Lets Improve Magi Ring Farming

Grim Dawn General Discussion
1

TLDR: Magi ring farming is a dreadful experience and RNG sucks, please improve. Let this madness be recorded in history now

155, 173, 177, 184, 202, 209, 209,
225, 229, 232, 247, 256, 260, 273,
277, 291, 319, 358, 365, 392, 406,
408, 429, 455, 489, 492, 500

OK. Some might be asking what are all these numbers.
And if you guessed , that those are the runs on which Legendary Magi Rings dropped, you would be correct.

But why does it start at 155 and end at 500 you may ask.

Lets just say the goals of farming for rings escalated a bit

It started off simple:

  • Get Anubar Ring for my warlock build. When did it drop? Run 155 ofc, the start of our stat run recording.

But why did I not do other things that I have the ring for the build?

  • Might as well get the last ring missing - Shuroth Ring

Did not take long, got on early 200-ish runs EZ. But hey

  • I am actually closer to having two of each ring now. Lets just see how long it will take to get the last few rings.

“This is where Stan should have stopped” is what the legends say.
But no, I had to have a complete set of 2. How long did it take, surely it did not drop on the last run #500?
No… The last ring, the 2nd Shuroth ring dropped as early as run 489!!!

  • this is where most players would stop, but LETS JUST GO BEYONG INSANITY AND COMPLETE THE 500 RUNS

What do 500 runs look like?

image
image1063Ă—292 92.2 KB

Up to this point it was only for insanity and comidic reasons.
lets got inot the expected values first then the stats:

% to get an epic ring is 16%
% to get a legendary ring is 4%
% chance to get a ring is the sum of the above, 20%
Expected spawnsof each magi: #Runs * 2 / Magis = 500 * 2 / 7 ~= 142.86
Expected rings: 142.86 * 0.2 ~= 28.57
Expected Epic rings: 142.86 * 0.16 ~= 22.86
Expected Legendary rings: 142.86 * 0.04 ~= 5.71

Below are is the sum of legendary + epic rings
Tawrot: 10 + 21 = 31
Sethris: 6 + 18 = 24
Basti: 4 + 24 = 28
Khonsar: 6 + 21 = 27
Anubar: 4 + 31 = 35
Shuroth: 2 + 17 = 19
Orisia: 8 + 17 = 25

Total Rings: 40 + 149 = 189

Based on the total rings collected, some of the legendary ones are bolow the expected , some above, but that is RNG for you.
Actually got the exact expect value for legendaries, but epics were a bit off.
All in all this was probably a lucky run, considering the absolute drought during the first 50 runs, sadly undocumented.

While we can absolutely toss all this under RNG, lets continue a bit more.
If we use the total rings collected and devide it by 20%, we will get a relative value for how many Magi’s have spawned.

Tawrot Estimated Spawn: 31 * 5 = 155
Sethris Estimated Spawn: 24 * 5 = 120
Basti Estimated Spawn: 28 * 5 = 140
Khonsar Estimated Spawn: 27 * 5 = 135
Anubar Estimated Spawn: 35 * 5 = 175
Shuroth Estimated Spawn: 19 * 5 = 95
Orisia Estimated Spawn: 25 * 5 = 125

As we can see Shuroth has spawned 40 times less than expected. Or just above 10%, where 14% expected.
Note: here the expected spawn is 189*5/7 = 135

I am not suggesting there is something wrong with the drop % or the spawn chance rng. But before we go to where I think the issue lies, a few disclaimers.

  • The runs were performed on the Community League mod. But if the game is working as expected then this should not be an issue, because
    – The mod does not modify any of the Magis dbr files
    – The mod does not modify the lua file which handles the spawning
    – While the mod takes advatage rng ambush events and setting a seed for the RNG, the Magi spawn function actually sets a new seed when it executes. Which should eliminate the ambush events as a culprit.

  • Apart from the first 50 runs in a single day, all other were done across 3 weeks with 20-30 per day, while restarting the game every 10 runs (give or so depending on ring drops).
    This prolonged farming period should elimite some king of RNG/seeding issue at game start (if this is still a thing from TQ days)

  • Actually number of runs could be 490, as I might have forgotten to update my counter before leaving for a few days…

I think the way spawning is hangled is wrong.
Currenly each time you go to the Magis you get 2 out of 7 on random. The next run it is the same, 2 out of 7 on random.
This leaves you completely at the mercy of RNG, as if you don’t get the Magi you are after you have 0% to get that ring.

