LT pulls...[Pull Streaks]

Buret0

0/30 streak ends with a Surfer Blue to make it 1/30... or 1/1?

OneLastGambit

Since the draw rate is per pull the odds of pulling any sequence at all is equal. Not higher or lower.

Since there are an infinite number of tokens in the pool from which you can draw it's both conceivable and expected that you could draw the same token multiple times in a row. It sucks.

This is the reason why we love vaults so much since every bad pull reduces the odds of a bad pull next time as the tokens are finite.

The answer to this conundrum is therefore to have a legendary vault.

ClydeFrog76

I've only opened around a dozen LTs and I just pulled my SECOND Phoenix green, my only 5* character.

I would love to know the odds of that

(Would've preferred an XFW black tbh, mine is next to useless at 5/2/2...sigh)

simonsez

ClydeFrog76 wrote:

I've only opened around a dozen LTs and I just pulled my SECOND Phoenix green, my only 5* character.

I would love to know the odds of that

Close to 7% (the odds of getting any two 5s in 12 pulls, not phoenix green specifically)

simonsez

OneLastGambit wrote:

Since the draw rate is per pull the odds of pulling any sequence at all is equal.

You are correct, but if you're saying that if you're looking at two sequences, one being:

XF, GT, KP, XDP, GT, Falcap, Carnage, Cyk, IW, KP
and the other being
XF, XF, XF, XF, XF, XF, XF, XF, XF, XF

and the latter doesn't make you question the randomness of pulls "because all sequences are equal", I have to insist that you find yourself a profession that has nothing at all to do with interpreting data.

Buret0

ClydeFrog76 wrote:

I've only opened around a dozen LTs and I just pulled my SECOND Phoenix green, my only 5* character.

I would love to know the odds of that

About one in 2,700 to get two of the same Phoenix cover in two consecutive pulls, so an 11 in 2,700 chance? Does that make sense?

1/(0.0333333*0.0333333/3) per chance?

Buret0

simonsez wrote:

OneLastGambit wrote:

Since the draw rate is per pull the odds of pulling any sequence at all is equal.

You are correct, but if you're saying that if you're looking at two sequences, one being:

XF, GT, KP, XDP, GT, Falcap, Carnage, Cyk, IW, KP
and the other being
XF, XF, XF, XF, XF, XF, XF, XF, XF, XF

and the latter doesn't make you question the randomness of pulls "because all sequences are equal", I have to insist that you find yourself a profession that has nothing at all to do with interpreting data.

10 is a pretty small sample size... you would probably start to ask questions after seeing the same result four or five times in a row and become pretty skeptical at 10, but if others can confirm their sequences seem more random you might just be looking at an outlier.

I mean, going 0/30 on 5*s when the odds of pulling a 5* doubled from the launch and pulling significantly less? That made me question the randomness of the pulls (or at least the stated odds), even though it was a pretty low sample size... you would expect to see about three 5*s in a normal sequence... three fewer isn't to the point where I'm crying foul.

I mean, I've sat at a roulette table that had every number on the past spins board come up red. That's an awful lot of coin flips coming up heads, but it doesn't make me believe the odds of that table's next spin are any different than approximately 50/50. We are betting on individual spins of the wheel, individual flips of the coin, individual openings of the packs. Just because I saw the last 16 spins came up red, doesn't mean I'm willing to put $1,000 on the next 16 spins coming up black in order to catch up.

simonsez

Buret0 wrote:

10 is a pretty small sample size... you would probably start to ask questions after seeing the same result four or five times in a row and become pretty skeptical at 10

Do you understand that the odds of getting the same character 10x in a row would be one in 32 trillion? It doesn't matter that it's 10 pulls. It's still one in 32 trillion. The occurrence of such an event should make you far more than "pretty skeptical".

simonsez

Buret0 wrote:

Just because I saw the last 16 spins came up red, doesn't mean I'm willing to put $1,000 on the next 16 spins coming up black in order to catch up.

No, no, no, that's not what anyone's saying. If the last 16 spins came up red, I'm saying you should consider leaving that casino.

Buret0

simonsez wrote:

Buret0 wrote:

10 is a pretty small sample size... you would probably start to ask questions after seeing the same result four or five times in a row and become pretty skeptical at 10

Do you understand that the odds of getting the same character 10x in a row would be one in 32 trillion? It doesn't matter that it's 10 pulls. It's still one in 32 trillion. The occurrence of such an event should make you far more than "pretty skeptical".

