Regular RNG unluckiness -or- sadistic gremlin manipulating my MPQ life?

Godzillafan67
Godzillafan67 Posts: 498 Mover and Shaker
Though I don't keep spreadsheets, the introduction of shards has given me a glimpse into my 5* pull rate. Since shargeting began, I've accumulated 249 shards, which means that I've opened 83 legendary tokens; mostly classics but some latests and specials as well.

Of those 83 tokens, I've pulled 3 5*s, two black Carnage and one blue Doc Ock. Therefore, instead of seeing a 1:7 droprate, I'm looking at about 1:27. And I firmly believe but have no data to back it up that I was already pulling at an abysmal rate prior to sharding.

Despite my natural affinity for logic, I'm just going to presume that it's the sadistic gremlin in the title's presented binary choice. Any spreadsheet folks with their own data to point out that this is fairly standard RNG bringing me down?

Comments

  • Rhipf
    Rhipf Posts: 294 Mover and Shaker
    To be fair, 83 pulls is a rather small sample size for comparing your pull rate to the stated pull rate.
  • bluewolf
    bluewolf Posts: 5,248 Chairperson of the Boards
    I mean do you want me to bring up again my 13.4% pull rate after 3960 pulls?  Someone posted here once about 0/77, I've gone 0/46.

    RNG is Random.  Except maybe for me, my account was set at "lower end of 5* pulls" apparently.
  • ThaRoadWarrior
    ThaRoadWarrior Posts: 9,126 Chairperson of the Boards
    83 classic, or a total of 83 classic and latest? Diluting you’re pulls across multiple stores makes it “feel” like you are below odds, but you might be in several simultaneous troughs on your independent strings of pulls in each location.
  • helix72
    helix72 Posts: 991 Critical Contributor
    It's generally held that the chance of getting a 5* is 15%. I say generally held: the game says the chance of a 5* is "~1/7", which means "about" 1/7. 1/7 is 14.2%, which is about 15%. And other users who have kept long logs of pull rates have validated that 15% seems to be the number.

    That said, in 83 pulls the chance of pulling 3 or fewer 5* covers is 0.09%, which is pretty low. But anomalies happen, and as much as I love a good conspiracy theory, it doesn't mean the system is rigged.
  • Godzillafan67
    Godzillafan67 Posts: 498 Mover and Shaker
    83 classic, or a total of 83 classic and latest? Diluting you’re pulls across multiple stores makes it “feel” like you are below odds, but you might be in several simultaneous troughs on your independent strings of pulls in each location.
    Aren't they independent of each other? I presumed that every time I opened any token the RNG calculator did its thing without consulting any sort of previous information.
  • helix72
    helix72 Posts: 991 Critical Contributor
    It was said to me previously on this forum (though I don't know the source) that each pre-determined sequence of pulls is separate for each type of pull. That said, they are independent, so the overall chance of getting 3 or less is the same.
  • IIAlonditeII
    IIAlonditeII Posts: 150 Tile Toppler
    83 classic, or a total of 83 classic and latest? Diluting you’re pulls across multiple stores makes it “feel” like you are below odds, but you might be in several simultaneous troughs on your independent strings of pulls in each location.
    Aren't they independent of each other? I presumed that every time I opened any token the RNG calculator did its thing without consulting any sort of previous information.
    If you mean "is there a pity timer", then the answer is sorta no.

    If you mean "does x failed pulls mean a successful pull later" then the answer is sorta yes.

    When you create your account, the game apparently flags your "big pulls" straight away, according to what some users here have reported the devs as saying in discord.

    This means that the lifetime pull rate for 5 Stars is always set to the correct number (we assume to be 15%). You are guaranteed to pull 15% if you pull long enough, because a string with exactly that % has already been determined.

    Now that may mean, for your particular account, 170 consecutive 4Star pulls followed by 30 consecutive 5Stars, so eh.
  • bluewolf
    bluewolf Posts: 5,248 Chairperson of the Boards
    @IIAlonditeII

    Oh, if only there was some way that the game corrected your draw rate towards the target of 15%. By the way, it IS 15%. 

    As I said. I have recorded 3,960 pulls and my 5 rate to date is 13.4%. Maybe it might round up to 13.5%. 

    Surely if there were some controlling mechanic my pull rate would be in the neighborhood of 14.5% or so, at worst, by now. 

