1. Hello!

    First of all, welcome to MapleLegends! You are currently viewing the forums as a guest, so you can only view the first post of every topic. We highly recommend registering so you can be part of our community.

    By registering to our forums you can introduce yourself and make your first friends, talk in the shoutbox, contribute, and much more!

    This process only takes a few minutes and you can always decide to lurk even after!

    - MapleLegends Administration-
  2. Experiencing disconnecting after inserting your login info? Make sure you are on the latest MapleLegends version. The current latest version is found by clicking here.
    Dismiss Notice

Drop rate is regulated by num of mobs you have killed lately

Discussion in 'Guides' started by Imberos, May 28, 2022.

  1. Imberos
    Offline

    Imberos Horny Mushroom

    46
    12
    46
    Nov 20, 2018
    12:48 PM
    Liffon
    Priest
    96
    Dear friends,

    im writing this down after i have started an experiment to test drop rates.

    The story (how it began): i was doing a quest at ToT that required me to kill 999 mobs of each kind. At the first kind of mobs i noticed i got 4 mob cards at the first 300 mobs, and then the 5th at around the 900th mob. that got me thinking, maybe the drop rates are somehow regulated by a mob counter? i kept recording data for the following 3,000 mobs and got similar result (i will post the results later on once i collect more data). Strangely at one point when i was about 600/999 done with another kind of mob i took a 10 mins break (after getting 4 cards at the first 300 and then 0 for the next 300). when i came back i got 3 cards from killing 200 mobs! this seems interesting.

    What am i saying?: im saying that i think the drop rates are regulated by some sort of counter on the server side. it is possible (and most likely is) that every mob (and item) has different value of drop rate. this experiment isnt about testing whats the drop rate for every item in the game, but to check whether this counter regulation is true or not.

    What do i need?: im looking for some1 to help me think of a testing methodology at the first stage (im an electrical engineer with signal processing major), and once we'll come up with one we will publish it and ask for volunteers for collect data in the easiest way possible. once this is all done we can share the results.


    thank you for reading, please help :)
     
    • Informative Informative x 2
    • Creative Creative x 1
  2. zeroxlr
    Offline

    zeroxlr Windraider

    437
    500
    296
    Oct 11, 2020
    3:48 AM
    xxxxSnuggles
    Beginner, Camper, Islander
    Usually when I’m playing on my camper, I get anywhere between 2 to 4 vote cash coupons or gacha tickets per hour.

    3 mob kills/ respawn cycle of 10 seconds

    60sec —> 6x3= 18mobs per min

    18 mobs x 60min = 1080 mobs per hour

    2 to 4 nx per hour

    540mobs to 270mobs killed per 1 drop

    There are also days when I get 0 nx drops lol
     
    • Informative Informative x 1
  3. Ainz
    Offline

    Ainz Zakum

    1,669
    1,081
    490
    May 2, 2015
    Male
    Netherlands
    11:48 AM
    So many I keep forgetting
    0
    Rng is gonna rng
     
    • Agree Agree x 6
    • Like Like x 1
    • Disagree Disagree x 1
  4. OP
    OP
    Imberos
    Offline

    Imberos Horny Mushroom

    46
    12
    46
    Nov 20, 2018
    12:48 PM
    Liffon
    Priest
    96
    That is not entirely true.
    If my hypothesis is true then the calculation for drop rate is something like:
    Base drop rate - (counter*factor)

    With the right methodology and enough data we can test if this is true or not regardless of rng being rng.

    Think about this: if we find out that the first 100 mobs is about 5 times more likely to drop something than the last 100, and getting it consistently numerous times is this rng?
     
  5. Ainz
    Offline

    Ainz Zakum

    1,669
    1,081
    490
    May 2, 2015
    Male
    Netherlands
    11:48 AM
    So many I keep forgetting
    0
    The reason I commented what I did is precisely because I don't believe your hypothesis to be true. People come up with similar ideas all the time. (an example is monster cards dropping more frequently after having completed the set)
    To properly test your hypothesis, you'd need people to log their grinds fairly extensively and then still get a sizeable dataset. I'd be very surprised to see the community work together like that for this.
     
