Zero 5* in 29 pulls?

1235

Comments

  • Vhailorx
    Vhailorx Posts: 6,085 Chairperson of the Boards
    ThaRoadWarrior said:
    My suggestion was more like “if all these number lines exist, just run a smoothing algorithm on them to avoid any wide gaps or wild clumps” before people have to complain
    That's just a pity meter. No need to to tinker  with their RNG algorithm directly (which is probably provided by their middleware anyway).

  • Vhailorx
    Vhailorx Posts: 6,085 Chairperson of the Boards
    edited February 2022
    entrailbucket said:
    I have heard stories of people getting compensation for very low rates after they made a very large number of pulls, but I've never seen any proof that it actually happened.

    That seems like something that you can ask for if you are dropping 4-5 figures on the ge annually, but not a realistic option for the baby seals.
  • Siv75
    Siv75 Posts: 55 Match Maker
    Blackstone said:
    My record for pulling with no 5* is 62.
    But I kept pulling and drop rate got much better. Like a 5* every 3-4 that just kept coming.

    It's how I learned to only pull when I had a a lot of pulls available.
    Really sorry you had to go through a spell of 62. No matter what you got after... that quite some pain to get through. 

    I think there is enough points of reference where this can happen to showcase that something is off. RNG systems, if truly random, would not have patterns of a draught before playing catch up. If unlucky, it would drop significantly less across the distribution and not distribute more later on. To me, that shows a pattern if it behaves that way (later payoff). 

    There is no way for anyone to prove one way or another from the position of players ... only to share one's own experience. The point I have made is such experiences impact negatively and it may be worth looking into it from a developers standpoint. I have also narrowed down the issue to special stores. The other stores seem to have more RNG consistent distributions. 

    I feel awful when I read about the misfortunes of others too. And this is a game ... should be fun... that's the whole point. Make it fun... do better. That's all I can say. Cheers. 
  • Siv75
    Siv75 Posts: 55 Match Maker
    entrailbucket said:
    I have heard stories of people getting compensation for very low rates after they made a very large number of pulls, but I've never seen any proof that it actually happened.
    Based on my interaction with them, I do not feel this is true. Anyone is welcome to prove me wrong but selectively compensating a few players would be alike to opening a pandora's box for customer service. If there was indeed genuine issues, more likely corrections would be made to the game instead so it does not recur. I would in fact prefer that to getting back tokens or being given compensatory shards or characters. 
  • Bad
    Bad Posts: 3,146 Chairperson of the Boards
    I think that after reading what players exposed on this thread common words are dry spells and clumps of 5*s.
    That proves that we are experiencing an algorithm that serves the pulls in that way.
    In that case it would  be perfectly normal, fair, and acknowledged by devs.
  • Siv75
    Siv75 Posts: 55 Match Maker
    bbigler said:
    I’m a software developer as well, and any good random number generator function will seed off of another constantly changing number, like microseconds. Thus, reaching true randomness.  It can’t be predicted. 

    But when you do a quick pull of 50 tokens, (for the sake of processing speed), it’s not calculating the next random cover in a fraction of a second, but instead just loading the graphic for the next predetermined cover.  So, at some point in time before pulling, the random sequence was generated. 

    As another poster stated, this is likely done on the server (perhaps during account creation).  But it makes me wonder how far out the sequence is calculated.  It must be finite, which means it must re-seed and regenerate periodically.  After 1000 pulls? 10,000 pulls? 
    Perhaps it generates a random sequence of 100,000 pulls and then checks if it’s close to 15%.  If it’s not, then it recalculates a new sequence.  I would have programmed it that way and it would be fair to everyone.  
    Cool to meet a fellow developer. Some nerd talk for the soul perhaps?

    I think the consistent feedback from all views, no matter how opposing they seem, is that with a large enough sample pool, the draws come close to the odds. Let's take that as fact 1.

    And and as you rightly pointed out, RNG systems by themselves are random and commonly seeded by time in micro seconds. If called randomly, would provide random numbers that do not relate with each other. However, anyone knows that there is no such thing as truly random in computer science or math, and the way it is implemented makes it as close to random as possible, as such implementations can be open to error.