    The game does nothing to control your rate. It’s all purely RNG. The devs have made comments here and there in the past and verified that there is no streak breaker nor do they plan to do that. The Targeted Hero/shard system shows you what that kind of system might look like. No bad runs but no lucky streaks, just slow and boring gradual progress.  
  • Borstock
    Borstock Posts: 2,520 Chairperson of the Boards
    I faced an AI the other day who collected 70 AP on turn 1.

    RNG stinks, but it's the nature of the beast.
  • Godzillafan67
    Godzillafan67 Posts: 498 Mover and Shaker
    Why is the return rate on 5* 15%? If the odds are 1:7, then it's 1 out of every 8 that should be a 5*, so 1/8 = 12.5%.

    @helix72 - I'd have to see someone from the dev crew say that they create and reference a sequence of predetermined pulls. While I can see value in doing such a thing, especially in enforcing quality control and addressing player engagement and experience, I bet they've created a "function" or two to return RNG results. Doing so would match with what IceIX was talking about in the "Bonus Heroes are dead. Yay there are shards now!" thread when it was said that, although the playerbase's BH rate was at/near the expected value, some players were getting Bonus Hereoed well below that (unlucky) while others were being blessed (lucky).
  • ThaRoadWarrior
    ThaRoadWarrior Posts: 9,126 Chairperson of the Boards
    83 classic, or a total of 83 classic and latest? Diluting you’re pulls across multiple stores makes it “feel” like you are below odds, but you might be in several simultaneous troughs on your independent strings of pulls in each location.
    Aren't they independent of each other? I presumed that every time I opened any token the RNG calculator did its thing without consulting any sort of previous information.
    It’s because they are all independent that it can feel worse or better depending on where you are on both pull tracks. Consider:

    your destiny in Latest Legends has you in a 25 pull dry streak. 

    Your destiny in Classic has you on a 26 pull dry spell.

    a release store pops up, and you are on a “normal” 1:7, but the 5* is coming on Pull 7. 

    It is conceivable that by splitting your pulls across all these stores, that you will hit 57 consecutive 4* pulls with no 5*, whereas you’d have broken through the trough had you committed to a single store. 

    Over time that isn’t meaningful, but when you average a pull or two per day that can be very demoralizing.
  • Godzillafan67
    Godzillafan67 Posts: 498 Mover and Shaker
    I was going to update this thread when I opened token #100, but #101 came along on its heels and gave me a 5*, so I'm including it too. Of the new 18, three were 5*. Although that run itself is an exact 1:5 beating the given 1:7, I'm actually doing worse than 1:15 since shards came along, which means that I've pulled less than half of what's expected.

    I neither believe that any nefarious plot is afoot nor that bad coding is affecting me personally. I just find this interesting and/or, as @Th@ThaRoadWarrior puts it, demoralizing (womp womp ;) ).

    It occurred to me that testing my results is comparable to testing an 8-sided die, on which 7 of the sides have a "4" and the 8th has a "5". Googling on how to carry out such an analysis suggests that some sort of chi-square computation and degrees of freedom lookup table would give a meaningful result. I'd take a stab at this myself, but I fear that I'd get it wrong and report an erroneous result.

    Image result for dilbert random number generator

  • ThaRoadWarrior
    ThaRoadWarrior Posts: 9,126 Chairperson of the Boards
    Start tracking pulls separately on a google sheet, see if per-store you have more “accurate” odds. This is sort of like recalculating your GPA by only factoring in courses within your major. I’d be curious to know what that looks like.
  • helix72
    helix72 Posts: 991 Critical Contributor
    @Godzillafan67 I agree with you on multiple fronts, but still note it does not change my observations. First the odds:
    In any real world gambling scenario, 1:7 odds would mean 12.5% as you correctly point out. However, based on feedback and testing from others (e.g. @The rockett mentioned in a thread customer service told him 15% LINK, and then others have been tracking pulls over a large number of tokens like this thread LINK) it appears the real number is 15%. Of course, that is based on feedback and observations from others so there is no guarantee.

