**Mastery Loadouts**
Due to issues related to the release of Mastery Loadouts, the "free swap" period will be extended.
The new end date will be May 1st.

Champ pool getting too big

Spurgeon14Spurgeon14 Posts: 1,665 ★★★★
As more and more champions keep getting added to the crystals, is there any hope of the regular crystals being split up or something? I'm tired of getting the same 4*s over and over...does this bother anyone else?
«13

Comments

  • V1PER1987V1PER1987 Posts: 3,474 ★★★★★
    Drooped2 wrote: »
    This complaint doesnt check out with math. Right now you have a .89 chance per champ ish

    Splitting it although removes some chamois from it ups your chances of pulling each.

    Ie 24 per crystal is 4 percent

    Why doesn’t that check out with math?
  • Darkstar4387Darkstar4387 Posts: 2,145 ★★★
    edited January 2019
    Some people just dupe the same champions over and over and over, I have many champions that are Max sig level or at high sig level but there is a chance to pull new champions as well it just depends on rng and luck.

    I doubt shrinking the pool would help any , unless they add more crystals and separate them by class for 3*,4*5* but even then you could keep getting the same ones over and over
  • ScottryanScottryan Posts: 467 ★★★
    "The pool is getting so big that i keep pulling the same champ over and over"...what?
  • GamerGamer Posts: 10,158 ★★★★★
    Zuko_ILC wrote: »
    Spurgeo14 wrote: »
    As more and more champions keep getting added to the crystals, is there any hope of the regular crystals being split up or something? I'm tired of getting the same 4*s over and over...does this bother anyone else?

    Would be nice if they split the crystals by years 2015, 2016, 2017, 2018 or at least combined 2 years

    God idies but combine with 2 year.
  • mostlyharmlessnmostlyharmlessn Posts: 1,387 ★★★★
    edited January 2019
    Spurgeo14 wrote: »
    As more and more champions keep getting added to the crystals, is there any hope of the regular crystals being split up or something? I'm tired of getting the same 4*s over and over...does this bother anyone else?

    It is very annoying, but it's the result of pRNG algorythm which Kabam uses. Low quality pRNG algorithms produce the kind of "Batchy" results we tend to see. Linear congruential generator (LCG), the one that Java uses by default, which while fast, has been known for decades to be pretty poor quality and produce the kind of results we see. (I do not know if this is the actual algorithm used, but it sure does feel like it.
  • belli300belli300 Posts: 704 ★★★
    I’m pretty sure cyclop is statistically that most likely champ you’ll pull he’s super lame and there are 2 of them plus nobody gets excited about cyclops. Not even cyclops would want to pull himself
  • Cranmer00Cranmer00 Posts: 527 ★★
    I think he’s saying

    He wants to stop getting trash champions over and over that serve no purpose in the game, thankfully AW diversity plays a roll in making some ok champions worthwhile, but there is no use for 25-40% of the champions
  • zarcorzarcor Posts: 28
    Maybe a good idea would be to have a 4* feature (just like we have with 5* ad 6*)
  • OrangieOrangie Posts: 76
    I don’t want any more 4 stars, they all trash
  • DNA3000DNA3000 Posts: 18,657 Guardian
    Spurgeo14 wrote: »
    As more and more champions keep getting added to the crystals, is there any hope of the regular crystals being split up or something? I'm tired of getting the same 4*s over and over...does this bother anyone else?

    It is very annoying, but it's the result of pRNG algorythm which Kabam uses. Low quality pRNG algorithms produce the kind of "Batchy" results we tend to see. Linear congruential generator (LCG), the one that Java uses by default, which while fast, has been known for decades to be pretty poor quality and produce the kind of results we see. (I do not know if this is the actual algorithm used, but it sure does feel like it.

    No pRNG in general use by any modern language has sufficiently bad sequential correlation that a human being could detect it with their eyeballs. I'm not familiar with Java's specific choice of parameters, but as there are very good ones with minimal correlation it would seem illogical they would go out of their way to pick a bad set.

    But again, even *horrible* pRNGs are unlikely to generate sequences humans can see. Especially in a multiplayer game where literally millions of people are doing things that require random numbers generated.

    Based on the literature I"ve read, Java's current choice for parameters suffers from tupling bins, aka the problem of "sets falling in planes" but it is otherwise reasonably random. This means it does not suffer from the correlation problem you describe: it is not "batchy." Instead, it doesn't distribute short sequences of random numbers randomly. To put it another way, to catch this implementation in the act of doing something non-random, you would have to be able to tell that the sequence Cyclops/X-23/MODOK occurs less often than MODOK/Wolverine/MsMarvel. No one does that analysis.

