(Now 0 for 73) - Devs, is the system working as you intend?

amusingfoo1

Tarheelmax wrote:

jobob wrote:

Wooodd wrote:

Whilst I sympathize with your recent poor 5 luck is it fair that you have been gifted a free (assuming you could choose your cover) whilst the rest of the player-base have not? Whilst RNG is not a fair way to distribute top end rewards, it is the system we all have to work with and unless it is changed it should be adhered to, particularly by CS. For them to give away a free cover to someone who's experience has not matched the listed odds on tokens seems like opening quite a can of worms to me.

It may be a can of worms... but again, I have serious questions about how random the system actually is. You see too many streaks that are so statistically improbable, that it casts doubt on the system IMO.

But every single string of 40 pulls, no matter what it is, is just as improbable as a string of no 5*s. Most likely it's confirmation bias, because people aren't on here clamoring to comment on their 3-6 5* pulls from 40 tokens.

edit: fixed typo

Well, sure, but there's forty ways of getting one 5*, so it's 40x as likely as getting none.

Tarheelmax

chamber44 wrote:

Tarheelmax wrote:

jobob wrote:

Wooodd wrote:

Whilst I sympathize with your recent poor 5 luck is it fair that you have been gifted a free (assuming you could choose your cover) whilst the rest of the player-base have not? Whilst RNG is not a fair way to distribute top end rewards, it is the system we all have to work with and unless it is changed it should be adhered to, particularly by CS. For them to give away a free cover to someone who's experience has not matched the listed odds on tokens seems like opening quite a can of worms to me.

It may be a can of worms... but again, I have serious questions about how random the system actually is. You see too many streaks that are so statistically improbable, that it casts doubt on the system IMO.

But every single string of 40 pulls, no matter what it is, is just as improbable as a string of no 5*s. Most likely it's confirmation bias, because people aren't on here clamoring to comment on their 3-6 5* pulls from 40 tokens.

edit: fixed typo

that's simply not true. RNG is not as random as it seems, because the multitude of stories of people pulling the exact same token 3, 4, 5 times in a row has to be statistically impossible (or improbable to a nearly unmeasurable degree).

I'll ignore the recent screenshots of the person who drew four 5* JG greens in a row, because that's too crazy to comprehend and instead use a more common example. In cashing in multiple DDQ tokens, I've drawn 4 sets of healthpacks consecutively. Now, even given that each draw from the Vault increases the chance that i'll draw healthpacks slightly, what is the statistical probability that I'll actually draw 4 in a row?

Now, what's the statistical probability that i'll do it again? (because I have - at least three times, to be exact, and that's not counting the times i'll draw 3-5 sets of healthpacks while cashing in 10-20 tokens at a time)

I will admit 100% that I hate math with a passion, so maybe the odds are higher than I think. But I doubt it. I'm also not saying the solution is to rig the system so that you can't draw two of the same tokens consecutively, but it does just look like the system isn't even random but that for whatever reason, these crazy streaks are possible. To me, that's almost worse, because there's an entire class of player that will never be able to climb out of the hole that "random" forces them into (as evidenced by the fact that I have more Dumpster Fire Hulk covers than I have of Iceman and JG combined).

I don't see how it's not accurate...

Each legendary token has the same odds independant of each other. You are plucking one example out of the tens of thousands of legendary token pulls and assuming that single example means there is something wrong with the RNG.

I can tell you that if D3 posted every single legendary token pull in order, you would see the randomness you are looking for (which includes clusters).

Tarheelmax

amusingfoo1 wrote:

Tarheelmax wrote:

jobob wrote:

Wooodd wrote:

Whilst I sympathize with your recent poor 5 luck is it fair that you have been gifted a free (assuming you could choose your cover) whilst the rest of the player-base have not? Whilst RNG is not a fair way to distribute top end rewards, it is the system we all have to work with and unless it is changed it should be adhered to, particularly by CS. For them to give away a free cover to someone who's experience has not matched the listed odds on tokens seems like opening quite a can of worms to me.

It may be a can of worms... but again, I have serious questions about how random the system actually is. You see too many streaks that are so statistically improbable, that it casts doubt on the system IMO.

But every single string of 40 pulls, no matter what it is, is just as improbable as a string of no 5*s. Most likely it's confirmation bias, because people aren't on here clamoring to comment on their 3-6 5* pulls from 40 tokens.

edit: fixed typo

Well, sure, but there's forty ways of getting one 5*, so it's 40x as likely as getting none.