    The implementation (and use) of RNG can be done in many ways. You can call it once for each pull (completely independent) or you can randomise a sequence (random number generated within a string of numbers), or you can randomise a table (random 2 dimensional strings). In each case one would need to consider all types of permutations of characters available for 4 and 5 stars each, which adding the odds of 1 in 7 for 5 stars separately. That's some math! And this needs to be done at the server for all its players exchanging tokens worldwide, accuracy and performance being crucial. Fact 2.

    Looking at the 2 facts, the most logical design solution (and this is my hypothesis) is one would create a sequence, or a table for each store to define what can be pulled. Then another sequence would be required to randomise the pulls. This can be done either once when a player starts pulling (randomised string of numbers) or repeatedly for each time a token is pulled. The latter would be less efficient as it would require a lot more server CPU time (fact 2). Fact 1 also would suggest randomising a string of numbers and following the outcome of that string in sequence. 

    So in effect, based on how that string is randomised, you either get lucky with some pulls ahead, or unlucky with nothing in front, or get fairly compensated close to odds if the sample of pulls are close to the length of the string being randomised..all of which have been experienced by players. 

    Alright. Nerd out. 


  • bbigler
    bbigler Posts: 2,111 Chairperson of the Boards
    Since the developers revealed that Bonus shards have a weighted algorithm, meaning that the longer the drought is the greater the probability is for Bonus shards.  Perhaps they implemented something similar for Legendary pulls?

    First, I want to say that random is random.  You don’t know what it’s going to be like, so saying that because pulls have droughts and clumps, it must prove that it’s not random; but that’s false logic because you’re assuming that a random sequence will be evenly distributed.  It could be “clumpy” or “smooth”, it’s random.  

    But for the sake of this thought experiment, their algorithm could be weighted like bonus shards.  Let’s say the probability of pulling a 5* goes up 1% with each 4* pull.  Then when a 5* occurs, the probability drops 6%.  This would facilitate a 1 in 7 rate.  So the probability of pulling a clump of 5*s is likely after getting a long drought.  
    For example, you go 20 pulls without a 5*, so the probability increases from 15% to 35%.  Then you get a 5* and the probability drops down to 29%.  You then have good odds to get another 5* in the next 3 pulls.  If that happened you would have a 25% rate afterward.  Only after getting a clump of 5*s would the probability drop back down to normal. 
    That’s just a hypothesis; it could also be random.  But if this was true, it would explain the clumps & droughts.  
  • entrailbucket
    entrailbucket Posts: 5,622 Chairperson of the Boards
    edited February 2022
    People who make a LOT of pulls, who obsessively track these things, would have noticed that.  They haven't.  Nobody has ever found an effect like that, and there are a lot of people specifically looking for it.

    You wouldn't necessarily see the results of this kind of study but it'd reach escape velocity and become common knowledge pretty quickly.
  • HoundofShadow
    HoundofShadow Posts: 8,004 Chairperson of the Boards
    I have records of n-th pull for my last few hoards. This is my recent pull, so you can see how RNG works:

    ss = Sersi bw = Big Wheel ab = Abigail Brand
    The number on the left is the n-th pull.

    11: ss
    30: bw
    37: ss
    39: ss
    40: bw
    43: ?? bonus shards
    44: bw
    46: bw
    47: 50 4* bonus shards
    50: ab

    52: bw
    53: bw
    74: ab
    95: 50 4* bonus shards

    108: ab
    109: ab
    112: bw
    115: ab
    124: ss
    125: ss
    129: bw
    137: bw
    140: bw
    145: ab
    148: 50 4* bonus shards

    151: bw
    161: ab
    171: bw
    173: bw
    174: ab
    190: ss
    193: bw
    196: ab

    205: 50 4* bonus shards
    223: ss
    225: ss
    233: ss
    234: ab
    238: ss
    241: ss
    242: bb
    247: ss
    248: 50 4* bonus shards