1 in 3 trillion actually.

The first character is excluded in your sample if same character. If it is getting a specific character ten times it would be 32 trillion.
However, if you drew XF nine times, the odds of pulling an XF on the tenth try is still 1/24.

Darknes21

I wish I had the same odds in the powerball as I do pulling the same character! Give me a break!

OneLastGambit

simonsez wrote:

OneLastGambit wrote:

Since the draw rate is per pull the odds of pulling any sequence at all is equal.

You are correct, but if you're saying that if you're looking at two sequences, one being:

XF, GT, KP, XDP, GT, Falcap, Carnage, Cyk, IW, KP
and the other being
XF, XF, XF, XF, XF, XF, XF, XF, XF, XF

and the latter doesn't make you question the randomness of pulls "because all sequences are equal", I have to insist that you find yourself a profession that has nothing at all to do with interpreting data.

You have misquoted me there, I did stipulate that in this case you are drawing from an infinite token pool where all options are available. In your example, rather like my quote, you made the parameters finite and therefore have more determined and weighted odds.

This rng system has an inifinite pool of tokens from which to draw so one draw does not affect the next draw, so every single possible sequence is also a probable sequence.

I'm not disagreeing that this system is horrible, none of us like it I was only pointing out that it is working perfectly as it should.

Buret0

All I know is, some people must be doing pretty well at drawing those covers. In Lightning Round last night I was hit by multiple people with championed 5*s with 4-6 covers.

And they must have infinite ISO (or just a massive amount), because looking at their rosters, the way they did it seems to have been grabbing a ton of LTs from championing 50+ 3 and 4 stars and adding that first cover.

For most of us, getting the 5 Million ISO we need to finish maxing out existing characters is a daunting enough task... I can't imagine how much ISO some of these people must have purchased.

Marvelfamily

I don't think that there is a failure in the pull/drop system. Maybe it's the "blue car syndrome" (you like to buy a blue car and you see tons of blue cars on the street)... Sometimes the luck has its own mind. I'd never send a request ticket for that, what did you expect?

'oh, we're changing that, what kind of cover do you prefer?' O.o

Be happy with your LT and make the best of it.

Unknown

simonsez wrote:

Buret0 wrote:

10 is a pretty small sample size... you would probably start to ask questions after seeing the same result four or five times in a row and become pretty skeptical at 10

Do you understand that the odds of getting the same character 10x in a row would be one in 32 trillion? It doesn't matter that it's 10 pulls. It's still one in 32 trillion. The occurrence of such an event should make you far more than "pretty skeptical".

The odds of pulling XF, GT, KP, XDP, GT, Falcap, Carnage, Cyk, IW, KP in that order are also one in 32 trillion, yet you're not exactly calling that a weird, unlikely circumstance...

Stax the Foyer

OneLastGambit wrote:

simonsez wrote:

OneLastGambit wrote:

Since the draw rate is per pull the odds of pulling any sequence at all is equal.

You are correct, but if you're saying that if you're looking at two sequences, one being:

XF, GT, KP, XDP, GT, Falcap, Carnage, Cyk, IW, KP
and the other being
XF, XF, XF, XF, XF, XF, XF, XF, XF, XF

and the latter doesn't make you question the randomness of pulls "because all sequences are equal", I have to insist that you find yourself a profession that has nothing at all to do with interpreting data.

You have misquoted me there, I did stipulate that in this case you are drawing from an infinite token pool where all options are available. In your example, rather like my quote, you made the parameters finite and therefore have more determined and weighted odds.

This rng system has an inifinite pool of tokens from which to draw so one draw does not affect the next draw, so every single possible sequence is also a probable sequence.

I'm not disagreeing that this system is horrible, none of us like it I was only pointing out that it is working perfectly as it should.

You can't prove that it's working perfectly. We understand that any given sequence is equally likely for a purely random selection, that's rudimentary and we're not dumb. However, clumping in the results may be indicative that something's hinky with the RNG that generates the pulls.

This is different than a roulette wheel example above, which is a well-known device that's been operating with the results in full view of the public for an extended period of time. When viewed in that context, you can trust that odd streaks are the outliers.