True, and I totally agree that it's not likely to make 40 pulls and not get a 5*, just that it isn't crazy that it happens. Or at least not really any crazier than any 40 pull string. I could post my last 40 pulls, but it happens to look random even though statistically it was improbable to pull those 40 covers, so people wouldn't ever say "wow, that's a crazy sequence."

simonsez

Wooodd wrote:

if I reply that statistically you could pull 10 million and never hit one as each pull is its own gamble with its own statistics applied then I'll be hit with the "easy for you to say with your perfect 10% pull rate" responses.

No, no one would say that. They would say you have absolutely no understanding of probability and statistics whatsoever, and you refrain from ever speaking of them again, because it's **** embarrassing.

simonsez

Tarheelmax wrote:

Yeah, if you literally pulled 1 million LTs and got none, then it would be reasonable to believe the odds aren't really 10%. But that is no where near anything that we are talking about. 40 is not the same thing as 1,000,000.

Not the point. I was addressing a specific comment, and he's standing by it, despite the exaggerated example.

simonsez

Tarheelmax wrote:

Each legendary token has the same odds independant of each other.

No, that's what we're being TOLD. A lot of evidence, as well as malenkov's project, suggest otherwise. So don't begin any post with a statement that's counter to what we're arguing about in the first place.

Wooodd

simonsez wrote:

Wooodd wrote:

if I reply that statistically you could pull 10 million and never hit one as each pull is its own gamble with its own statistics applied then I'll be hit with the "easy for you to say with your perfect 10% pull rate" responses.

No, no one would say that. They would say you have absolutely no understanding of probability and statistics whatsoever, and you refrain from ever speaking of them again, because it's tinykitty embarrassing.

Your obviously a statistician then Simonsez, enlighten me.

Is it impossible to not hit one out of 1 million or improbable?

mpqr7

I don't think that people are complaining that the code doesn't work as expected.

People are complaining that the code IS working as expected, ie that due to the way the code is constructed, there are plenty of times when they will work hard for months and not receive any reward.

In a regular game, if you played it as hard as we do for months, you would have probably beaten it by now, not be struggling to receive a single 1/13th of a power-up.

simonsez

Tarheelmax wrote:

True, and I totally agree that it's not likely to make 40 pulls and not get a 5*, just that it isn't crazy that it happens. Or at least not really any crazier than any 40 pull string. I could post my last 40 pulls, but it happens to look random even though statistically it was improbable to pull those 40 covers, so people wouldn't ever say "wow, that's a crazy sequence."

You people who make that kind of statement are missing the point. If you pull a GT, SL, Thing, Antman, and Elektra, yes, that's a unique sequence, just as unique as any other. BUT BEFORE YOU MAKE THOSE PULLS, no one was entertaining the hypothesis, "I bet you pull a GT, SL, Thing, Antman, Elektra". On the other hand, the notion that people's pulls cluster and show a preponderance of duuplication is something WE ALREADY SUSPECT PRIOR TO THE PULLS. It is not valid to look at the above sequence after the fact and claim it's an unlikely outcome. On the other hand, if you state beforehand, "I believe pull results cluster", and you end up with GT, GT, SL, SL, Thing, it is completely valid to point out the unlikely probability of drawing two pairs out of 5 pulls, because clustering was your stated hypothesis before any pulls happened.

I hope this is helpful to someone.

simonsez

Wooodd wrote:

Is it impossible to not hit one out of 1 million or improbable?

It depends on how you define "impossible". Most people with an ounce of common sense would consider a decimal point, followed by page full of zeroes and a 1 to be the same as "0", but you're always going to find a Lloyd who will look at that and insist, "So, you're saying there's a chance!".

Tarheelmax

simonsez wrote:

Tarheelmax wrote:

True, and I totally agree that it's not likely to make 40 pulls and not get a 5*, just that it isn't crazy that it happens. Or at least not really any crazier than any 40 pull string. I could post my last 40 pulls, but it happens to look random even though statistically it was improbable to pull those 40 covers, so people wouldn't ever say "wow, that's a crazy sequence."

