Magic Find and its efficiency: A statistical insight

  • #1063
    How can you be so unlucky? I think the magic find gods are not pleased with you, probably from the moment you started deciphering their randomness in giving and taking.
  • #1064
    Quote from HuiTzi

    How can you be so unlucky? I think the magic find gods are not pleased with you, probably from the moment you started deciphering their randomness in giving and taking.


    well, it took me like 7 full clears, but I finally found one!


    And on my last off-day too ;) RNG can indeed still kick in.
  • #1065
    awww that's an awful skorn... those useless legendarys... just vendor it for the more usefull 3900g.
  • #1066
    What about a 1.0.5-list of ilvl/slot distributions? Could reveal the changes to 1.0.4.


    I think I can provide plain texts of items like this:
    Name: CINCH JUSTICE
    Slot: Waist
    Quality: Rare
    Type: Belt
    Armor: 256
    Attributes:
    +84 Intelligence
    +43 Physical Resistance
    +6% Life
    20% Better Chance of Finding Magical Items
    iLvl: 62
    rLvl: 60


    Edit: We should check the color distribution for the bonus drop: Similar to slot 1/4? Similar to slot 2/3? Similar to NV slot?
  • #1067
    So... that bonus drop that has a chance of occurring when a monster is killed in MP>1... think it's like one of the earlier slots? Or something new?
  • #1068
    Quote from ztking

    So... that bonus drop that has a chance of occurring when a monster is killed in MP>1... think it's like one of the earlier slots? Or something new?


    It looks to me to be identical to the existing trash item slots. It feels like it's better than elites slot 2, but not quite as good as elites slot 1.
    In any case I am not sure how we can figure it out. Does it apply to elite/boss kills? I think I read somewhere that it only applies to "normal" monster kills, meaning pack members of elite packs that isn't the last/the leader and other trash. If that's true it'll be difficult to determine.

    For the spreadsheet I treat it like an additional trash item-drop.
  • #1069
    Just wanted to give my thanks as I was doing some data gathering on the AH and this helped.

    Not sure if anyone pointed out this on 54 pages but;
    update on the max MF

    - Amulet 45
    - Ring 30 (Nagelring)
    - Quiver / Orb / Mojo 20
    and you may add Weapon 45 (Sun Keeper)

    This is relevant to section 2,7
  • #1070
    Quote from CamogaFly

    Just wanted to give my thanks as I was doing some data gathering on the AH and this helped.

    Not sure if anyone pointed out this on 54 pages but;
    update on the max MF

    - Amulet 45
    - Ring 30 (Nagelring)
    - Quiver / Orb / Mojo 20
    and you may add Weapon 45 (Sun Keeper)

    This is relevant to section 2,7


    I updated section 2.7 with new numbers. I used to have weapons on the list too, but I decided to leave it for armor only for the time being and see if blizzard ends up completely removing the affix from items.
    Thanks for the input.
  • #1071
    Wow the amount of data collected and analyzed here is staggering.
  • #1072
    Quote from Crotazzo

    awww that's an awful skorn... those useless legendarys... just vendor it for the more usefull 3900g.


    instead vendoring you could salvage this, 10k for brimstone
  • #1073
    So much Information, very helpful!
  • #1074
    do you plan on doing a study on 'randomness' of legendary drops, you might say quality here, ie. mempo or andariel, in relation to mf/mp?

    Or maybe that is much more complex than a mf.

    Reason why I ask this is because I noticed that on plvl90-95 (no additional mf on mp1) i was getting at least 5 times a day gladiators gauntlets, out of 15-20 leg/set items.
    on plvl90-95 (no additional mf on mp1) i was getting at least 5 times a day the gidbinn ceremonial knife, out of 15-20 leg/set items.

    On plvl100 I never got any of those not even once and I played on all mp's.

    Also one more thing I noticed is set chest armor, on 10 natalya's chest I am getting 2 tal rasha chest (got arount 30 nataly and 5 tal rasha's, 1 inna's and 1 zunnimasa chest all of that is in lifetime playing cant really give some avrage mf/mp)
  • #1075
    I would be really surprised if there is anything related to specific legendary items. I think your drops are just a random fluctuation.
  • #1076
    to The Master of Magic Find
    Hi, I've only read initial post, sry can't handle 54 pages, but question is very simple. in 1.0.5 legendary drop was doubled, is this reflected in graphs/tables in initial post? 0.001 drop rate you use for legendary is for 1.0.4 or 1.0.5? also where this number comes? i doubt you have big enough data pool to get that specific on legendarily, like you totally can on rares/blues.

    tnx for answer :)
  • #1077
    Although I'm not one of the masters of magic find, I dare to answer.