    • Agree Agree x 3
  6. -ovv
    Offline

    -ovv Horntail

    2,246
    880
    500
    Feb 23, 2020
    Male
    3:48 AM
    -ovv
    Beginner
    200
    Honor
    Confirmation bias

    It’s a known fact that complaining about ninja nerfs will bring you good luck.
     
    • Agree Agree x 2
    • Friendly Friendly x 1
  7. OP
    OP
    Imberos
    Offline

    Imberos Horny Mushroom

    46
    12
    46
    Nov 20, 2018
    12:48 PM
    Liffon
    Priest
    96
    i dont understand, u say im wrong because people come up with similar ideas? or u say im wrong because it will be hard to test?

    the right argue to why i am wrong should include something about RNG properties that explain how drops behave. if you have this info please share it with us, if u dont please dont throw general unbiased claims
     
    • Great Work Great Work x 1
    • Creative Creative x 1
  8. zeroxlr
    Offline

    zeroxlr Windraider

    437
    500
    296
    Oct 11, 2020
    3:48 AM
    xxxxSnuggles
    Beginner, Camper, Islander
    The easiest and the hardest place to test is in Tutorial.
    • count the number of kills per cycle (every 10 seconds)
    • tutorial mobs ONLY drop NX... other than Event Etc/Tutorial Etc
    In normal maps, you have ERROR... LOTS OF IT.
    • unable to map clear before respawn cycle in most maps without multiclienting or help
    • Questionable... is NX Drops based off of Variable of each:
      • Server Time
      • Channel
      • Map
      • Mob
      • Number of Mob Killed per Cycle
      • Number of Total Mob Killed per Time Frame
      • 1h per 1 kill
      • Multi Hits per 1 kill
    As you can see... there's a lot of questions and hypothesis that can be made about NX Drops, especially without the exact formulation.

    Since we only see results, it'd most likely need lots of data to analyze, by using a good Machine Learning Cluster program.

    Of course... this is all approximation of Unknown Variables for the resultant NX Drop of at least 1.
    • we can only make Estimates based on Known Variables.

    TLDR:
    Pray to RNGesus
     
  9. OP
    OP
    Imberos
    Offline

    Imberos Horny Mushroom

    46
    12
    46
    Nov 20, 2018
    12:48 PM
    Liffon
    Priest
    96
    your right about variables we need to consider, yet we can factor them out partially by considering:
    • Server Time - i dont think drop rate is based on server time (it will cause unfair gameplay)
    • Channel - we can restrict data to be collected from specific ch only
    • Map - we can test different maps and compare them to themselves (like at clear "mob counter" vs non clear)
    • Mob - same as map
    • Number of Mob Killed per Cycle - TBD
    • Number of Total Mob Killed per Time Frame - TBD
    • 1h per 1 kill - TBD
    • Multi Hits per 1 kill - i dont think this will effect drop rate (but this is just gut feeling)
    about the use of machine learning (after i have practiced it myself in the past) i try to keep this data collection as low as possible, meaning we will not have enough data to rely on machine learning (we cannot use CNN for example but i think we can use MLE or MAP since it requires less data with our prior)

    maybe the 999 quest is a good place to consider since you have the game itself to keep u updated with ur mob count, just need to add timer to that.
     
    • Informative Informative x 1
  10. Ainz
    Offline

    Ainz Zakum

    1,669
    1,081
    490
    May 2, 2015
    Male
    Netherlands
    11:48 AM
    So many I keep forgetting
    0
    As irony would have it, this is actually one of the very few 'theories' with regards to factors that affect droprate that have had any sort of confirmation, as far as I'm aware. (As well as it being one of the very few factors that would be reasonable to have an effect on droprate)

    Edit: Wait, I may have misunderstood. I thought this was referring to the amount of mobs killed per hit as opposed to vice versa. Apologies if I got it the wrong way around!
     
    • Agree Agree x 1
  11. Eleza
    Offline

    Eleza Mano

    11
    7
    10
    Apr 8, 2022
    6:48 AM
    Eleza
    F/P Wizard
    103
    Community
    Interesting, but I think your drops may have been circumstantial at best. Sadly the best way we can look into something like this would be to look at the raw code and understand it, so if it's possible for you, I would just look there, since the answer would be there.