You people who make that kind of statement are missing the point. If you pull a GT, SL, Thing, Antman, and Elektra, yes, that's a unique sequence, just as unique as any other. BUT BEFORE YOU MAKE THOSE PULLS, no one was entertaining the hypothesis, "I bet you pull a GT, SL, Thing, Antman, Elektra". On the other hand, the notion that people's pulls cluster and show a preponderance of duuplication is something WE ALREADY SUSPECT PRIOR TO THE PULLS. It is not valid to look at the above sequence after the fact and claim it's an unlikely outcome. On the other hand, if you state beforehand, "I believe pull results cluster", and you end up with GT, GT, SL, SL, Thing, it is completely valid to point out the unlikely probability of drawing two pairs out of 5 pulls, because clustering was your stated hypothesis before any pulls happened.

I hope this is helpful to someone.

I don't think I'm missing the point. And I also think you have a clear grasp on the math, so please don't mistake my responses as looking down on you, because I don't.

I don't disagree with your point that there may be something messing with the true RNG. I'm just saying that the evidence is a few people plucking clustered results and saying there is a problem with the RNG. It's confirmation bias at its best.

What I am saying is people DO go into pulling 40 tokens expecting a 5*. And I am also saying that its obvious people do not go into it expecting to pull duplicates or any random assortment of 4*s. So coming out of it, people notice a couple things: 1) lack of 5*s because they are expecting them and 2) clusters, because they aren't expected and stand out (since they don't look random). This again leads to confirmation bias, because you forget about all the other seemingly normal pulls.

So, unfortunately we are left with poor evidence based on a small number of people (relative to the player base) raging about bad luck. Then that is being extrapolated into an assumption that it is the norm, and therefore the RNG process is not working properly.

On the other hand, we have poor customer service that is replying with a canned response of "It's working as intended" without any evidence to back them up.

So, right now, I'm siding with the developers and assuming that they have looked at the numbers and that they flesh out on a game-wide basis to line up with the listed odds.

Wooodd

simonsez wrote:

Wooodd wrote:

Is it impossible to not hit one out of 1 million or improbable?

It depends on how you define "impossible". Most people with an ounce of common sense would consider a decimal point, followed by page full of zeroes and a 1 to be the same as "0", but you're always going to find a Lloyd who will look at that and insist, "So, you're saying there's a chance!".

Then I guess you would have to call me Lloyd.

Any whilst I did use the phrase "no matter how many times" I didn't directly refer to a sample size of 1,000,000.

The highest number of misses I have seen people complain about in a row is <100, there may be people but I haven't seen any posts myself.

Now 0/100 whilst highly unlikely is nowhere near impossible with a 10% pull rate.

What we need to do if we are to provide any evidence that the system in inherently cheating people is get the pull rates of everyone. As has been said in the past, there will be people hitting well above the 10% who aren't on here complaining about it who's results are not being factored into this discussion.

Eddiemon

Pongie wrote:

Coltsjustwin wrote:

I have seen some "top" players post their results from LT pulls and they seem to be incredibly lucky by getting a near 25-33% 5* pull rate
Might have been just that day but damn...I don't hear anybody else getting that lucky.

Random is random. With a 10% chance of hitting a token large winning streaks are very uncommon where large cold streaks are not. And people tend to overexaggerate adverse events and take for granted positive events.

I have a theory about this. The vets and whales that open heaps of tokens immediately after a character release are close to, if not opening the majority of token pool. Eg they open 70 tokens, chances are its 70 out of 100 tokens being opening at the time. Therefore they get a more even distribution. For the rest who are opening 70 tokens over an extended period, this is just a fraction of how many tokens being opened. It could be in the thousands if not more. In this case the distribution is more random.

It's a nice theory but makes no mathematical sense, even if such pools existed rather than just being randomly generated numbers.

First you are assuming that the only person opening from that pool at that time is one whale. If two whales are then one could have sensational luck and the other truly terrible luck. You also need to start a new pool because if you get the tail end of one pool and teh start of another you get no guarantees.

If a whale takes 2/3 of a pool then they will, on average, take 2/3 of the 5 stars from that pool. The odds have not changed in the slightest. And whether you draw your 5* in the middle of a whale's rampage or at an idle moment you're still in with a 10% chance on average. A whale may have taken 70 tokens from the pool which may have left more or less than 3 5 stars in the remainder of the pool, which may increase or decrease your odds on this particular draw, but averaged out over 100 draws you would find that your lucky and unlucky pools would have averaged themselves out to that 10% chance.

I know it feels strange, but say you are taking the last token from a pool. If it were a 5* token your odds would be 100% of drawing a 5 star while if it were a 4*. But your odds of finding that first pool is 10% while the second pool is likely to exist 90% of the time.

Not that I believe they use any kind of 'pool' system, because some of the past exploits would have been very easy to deal with in a pool system, but much harder in a random number generated system.