How do magis event spawn work?

  • First the game rolls a random number to chose 1 from 7
  • Then the game rolls a random numbe to chose 1 from 6

Doing some calculations again:
% to NOT get Magi X on first spawn is 1 - 1/7
% to NOT get Magi X on second spawn is 1 - 1/6
% to NOT get Magi X in every encounter is multiple of the above:
(1 - 1/7) * (1 - 1/6) = 0.857 * 0.833 = 0.714 or 71.4%

What I propose is to not allow Magis to repeat spawn in consecutive runs.
Meaning that if Anubar and Tawrot spawn on the first run, then they can spawn on the next run, and you will get 2 out of the 5 remaining Magis (new magis here are excluded for next run, and so on)
( I would actually apply this if the Magi was killed, instead of on spawn)

Math… again:
The math for the first run is the same as above. However, for the second run:
% to NOT get Magi X on second run, first spawn is 1 - 1/5
% to NOT get Magi X on second run, second spawn is 1 - 1/4
% to NOT get Magi X second encounter is multiple of the above:
(1 - 1/5) * (1 - 1/4) = 0.8 * 0.75 = 0.6 or 60%

But why calculate the chance to NOT have a Magi spawn?
Because now we can do a small simulation of 5 runs.

With the current way spawning work, here is the % to NOT have a Magi X spawn at all for 5 runs: This is just multiplying 71.4% five times
0.714^5 = 0.1856 or 18.56%

But if we use the new method we have this:
0.714 * 0.6 ^ 4 = 0.0925 or 9.25% to NOT have Magix X spawn in 5 runs

All this means, is that with the new method we will have a better distribution on spawns and are more likely to got the one we are after and is a small safeguard vs RNG bs.

I have also written a simulation in javascript for Magi runs, but I think the above reasoning is enough to prove the improvement this will bring.

Anyway here are tips on farming Magi rings:

  • Sustaining Skeleton keys.
    – Learn the treasure trove spawns, 3 out of 4 times you get it for free without major impact to time spent.
    – If you are not hording epics/legendaries dismantle what you do not need.
    – If you do not wish to dismantle you can run 2 troves in Gloomwald + celler + chance for nemesis and totem.
    – On your last skeleton key you can continue the dungeon run and get all the rare mats you need from the shop. When you buy one, run back to the staircase a few screens from where you came and wait 20-25s, then the shop will reset.
    – Convert brains and hearts into blood
  • Restart the game if you do not get any rings in 5 runs (I DO NOT BELIEVE GD IS FREE FROM TQ RNG)
  • disengage skill can help you cut corners if there are not targets for other movement skills.

Thoughts? Math Errors? Mad Errors? Other (GD Stash comments)?

5 Likes
2

still better than alkamos rings farm :grin: image

4 Likes
3

Isn’t this impossible to do in Grim Dawn engine? Or does it keep some data in-between returns to the main menu?

7r9YpEoErV
(gotta get that max conversion rolls)

Why not make it even less RNG, just repeat all the possible pairs of Magi one after another:
<1, 2>, <1, 3>, … , <1,7>, <2,3>, … , <6,7>
Well maybe not like that or in random order because then the 1st one repeats 6 times in a row.

4

Maybe then it wouldn’t have taken me…checks Steam achievements over 2 years since FG release to actually kill all of the Magi.

1 Like
5

to be fair, you can’t have done that much magi farming/ToH runs on 1/same character for that to happen :sweat_smile:
a lot of people had it pretty briefly after ToH release
there is a big difference in 489 runs needed to get all rings twice, vs simply just getting all 7 magi to spawn once for 1 char :grinning_face_with_smiling_eyes:

6

Well, I wasn’t actively doing ToH runs for rings at the time, but when ToH released, I went and completed it once with every character I had waiting at the start of Ultimate (which at that point was around 28), then I also did it with every character I created after that point (around 10) twice (once on Veteran, once on Elite) and then, I believe, I completed at least 6 characters on Ultimate before the achievement popped up. So that’s around 54-ish runs before I even met the seventh magus.