This is a black box where you only see your own results, and you can see the effect of pulls on other people's rosters. All we have is anectodal evidence from other players, but with a fairly small player base (relative to the pull odds here), there's a lot of anecdotal reports of clumping. There's selection bias there, as well as unverifiable claims, but it can't be completely overlooked.

With the constant parade of bugs and broken features, do you really trust that the token draw system has been operating correctly throughout the lifetime of the game?

Dwarfsteel

I've been thinking about peoples pull rates and what not and it got me to thinking about how exactly do the numbers get generated? From what I can tell, the token pulls are generated client side, right? There must be some kind of algorithm that determines the number as we all know there really is no such thing as true random in computers. So is it possible (not really a programmer type) that the algorithm is pulling some number or numbers from the clients computer, phone, tablet whatever that could bias the results. As say for simplicity's sake the algorithm is something like (3a + 4 = token reward) and say that "a" is the time of day, the last phone number in your call history, the devices' serial number, etc. Again, I understand that this is far too simplistic and I'm sure the algorithm(s) change, alternate, or some such, but I can't help thinking that this clumping that occurs isn't as coincidental as pure logic would dictate.

Yes, yes, confirmation bias and all that fun psychological babble, but we're not really talking about a truly random system. There are covers that seemingly drop more often for certain people. I've mentioned it elsewhere and I'm certain the global pull rates fall within acceptable ranges of posted percentages, but perhaps they're not really being distributed to the degree one would expect in a truly random environment. Any validity to this line of thought?

chamber44

one of my first (and only, for a long time) purchases was enough HP to buy a 10-pack. I drew 6 bagmans, prompting an email to Customer Service (which ended, as you might expect, with their assurance that the system worked correctly).

Then, the all-too-common situation where one pulls three of anything in a row out of the vault, whether it's three Ares covers, three health packs, or what.

Clumping seems to be a thing. Whether it's something the devs can (or want to) fix in another, but i wonder if we would accept a system that guaranteed that we didn't draw the same thing two times in a row

hsk808

Dwarfsteel wrote:

I've been thinking about peoples pull rates and what not and it got me to thinking about how exactly do the numbers get generated? From what I can tell, the token pulls are generated client side, right? There must be some kind of algorithm that determines the number as we all know there really is no such thing as true random in computers. So is it possible (not really a programmer type) that the algorithm is pulling some number or numbers from the clients computer, phone, tablet whatever that could bias the results. As say for simplicity's sake the algorithm is something like (3a + 4 = token reward) and say that "a" is the time of day, the last phone number in your call history, the devices' serial number, etc. Again, I understand that this is far too simplistic and I'm sure the algorithm(s) change, alternate, or some such, but I can't help thinking that this clumping that occurs isn't as coincidental as pure logic would dictate.

Yes, yes, confirmation bias and all that fun psychological babble, but we're not really talking about a truly random system. There are covers that seemingly drop more often for certain people. I've mentioned it elsewhere and I'm certain the global pull rates fall within acceptable ranges of posted percentages, but perhaps they're not really being distributed to the degree one would expect in a truly random environment. Any validity to this line of thought?

I bet something along these lines is happening. Probably a random number generator on the client side for tokens and gem drops. I think numbers really are generated randomly, but sometimes the generator gets "stuck", or at least bogged down. I'm sure most long time players have seen weird stuff like when you get a 4 in a row match that clears a vertical line and every single tile that drops in is the same color. I play MPQ on my phone and maybe something running in the background takes up so much cpu momentarily that it interferes with the random number generator and makes it lag?

Dwarfsteel

hsk808 wrote:

I bet something along these lines is happening. Probably a random number generator on the client side for tokens and gem drops. I think numbers really are generated randomly, but sometimes the generator gets "stuck", or at least bogged down. I'm sure most long time players have seen weird stuff like when you get a 4 in a row match that clears a vertical line and every single tile that drops in is the same color. I play MPQ on my phone and maybe something running in the background takes up so much cpu momentarily that it interferes with the random number generator and makes it lag?

I don't think that this sort of problem would cause the algorithm to change. At best it would slow down the time it takes to compute the number, but the end result would be the same. Unless one of the variables is time dependent, but then again, I'm not really up to speed on programming.