    This is separate from the fact that the Java implementation is not generally used in server-side applications. The generators you'd use from something like the Microsoft or GNU libraries are much stronger.
  • IronDevilIronDevil Posts: 9
    It would be nice if crystals are splitted class wise. That way anyone hoping for a particular champ can open that particular class crystal
  • Darkstar4387Darkstar4387 Posts: 2,145 ★★★
    Scottryan wrote: »
    "The pool is getting so big that i keep pulling the same champ over and over"...what?

    Means duping the same champions over and over and over again, and again i know how op feels since I have a lot of subpar champions at Max sig like Groot, cyclops, Netflix Daredevil, winter solider etc

    I also have a lot of champions at sig 60-80 some not so good like civil warrior, both Magneto's and, vulture, hulk buster and many more.

    It's the fact that even with a pull this big you coul end up never pulling certain champions and just pulling the same ones over and over.

    I am missing a number of champions in the 4* pool and have a habit of duping some in quick succession right after I just pulled them.
  • SuvenduSuvendu Posts: 177
    Spurgeo14 wrote: »
    As more and more champions keep getting added to the crystals, is there any hope of the regular crystals being split up or something? I'm tired of getting the same 4*s over and over...does this bother anyone else?

    Kabam might introduce 4* featured pool like 5* featured.. Like 3k shards for one featured 4*
  • St333lllSt333lll Posts: 278
    Either way, whether the math works or not, the pool is getting huge. There are champs that I will never get the luxury of even seeing. In all my Years of this game I've never pulled a 4 star Scarlet Witch.
  • Midknight007Midknight007 Posts: 764 ★★★
    What might be a nice idea is a generic class specific gem, or one that puts 2-3 classes only.
  • WorknprogressWorknprogress Posts: 7,233 ★★★★★
    Why would they split stuff into classes? It wouldn't make any sense to even give random AG and rank up gems as rewards at that point. Then anyone could just target a specific class for whatever rewards they got.
  • MavRCK_MavRCK_ Posts: 469 ★★★
    edited January 2019
    Mathematically, the pool has grown faster than the rate we can reasonably earn 5* shards so the OPs comment when interpreted comparing as these two factors is reasonable. The probability of getting the champion(s) you want is extremely rare now. <1% from a basic crystal. 4% from a feature crystal that cost 50% more than a basic.
  • SDPSDP Posts: 1,622 ★★★★
    Dude. Eventually, you will open multiple 4* a week. You have to suffer through it like the rest of us. Also, your post makes no sense.
  • MaatManMaatMan Posts: 958 ★★★
    Why would they split stuff into classes? It wouldn't make any sense to even give random AG and rank up gems as rewards at that point. Then anyone could just target a specific class for whatever rewards they got.

    well they really should be split in some way or another. i was merely offering a suggestion.
    and a good one for us TBH.

    u got a better one?
  • WorknprogressWorknprogress Posts: 7,233 ★★★★★
    MaatMan wrote: »
    Why would they split stuff into classes? It wouldn't make any sense to even give random AG and rank up gems as rewards at that point. Then anyone could just target a specific class for whatever rewards they got.

    well they really should be split in some way or another. i was merely offering a suggestion.
    and a good one for us TBH.

    u got a better one?

    Think it's fine as is personally. Want a reduced pool? Open featured crystals
  • DrZolaDrZola Posts: 8,543 ★★★★★
    DNA3000 wrote: »
    Spurgeo14 wrote: »
    As more and more champions keep getting added to the crystals, is there any hope of the regular crystals being split up or something? I'm tired of getting the same 4*s over and over...does this bother anyone else?

    It is very annoying, but it's the result of pRNG algorythm which Kabam uses. Low quality pRNG algorithms produce the kind of "Batchy" results we tend to see. Linear congruential generator (LCG), the one that Java uses by default, which while fast, has been known for decades to be pretty poor quality and produce the kind of results we see. (I do not know if this is the actual algorithm used, but it sure does feel like it.

    No pRNG in general use by any modern language has sufficiently bad sequential correlation that a human being could detect it with their eyeballs. I'm not familiar with Java's specific choice of parameters, but as there are very good ones with minimal correlation it would seem illogical they would go out of their way to pick a bad set.

    But again, even *horrible* pRNGs are unlikely to generate sequences humans can see. Especially in a multiplayer game where literally millions of people are doing things that require random numbers generated.