Eddiemon

simonsez wrote:

You people who make that kind of statement are missing the point. If you pull a GT, SL, Thing, Antman, and Elektra, yes, that's a unique sequence, just as unique as any other. BUT BEFORE YOU MAKE THOSE PULLS, no one was entertaining the hypothesis, "I bet you pull a GT, SL, Thing, Antman, Elektra". On the other hand, the notion that people's pulls cluster and show a preponderance of duuplication is something WE ALREADY SUSPECT PRIOR TO THE PULLS. It is not valid to look at the above sequence after the fact and claim it's an unlikely outcome. On the other hand, if you state beforehand, "I believe pull results cluster", and you end up with GT, GT, SL, SL, Thing, it is completely valid to point out the unlikely probability of drawing two pairs out of 5 pulls, because clustering was your stated hypothesis before any pulls happened.

I hope this is helpful to someone.

I hope it isn't.

Clustering and a preponderance of duplication is a sign of randomness. In a room of just 23 people there is a 50% chance that 2 of them share the same birthday. In a room of 75 people there is a 99.9% chance of 2 people sharing the same birthday. If random meant equal distribution then these things shouldn't be true.

Even better when you take the pessimistic view and discard any results that disagree with you. So when the results don't cluster you are pleasantly surprised and don't post on here about how you were wrong, but when they do cluster you take that as a sign that you knew something was wrong and that your odds of predicting a common occurrance is phenominal.

You predicted 'clustering'. Which could be two duplicates, or three duplicates, or 2 pair, or a full house or a poker of covers. Whatever. But that whole set has a far higher probability than the probability of 'two pairs', which you have then specifically calculated even though that wasn't your prediction at all.

In a 5 cover pull you have about a 1 in 3 chance of a duplicate somewhere given the current stable of 4 stars. If you were making the prediction when there were less than 27 4 stars on the market those odds only go up.

jobob

At some point, "mathematically possible" doesn't mean much, as you cross a line of "practically impossible."

It is mathematically possible that I will hit the next 3 powerballs, buy my own island, and as I am swimming along the coast, be attacked by a shark and only be saved when the two of us are struck by lightning. All that has a >0% chance of happening... But is practically impossible.

So when I make a pull that has a 0.000053% chance (if I calculated right) of occurring randomly, I am skeptical.

It's Occam's razor... What's the simpler theory, that all these statistically improbable streaks are random, or that there is indeed something else that factors in sometimes.

Tarheelmax

jobob wrote:

At some point, "mathematically possible" doesn't mean much, as you cross a line of "practically impossible."

It is mathematically possible that I will hit the next 3 powerballs, buy my own island, and as I am swimming along the coast, be attacked by a shark and only be saved when the two of us are struck by lightning. All that has a >0% chance of happening... But is practically impossible.

So when I make a pull that has a 0.000053% chance (if I calculated right) of occurring randomly, I am skeptical.

It's Occam's razor... What's the simpler theory, that all these statistically improbable streaks are random, or that there is indeed something else that factors in sometimes.

This issue is, before you started pulling, you literally must make a string of pulls with that probability at some point. I don't know if that's 5 pulls, 10 pulls or 100 pulls. But at some point, the chances of pulling the string you pulled will be 0.000053%.

You are only skeptical about it if you don't like the results of whatever your string actually is.

Edit:

And in this case, I would say the Occam's razor approach would be to believe the Devs have used a very easy to implement RNG open source code for the game instead of some weird elaborate system for token pulls.

jobob

simonsez wrote:

Tarheelmax wrote:

True, and I totally agree that it's not likely to make 40 pulls and not get a 5*, just that it isn't crazy that it happens. Or at least not really any crazier than any 40 pull string. I could post my last 40 pulls, but it happens to look random even though statistically it was improbable to pull those 40 covers, so people wouldn't ever say "wow, that's a crazy sequence."

You people who make that kind of statement are missing the point. If you pull a GT, SL, Thing, Antman, and Elektra, yes, that's a unique sequence, just as unique as any other. BUT BEFORE YOU MAKE THOSE PULLS, no one was entertaining the hypothesis, "I bet you pull a GT, SL, Thing, Antman, Elektra". On the other hand, the notion that people's pulls cluster and show a preponderance of duuplication is something WE ALREADY SUSPECT PRIOR TO THE PULLS. It is not valid to look at the above sequence after the fact and claim it's an unlikely outcome. On the other hand, if you state beforehand, "I believe pull results cluster", and you end up with GT, GT, SL, SL, Thing, it is completely valid to point out the unlikely probability of drawing two pairs out of 5 pulls, because clustering was your stated hypothesis before any pulls happened.