7

isn’t the magi achievement, like most/“all” GD achievement, counting for single/same char only tho?
which would maybe have delayed it a bit

8

It “must” be because if it was not then well 0.714^54 is very small chance :slight_smile:

[edit] well it’s not exactly that but probably something similar, that’s lot of counting

9

ye reading up on old posts seem people got it in like 1 hour or 15’ish runs or so back then, so 54 sounds high

10

This can be tracked by quest tags. Whenever the magis spawn you will check for the quest tags and if present take out the boss from the pool.

The same method I suggest can be extended, so that you have to do all 7 magis before getting a repeat. This would offer perfect distribution on spawns.

However, that would mean doing/waiting 2.5 before have a chance for the Magi you are after to repeat, and might just turn out into a feels bad moment when you only want 1.
Think what I suggest is sweet spot.

Yeat this one like many has to be on a single character.

1 Like
11

Stopping shortly after a long run induces some kind of bias (if you would have defined the 500 runs before you started, it would be different, but your number of runs is a result of your experiment - so it is slightly problematic here). And obviously, using the expected number of spawns, also makes the stated number of spawns rather senseless (as there is quite some uncertainty) [/bean counter :wink: ]

But well, got me motivated to make a small simulation (long live monte carlo if you are too lazy to calculate the exact distribution). So I repeated the 500 runs 100 times to see how bad your experience was:

  • the lowest number of an individual magi “spawn” was 105, the highest 185 (5% quantile: 128, 95% quantile: 160) [sampling 2 out of 7 in each run]
  • the number of rings (epic and legendary) was between 14 and 49 (5% quantile: 20, 95% quantile: 38) [assuming a binomial distribution using the number of spawns and as prop = 0.2]
  • the number of lengendary rings was between 0 and 14 (5% quantile: 2, 95% quantile: 10) [assuming a binomial distribution using the number of rings (legendary or epic) and as prop = 0.2]
  • the total number of rings was 19989 (slightly below the expected 20k)
  • the total number of legendary rings was 3934 (also slightly below the expected 4k)

Now, what can be expected from a 500 farming run

  • 23/100 resulted in less than 20 of one of the rings (legendary or epic)
  • 19/100 resulted in less than 2 of one of the legendary rings
  • 3/100 resulted in one legendary ring not got at all

Concluding remark, you was lucky - could have been much worse :stuck_out_tongue:

1 Like
12

13

(GDStash comment): And yet people still use x2 Shuroth ring on every acid build imaginable ::sigh::

(Meanwhile, me farming rings):

1
11280Ă—1024 200 KB

I need 2 Tawrot rings, don’t judge me :C

2 Likes
14

RNG is the problem here. You can do 500 runs and never get the ring you need. This kind of RNG is imo too much for a single player game.
Ideally, there should be a way to transmute Magi Rings into each other but this is only possible with set items.
Other way would be to add them to Egellon so you can exchange one for another.
Yet another solution would be to add a blueprint for those rings while lowering their chance to drop so overall chance to get either a ring or a blueprint stays the same. Not the most elegant solution because there’s still a possibility for a 500 runs without dropping what you need but at least you have to do it only once.

15

I think you could probably also make a crafting recipe that makes a random Magi ring out of some Magi ring

3 Likes
16

yeah, as a scientific experiment this is shit.
Was never meant as an experiment, just resembled something at the end.

Maybe someday someone will record all of it :rofl:

Wonder what your simulation will give as numbers if my idea was taken into account.

Mine gives me a boost of minimum spawns of about 15-30

17

yeah RNG sucks.

All good ideas. Sadly, looking at how the alkamos ring grind has improved over the years, I doubt Zantai will introduce any of this :frowning:

For a starter I would hope Zantai actually read something from all of that wall of text :joy:.

2 Likes
18

I’m just so glad that my favourite damage type (chaos) doesn’t have Magi’s rings as best in slot :smiley:

3 Likes
19

Why would you rob single players of the joy of finding something very rare?

1 Like
20

This is why triple-rare green MIs exist :grin:.