    Based on the literature I"ve read, Java's current choice for parameters suffers from tupling bins, aka the problem of "sets falling in planes" but it is otherwise reasonably random. This means it does not suffer from the correlation problem you describe: it is not "batchy." Instead, it doesn't distribute short sequences of random numbers randomly. To put it another way, to catch this implementation in the act of doing something non-random, you would have to be able to tell that the sequence Cyclops/X-23/MODOK occurs less often than MODOK/Wolverine/MsMarvel. No one does that analysis.

    This is separate from the fact that the Java implementation is not generally used in server-side applications. The generators you'd use from something like the Microsoft or GNU libraries are much stronger.

    What pRNG program do you believe they use? I’d be curious to know if you gained insight from your stints helping vet content in Betas and elsewhere, or if you’ve been able to glean anything. As far as I am aware, I haven’t seen anything that describes the actual process used in game to generate things like crystal pulls (presumably fight pRNG would have to be pretty simple, since the timing demands speed and simplicity). But I’m no longer heavily involved in generating simulations and models, so I may be behind the curve.

    What @mostlyharmlessn is saying is qualitatively different from the usual tinfoil hat rubbish that suggests the game is secretly working to hose the player base. What I understand him to say is that the pRNG algorithm, combined with the way seed numbers are generated and refreshed, can tend to produce the same result more frequently than a better randomizer would. “Better” doesn’t have to mean perfectly random, but it can mean something akin to less flawed or (using his terms) less inclined to produce batchy results (a la the four 5* Punisher 2099’s your alliance feed shows mates all pulling at approximately the same time).

    That’s a fair argument—it isn’t the stuff of fever dreams, but rather an inquiry into the actual mechanics of the system and whether they perform effectively. Right now it’s all a black box—as far as we know, there’s no audit or analysis to ensure sufficient randomness. We are required to take it as an article of faith that it works like we are told it works.

    Dr. Zola
  • mostlyharmlessnmostlyharmlessn Posts: 1,387 ★★★★
    DNA3000 wrote: »
    Spurgeo14 wrote: »
    As more and more champions keep getting added to the crystals, is there any hope of the regular crystals being split up or something? I'm tired of getting the same 4*s over and over...does this bother anyone else?

    It is very annoying, but it's the result of pRNG algorythm which Kabam uses. Low quality pRNG algorithms produce the kind of "Batchy" results we tend to see. Linear congruential generator (LCG), the one that Java uses by default, which while fast, has been known for decades to be pretty poor quality and produce the kind of results we see. (I do not know if this is the actual algorithm used, but it sure does feel like it.

    No pRNG in general use by any modern language has sufficiently bad sequential correlation that a human being could detect it with their eyeballs. I'm not familiar with Java's specific choice of parameters, but as there are very good ones with minimal correlation it would seem illogical they would go out of their way to pick a bad set.

    But again, even *horrible* pRNGs are unlikely to generate sequences humans can see. Especially in a multiplayer game where literally millions of people are doing things that require random numbers generated.

    Based on the literature I"ve read, Java's current choice for parameters suffers from tupling bins, aka the problem of "sets falling in planes" but it is otherwise reasonably random. This means it does not suffer from the correlation problem you describe: it is not "batchy." Instead, it doesn't distribute short sequences of random numbers randomly. To put it another way, to catch this implementation in the act of doing something non-random, you would have to be able to tell that the sequence Cyclops/X-23/MODOK occurs less often than MODOK/Wolverine/MsMarvel. No one does that analysis.

    This is separate from the fact that the Java implementation is not generally used in server-side applications. The generators you'd use from something like the Microsoft or GNU libraries are much stronger.

    The problem which I have stated, the batchy results is exactly what we are observing. I didn't say they were using Java's built in one, I do not have access to their code to see what they are using.

    Those that have been tracking their own pulls see that they get stuck in a batch. I've pulled JF/Drax/WS/RR consistently where every other 5* basic crystal pull is one of those. That is the kind of results which a LCG type algorithm produces, and exactly what the OP is talking about, exactly what we see when we open 5 crystals and get 2 of the same champ in that batch.

    With the number of champs in the pool at this point we should be complaining that we can't ever pull a dupe.
  • DrZolaDrZola Posts: 8,543 ★★★★★
    https://forums.playcontestofchampions.com/en/discussion/comment/706565#Comment_706565

    Related thread about size of basic pool. Should probably be joined with this one.

    Dr. Zola
Sign In or Register to comment.