters have randomly distributed resists between 15.3% and 27.9%                                       |
---------------------------------------------------------------------------------------------------------|
|CR  | Casts    | Targets  | Hits     | Resists  | TargRes% | CastRes% | P.Loss   | P.Loss% | Avg Casts  |
|0%  | 4350610  | 98421050 | 76898733 | 21522317 | 21.87%   | 49.41%   | 15561855 | 55.58%  | 4357       |
|40% | 36788146 | 88417084 | 76900713 | 11516371 | 13.03%   | 31.3%    | 8788060  | 31.39%  | 3679       |
|70% | 32287168 | 82254298 | 76901115 | 5353183  | 6.51%    | 16.58%   | 4287638  | 15.31%  | 3229       |
|84% | 30266374 | 79653003 | 76897175 | 2755828  | 3.46%    | 9.11%    | 2266277  | 8.09%   | 3027       |
==========================================================================================================

Key:
CR - Counter-resist
Casts - Imperils cast total
Targets - Total number of targets imperil was cast against (does not include already-afflicted targets)
Hits - How many of the imperils hit against their target.
Resists - How many of the targets resisted imperil.
TargRes% - The rate at which individual monsters resisted imperil.
CastRes% - The rate at which casts of imperil were met with a resist by any of its (up to) 3 targets.
P.Loss - the amount of extra casts needed as opposed to a theoretical perfect run with absolutely 0 resists.
P.Loss% - same as above, expressed as a % increase in turns compared to the perfect run.
Avg Casts - the average amount of imperil casts per fest.

Hopefully this is adequately explained. Again, if you have feedback on the methods I used to model uncertain parts, I will happily re-run the simulation given better data regarding these problems.

edit: note that the max is actually 27.1%, not 27.9%. I'm not sure if I used the correct figure in the calcs and just wrote it wrongly here, or if it was wrong in there too. that said it doesn't really make any significant difference to the results (if anything, greater distance between min and max values still coming out roughly the same just serves to highlight that it doesn't matter which way it was simulated)

This post has been edited by lestion: Nov 16 2019, 00:13

That monster data is amazing, thanks! I assume sssss2 gathered it playing as mage because the average elemental mitigations are slightly lower than his recent 1H tests. For example Dark mitigation is 51.51% instead of 1 - 2 * 22.67% = 54.66%.

Monster Selection

The game tries to artificially balance out the PL of monsters you encounter. Therefore I expect average PL will not increase much in the future. So Demonic/Hallowed melee should not continue to get worse.

The game does not balance species like I expected. Players have made tons of Giants; so you will fight lots of Giants. This explains why Arctic could do more damage than Shocking compared to years ago. If players made different monsters, the relationship can change to anything.

The weakening of Demonic/Hallowed compared to years ago is not because of monster species, but because of spike shield: it is more effective against high PL monsters with more elemental resistance.

The "Average" column isn't equal among species and its variation is complex. Assuming there is no Natural Weighting to favor some species, I believe it's instead the result of the game trying to balance the encountered PL. The 4 species maximum per round can also affect it.

Giant: 362 total, 60 average. Most common by far, but most have a low PL.
Dragonkin: 248 total, 74 average. Common and tending to have high PL.
Celestial: 249 total, 73 average. Common and tending to have high PL.
Undead: 220 total, 75 average. Common and tending to have high PL.
Arthropod: 260 total , 63 average. They are common, but their PL is low.
Mechanoid: 188 total, 68 average.
Elemental: 152 total, 68 average.
Beast: 154 total, 66 average.
Daimon: 142 total, 71 average. Tending to have a high PL.
Humanoid: 128 total, 75 average. Though few in number, I think many have PL 2250 so are selected often.
Sprite: 124 total, 68 average. Though few in number, I see many have PL 2250 so are selected often.
Reptilian: 152 total, 55 average. Few in number and selected rarely due to 4 species maximum.
Avion: 130 total, 52 average. Few in number and selected rarely due to 4 species maximum.

Monster Selection Rules

1. Choose a random desired PL between 1500 and 2250.
2. Choose a random monster who is close to this PL from the pool of all monsters.
3. Repeat until all monsters or 4 species are chosen.
4. If 4 species are chosen then future selections are restricted to those species.

This is my guess for the rules.

What is the Future of Monsters?

1. PL gap between 1800 and 2250 may fill in. This might increase the average PL a little more, but it definitely will never be all 2250 or even close to that, because the game always rebalances it.
2. Army of stupid Giants will get stronger. Fiery and Arctic may regress and get weaker.
3. Army of primitive Arthropods will get stronger. Fiery could turn to crap.
4. Giants might be good at killing mages or something, so people might make even more.

I think there have always been tons of Giants. In Research for 1H I wondered why Arctic was the same as Shocking/Tempestuous without spike shield. If you look at the original Monster Lab elemental resistance table then Cold should be the best elemental and Fire also good. The easiest way to explain it was Giants, but why should they be most important? Sure they have a lot of HP and PMit but I felt it wasn't enough. Now that I saw the population it makes sense.

If the army of stupid Giants rises in PL not only will their elemental resistances increase, but they will also be selected more often. In effect there will be more Giants.

I previously estimated that 1/4 of monsters have the burn ailment at any given time. Against a PL 2250 Giant that means Cold on average does (25% + 25% + 25% + 50%) / 4 = 31% damage while Dark/Holy does 38% damage. This means Dark/Holy could actually improve in the future again.

Recent data suggested Cold became 6% stronger than Electric when it used to be the same. This is probably because players made more Dragons and Arthropods, who are weak to Cold.

Upon further thought, Cold with flame shield became 9% stronger than Dark but it's likely 6% of this is due to more Dragons and Arthropods. So the phenomena of increased PL over many years only made a 3% difference.

I confirm this is expected with a separate calculation: In Research for 1H, 18.60% / 16.90% = 1.10 so Dark was 10% stronger than Cold. Flame shield gives 18.74% - 16.90% = 1.84% improvement to Cold, or 1.5 * 2 * 1.84% = 5.52% adjusting scaling for consistency.

Typical Dark non-mitigated damage dropped from 52% to 45.34%. Ignoring the Dragons and Arthropods change, Cold would otherwise have dropped almost equally, 0.91 * 45.34 = 41.22%. With flame shield it becomes 46.74%. Comparing to Dark this is 46.74%/ 45.34% = 1.03 = 3% improvement.

Therefore, improvement in Cold over Dark is actually mostly due to species, just like Electric. Since players can change this as they wish, we should not overreact. Maybe next year Cold will be worse. Or it could be stronger because of the Arthropods and Reptilians currently growing.

I think innately, Cold should be best because it has the best original value in the elemental resistance table. Later I will try to calculate it theoretically when all monster species are made equally.