    It's like playing roulette and putting everything you have in the same spot every time. Just because it didn't hit 25 black last round doesn't mean your chances increase or decrease. However at the same time, who is to say the wheel isn't rigged to prevent heavy bidders like yourself from winning that pot? You won't know unless you can take a closer look at how everything works.

    Almost all games I play with a loot table usually just roll and whatever you roll is what you get. It feels like you should get a zak helm every time you kill zakum, and then suddenly you don't get one in a run despite you just having a run when it drops 4.

    Same goes for gach or nx. I could get them to drop more frequently within an hour, but then the next hour I come up dry. if something has a .1% chance, you're throwing the equivalent of a d1000 every time you kill something, but that's not saying you will get it within 1000 kills, you have a 1/1000 chance every time you kill it.

    Also; this game is so rng heavy, I would actually be surprised if the drops acted any differently. And if it DOES act differently, then we need to look into how other things like scrolling or gaching works then, as well.
     
    • Agree Agree x 1
  12. Ainz
    Offline

    Ainz Zakum

    1,669
    1,081
    490
    May 2, 2015
    Male
    Netherlands
    11:48 AM
    So many I keep forgetting
    0
    [​IMG]
     
    • Agree Agree x 3
    • Informative Informative x 1
  13. LeonardoJF
    Online

    LeonardoJF Zakum

    1,820
    336
    460
    Jun 16, 2021
    Male
    Rio grande do Sul - BR
    7:48 AM
    ItzLeo
    Paladin
    200
    Favela
    I remmember i played a server that had a estimated like 1 drop per 400 mobs when u used a command, but, rng is rng
     
    • Agree Agree x 1
  14. smellypete
    Offline

    smellypete Blue Snail

    3
    0
    17
    Nov 27, 2021
    Male
    10:48 AM
    goucha
    Hermit
    77
    In my opinion, machine learning would be hard to apply due to the amount of variables that can interfere with the probability; and the construction of your dataset. I think a alternate path you could take is to research maplelegends. Since no staff member (atleast that I know of) was transparent about the topic of rng and people just believe it works as intended without understanding the topic you have to do it yourself, namely reverse engineering.

    It could be entirely possible that the drops are defined by a pseudo random number generator that does not produce a uniform distribution or is a non cryptographical prng. (check pdf on: Serious Cryptography: A Practical Introduction to Modern Encryption - Jean-Philippe Aumasson - chapter 2 on Randomness).

    Regarding reverse engineering, either by debugging using ollydbg or cheat engine, so that you can find memory addresses which are related to this rng process. Then, disassembling using IDA the code itself until you find the functions you want so that you can understand how they work and clarify your post goal. for example (https://www.youtube.com/watch?v=___jEOjGCOY - there are communities that do these type of research, although I'm not sure if you have permission to do these things in maplelegends).
     
    • Informative Informative x 1
  15. Cornwall
    Offline

    Cornwall Chronos

    85
    34
    93
    Dec 11, 2020
    Male
    6:48 AM
    Cornsair
    Corsair
    Pasta

    My best recommendation for you right now is to gather some friends and do your experiment privately and if you then deem the information worth sharing then you present it here. Its almost human nature for people to reject your ideas on the outset just because they cant envision/believe it themselves (aka troll sometimes).

    Good luck on your data collection.
     
    • Creative Creative x 1
  16. zeroxlr
    Offline

    zeroxlr Windraider

    437
    500
    296
    Oct 11, 2020
    3:48 AM
    xxxxSnuggles
    Beginner, Camper, Islander
    smellypetesmellypete CornwallCornwall
    • It’s human nature to be invested in your own ideas
    Why would anyone be willing to spend their time and energy to help you if you’re not helping yourself?
    • If you’re an entrepreneur, you’d still need to prove that your idea/product is worthwhile and safe to garner investors’ help.
    You guys are welcome to privately assess the mechanics of some other copy of MS GMS v63 elsewhere, or by signing up to be a staff member.
    • Since as a basic user, we’re not allowed access to see the server database, nor code.
    • Even the staff can’t disclose certain information to us either.
    If you wish to try the Pseudo RNG method with external programs…
    • Dont… you may get banned
    • I don’t condone any ban-able behaviors
    • Do so at your own risk bruhhh
     
    • Informative Informative x 1

Share This Page