I hope this is helpful to someone.

Exactly. You are just as likely to shuffle a deck of cards into sequential order as any other specific combination.

It's like saying everyone should hit the powerball, because my number was just as improbable as the winner's.

Unknown

My 2 cents:

Code: A lot of people in this thread probably don't code/develop as a profession. Generally speaking, it's a bit silly to say the devs wrote some convoluted code to obfuscate their random pulls. Most likely, they will code this in the simplest manner they can think of, which is probably dividing the available pool up in some distribution and just pulling a random number to get it. There's no guarantee that their code works as intended, but you have to put some faith in their internal testing.

Math: Assuming the above is true, there are definitely some people who are going to get the short end of the stick. That's just that.

My Solution: Add a 10-pack of Legendaries, that would cost 250/200 CPs, which when pulled, will give exactly 9 4* covers and 1 5* cover. This is a simple approach which would not change the percentages like some of the other suggestions.

Tarheelmax

jobob wrote:

simonsez wrote:

Tarheelmax wrote:

True, and I totally agree that it's not likely to make 40 pulls and not get a 5*, just that it isn't crazy that it happens. Or at least not really any crazier than any 40 pull string. I could post my last 40 pulls, but it happens to look random even though statistically it was improbable to pull those 40 covers, so people wouldn't ever say "wow, that's a crazy sequence."

You people who make that kind of statement are missing the point. If you pull a GT, SL, Thing, Antman, and Elektra, yes, that's a unique sequence, just as unique as any other. BUT BEFORE YOU MAKE THOSE PULLS, no one was entertaining the hypothesis, "I bet you pull a GT, SL, Thing, Antman, Elektra". On the other hand, the notion that people's pulls cluster and show a preponderance of duuplication is something WE ALREADY SUSPECT PRIOR TO THE PULLS. It is not valid to look at the above sequence after the fact and claim it's an unlikely outcome. On the other hand, if you state beforehand, "I believe pull results cluster", and you end up with GT, GT, SL, SL, Thing, it is completely valid to point out the unlikely probability of drawing two pairs out of 5 pulls, because clustering was your stated hypothesis before any pulls happened.

I hope this is helpful to someone.

Exactly. You are just as likely to shuffle a deck of cards into sequential order as any other specific combination.

It's like saying everyone should hit the powerball, because my number was just as improbable as the winner's.

This isn't what I'm saying at all. I'm saying that if you shuffle a deck of cards, draw the top card, place that card back, re-shuffle and repeat the process 3 times (4 draws total), then you have the exact same likelihood of doing the following:

Pull 4 Aces
Pull 4 Twos
Pull a 2, 7, Jack, Ace

Now, if you have 10,000 people do this, the only people that are likely to really speak up are the ones that had the horrible experience of pulling 4 Twos. And just because a small minority of people start yelling that they only pulled 2s, doesn't mean the deck was rigged. It means they were expecting a "random-looking" or a "lucky" outcome but didn't get it, so they complain about it.

Then you have people coming along and saying "hey look, that small minority of people must be representative of the entire 10,000; that deck of cards is RIGGED."

jobob

Tarheelmax wrote:

This isn't what I'm saying at all. I'm saying that if you shuffle a deck of cards, draw the top card, place that card back, re-shuffle and repeat the process 3 times (4 draws total), then you have the exact same likelihood of doing the following:

Pull 4 Aces
Pull 4 Twos
Pull a 2, 7, Jack, Ace

Now, if you have 10,000 people do this, the only people that are likely to really speak up are the ones that had the horrible experience of pulling 4 Twos. And just because a small minority of people start yelling that they only pulled 2s, doesn't mean the deck was rigged. It means they were expecting a "random-looking" or a "lucky" outcome but didn't get it, so they complain about it.

Then you have people coming along and saying "hey look, that small minority of people must be representative of the entire 10,000; that deck of cards is RIGGED."

And what I am saying is that you can't put "pull a 2, 7, Jack, Ace" in the same list as pulling 4 aces or 2s, because no one cares about pulling a 2, 7, Jack, Ace. You need to replace that with basically ALL the combinations that include one Ace, because those are viewed as being about the same. And if you get people pulling a distinct set nearly as frequently as the sum of all the other possible sequences... something is very wrong.