    "The following presented data have been collected in Diablo III v. 1.0.3.10235 - v. 1.0.4.11327."
    => It's based on pre-1.05.

    Many of the relevant sections have been updated, e.g., section 2.8, as mentioned at the beginning of the post. However, as you said, I don't think this study is representative for the drop rate of legendaries. It's mainly to find out about the affix rolls on rares - and there's good reason to believe that nothing has changed here since 1.05. The only way to find out if all of this is still true is of course to collect new data...
  • #1078
    The results on rares should still be valid, although there was no testing done to confirm this 100%, if I recall correctly. Regarding the doubling of the legendary drop rate: the excel file has been changed to reflect this change, but the initial number (0.001) was never 100% confirmed due to the rather low drop rate of legendaries. However, the results outputted from the excel file seem to reflect my average legendary drop rate quite well, on a 'per run' basis. So the number shouldn't be too far off.
  • #1079
    I Just want to thank you for the effort you put into this, it was amazing.
  • #1080
    Quote from Qwaygon

    to The Master of Magic Find
    Hi, I've only read initial post, sry can't handle 54 pages, but question is very simple. in 1.0.5 legendary drop was doubled, is this reflected in graphs/tables in initial post? 0.001 drop rate you use for legendary is for 1.0.4 or 1.0.5? also where this number comes? i doubt you have big enough data pool to get that specific on legendarily, like you totally can on rares/blues.

    tnx for answer :)


    The drop rate that is given in the calculation is the one that has been doubled.
    We estimated it to be approximately 1/2000 chance for each item slot in pre-1.05. When 1.05 hit I simply doubled it to 1/1000 (= 0.001).
    We have no legitimate way of saying this is near the true value. As a matter of fact, the random deviation there is for our data around the legendary drops is, simply speaking, huge compared to the approximated distribution. We did, however, narrow down the options and eventually we settled with a nice round number (being 1/1000 at 0 MF atm).

    Keep in mind that the legendary drops are only directly reflected in the model system and this model system is, well, a model! It will never be able to actually show the true nature of the drops, only come close to it, and I believe the approximation of 1/1000 chance for legendary item at 0 MF is reasonable.

    If there is a disagreement on this then of course I will take it into consideration, but as people have already mentioned this study does not answer the underlying questions that are of legendaries. It's more about showing the trends.

    Glad you read it though, and don't worry, just reading the original post is a task in itself.

    - - -

    Quote from Bagstone

    Although I'm not one of the masters of magic find, I dare to answer.

    "The following presented data have been collected in Diablo III v. 1.0.3.10235 - v. 1.0.4.11327."
    => It's based on pre-1.05.


    Updated to:
    The following presented data have been collected in Diablo III v. 1.0.3.10235 - v. 1.0.4.11327.
    The doubled drop chance for legendaries in 1.05 have been considered for the calculations.


    Like a ninja...

    - - -

    Quote from dulac84

    do you plan on doing a study on 'randomness' of legendary drops, you might say quality here, ie. mempo or andariel, in relation to mf/mp?


    No.

    I believe there is no relation between mf/mp and the actual legendary roll. I believe that is totally up to iLvl and randomness to a certain degree.
  • #1081
    Quote from Cyeron

    Updated to:
    The following presented data have been collected in Diablo III v. 1.0.3.10235 - v. 1.0.4.11327.
    The doubled drop chance for legendaries in 1.05 have been considered for the calculations.


    Like a ninja...



    this solves everything, and somehow reflect what i was expect things to be :) i do agree that 1/1000 looks reasonable assumption

    tnx for answer )
  • #1082
    IMO, legendary drop rate analysis is definitely needed. Also patterns need to be tracked. I notice if i save 40-50 legendary over 3-4 days of grinding i will see a pattern of every 5-6 legendary a repeat and sometimes even a triple may just be a trap or trick, as i have only gotten maby 400-600 legendary since 1.05 patch. I will farm extra hard if I get a good legendary and so far am rewarded more often than not on a double in the next 2ish days from initial drop.

    I would also like to think investigations into legendary drop locations could be pursued(woudl take months of data but not alot of work depending on amount of participants). Doing alkaizerish runs dayin and dayout there seems to be something going on with legendaries dropping in first 5 stacks NV/whites-in-amungst usually at the end of zones(going backwards in alkaizer run its your start into that zone). I even do arreat 2 from cursed tower 1 wp + backtrack from arreat2 wp to that hotspot at end of heart zone +go keep dephs3 clear to keep dephs 2. But again maby the nature of my runs is just feeding into that perception(more reason to look into it can you say better farming routes for legendary gathering specifically) without tracking all the data from many people theres no way to tell.
  • To post a comment, please or register a new account.
Posts Quoted:
Reply
Clear All Quotes