    As for the predetermined pulls, again I'm going off things said by other vets who frequently mention it. As I understand it, before predetermined pulls, there was some way to game the system, though I don't know exactly how it worked. Again, I'm relying on information supplied by other vets including @pheregas so perhaps they can give more color on the background of where this came from.
  • TPF Alexis
    TPF Alexis Posts: 3,826 Chairperson of the Boards
    helix72 said:
    @Godzillafan67 I agree with you on multiple fronts, but still note it does not change my observations. First the odds:
    In any real world gambling scenario, 1:7 odds would mean 12.5% as you correctly point out. However, based on feedback and testing from others (e.g. @The rockett mentioned in a thread customer service told him 15% LINK, and then others have been tracking pulls over a large number of tokens like this thread LINK) it appears the real number is 15%. Of course, that is based on feedback and observations from others so there is no guarantee.
    I think it might just be that whoever set the format for that didn't know the difference between 1:7 odds and 1 in 7 odds. 1 in 7 odds is about 14.3%, so 1 in 7 is a pretty reasonable fractional approximation for 15%.
  • bluewolf
    bluewolf Posts: 5,248 Chairperson of the Boards
    Look at the game display, it shows "~1:7" as the odds.  I'm not sure why they came up with 15% but that's the number and 1:7 is pretty close - but of course not truly 15%, hence the ~.

    You really need to track a LOT of pulls before you could feel bad about a run that isn't close to 15%.  I'd say at least 500 or so?  There may be some people who are better at stats and things who could come up with a number of pulls that would close in on that (most of the time).

    But really it's all sort of moot because even if you had some crazy run of, say, 1 5* in 100 pulls, that's a possible result (very unlikely but not impossible) and since the game DOES NOT TRY TO ADJUST YOUR RESULTS in any way, even a very bad result could happen and complaining about it would probably just result in a reply saying "sorry, it's all random".  I mean even if there was a .000001% chance of getting 1 5 in 500, well, that's still possible.

    It's kind of interesting to consider that there have been a lot of accounts activated, and maybe there's been some extremely bad run that either no one complained about (just gave up on the game), or just didn't get far enough along (didn't play hard enough) to see it, or never tracked anything and similarly never complained.  So maybe someone has gotten 1 5 in 200 pulls, or that was the predetermined result, but we just don't know about it.

    Lastly re: predetermined pulls, there is something built into the code so that your next pull and the next and the next is locked in.  Years ago you could preview pulls, see what was coming and reject the pulls to try again until you saw what you liked and accepted the pulls.   The thing that most people would bother trying to manipulate would obviously be 5's, so some people would re-roll their pull and keep trying until they got a 5 instead of a 4. 

    That's gone now because they set it up so you can't change the next cover you get no matter what (other than, when a new character enters the token pool, you might get a different cover because the list changes). Now, the tier you pull is locked in 100%. It could be that your next 5 in Latest is (for example) 10 away, and that cover could change when they change who is in  Latests.  But you'll always get a 5 from that pull regardless.  (Same thing applies when a new 4 comes into the pool, it might change which 4 you get next but you'll still get a 4.)
  • Vhailorx
    Vhailorx Posts: 6,085 Chairperson of the Boards
    The odds splash page is total garbage and has been since bonus heroes was introduced.  It emphasizes simplicity over accuracy to the point of blatantly misleading players about bonus hero odds for the entire lifespan of that feature.

    1:7 was chosen because 1:6 risks overrepresenting the drop rate (16.667% > 15%).  The devs explicitly said that the LT drop rate did not change when the odds splash page was revised (for bonus heroes), and it had previously been shown as exactly 15%.

    As has been said, the string of pulls is predetermined out to infinity for everyone for every store.  So assuming you play a lot, you should eventually get to ~15% from both the LT and classic stores. (Sorry for your **** luck bluewolf, but I think being off by less than 2% over 4k pulls probably isn't all that unusual.  Anyone want to calculate the standard deviation for that?).

    The only way to have a pull rate above 15% over the long haul now is to get very luck in some special stores.  E.g., I pulled ~700cp from the rescue release store and got something like a 45% drop rate.  And I got 3/6 5*s from the BRB release store.  So if I could magically only pull from stores where I had randomly good distributions in the first 5-20 pulls, theny lifetime draw rate could stay above 15%. 

    Otherwise, basically all of this stuff is just confirmation bias and the human inability to recognize true randomness (subconsciously, we all expect randomness to mean "evenly distributed").
  • T_REZ5000
    T_REZ5000 Posts: 147 Tile Toppler
    edited December 2019
    I’ve only recently started tracking my pulls out of curiosity.  So far have 348 pulls with 47 5*s.  Thats a 13.5% 5* rate.  I know it’s a small sample size, just thought I’d put it out there.

    edit:  this is only for latest legends.