    254: ss
    257: ss

    Total: 39 / 257 = 15.175%
  • KGB
    KGB Posts: 3,177 Chairperson of the Boards
    edited February 2022
    Some fairly dry streaks in Hounds numbers
    1) A 19 streak between 11-30
    2) A 21 streak between 53-74
    3) A 34 streak between 74-108 and more importantly, only 1 5* between 53-108 (55 pulls!)
    4) A 27 streak between 196-223
    Yet despite those droughts he still ended up with 15% over 257 pulls. This shows the value of only opening when you plan to open a very large amount.
    KGB
  • bbigler
    bbigler Posts: 2,111 Chairperson of the Boards
    From everyone’s feedback & numbers I’ve seen, pulls are “clumpy” instead of “smooth”. Interpret that however you like, but the end result always approaches 15%
  • dianetics
    dianetics Posts: 1,588 Chairperson of the Boards
    Bonus shards may be weighted. But legendary pulls certainly are not.
     If we expect 1 in 7 and it is weighted you would probably never see 30 pulls with no legendary tokens.
    If each pull increased the chances to pull a 5* then anything less than 5% (1/3 the advertised pull rate) probably would not happen.
    I get shards bonuses regularly, but certainly I do not see 5*s at the same rate
  • fractalvisions
    fractalvisions Posts: 308 Mover and Shaker
    This shows the value of only opening when you plan to open a very large amount.
    KGB

    Only if you're the kind of person who grumbles about long dry streaks. ;) How many you open in one go has no effect on the outcome. Although, perhaps an alternative is to open in small groups. You'll hardly be surprised if you open four tokens and they are all 4*.

    P.S. @entrailbucket I was literally just about to post about the shuffle story.  :D

  • bbigler
    bbigler Posts: 2,111 Chairperson of the Boards
    As I said above, it’s probably random and not weighted.  I just thought it was a fun hypothesis that fits the results.  
  • entrailbucket
    entrailbucket Posts: 5,622 Chairperson of the Boards
    fractalvisions said:
    This shows the value of only opening when you plan to open a very large amount.
    KGB

    Only if you're the kind of person who grumbles about long dry streaks. ;) How many you open in one go has no effect on the outcome. Although, perhaps an alternative is to open in small groups. You'll hardly be surprised if you open four tokens and they are all 4*.

    P.S. @entrailbucket I was literally just about to post about the shuffle story.  :D

    I love that story.
  • KGB
    KGB Posts: 3,177 Chairperson of the Boards
    fractalvisions said:
    This shows the value of only opening when you plan to open a very large amount.
    KGB

    Only if you're the kind of person who grumbles about long dry streaks. ;) How many you open in one go has no effect on the outcome. Although, perhaps an alternative is to open in small groups. You'll hardly be surprised if you open four tokens and they are all 4*.

    P.S. @entrailbucket I was literally just about to post about the shuffle story.  :D

    It's more than avoiding grumbling about dry streaks. Opening a bunch piecemeal means there is an excellent chance you get nothing / way under what you expect to get for a given special store (or latest legends).
    As I posted above someplace, getting a dry spell matters a lot in the short run because not all special stores/latest legends stores are the same. They are theoretically the same in that they contain 3 5*, but the quality of those 3 varies wildly so having a dry spell in a high value store hurts way more than in a low quality store.
    So someone who got a 10% draw rate this store can really fall far behind someone who got 20% in the same store. Even though neither player did anything to deserve the extra or lesser luck.
    KGB
  • bbigler
    bbigler Posts: 2,111 Chairperson of the Boards
    Yep, randomness is both fair and unfair.  
  • Bad
    Bad Posts: 3,146 Chairperson of the Boards
    HoundofShadow said:
    I have records of n-th pull for my last few hoards. This is my recent pull, so you can see how RNG works:

    ss = Sersi bw = Big Wheel ab = Abigail Brand
    The number on the left is the n-th pull.

    257: ss

    Total: 39 / 257 = 15.175%
    Am I right to guess that you champed the 3?

Categories