Nothing strange about consecutive Lotto numbers

[3WA]

Guys, I analysed all the lotto results from the implementation of the RNG and I cracked the code.
For R100k, I'll sell you next week's winning numbers.

 

[gboy]

It depends on how they create the random numbers, purely mathematics is not random. Read a book called the chaos theory. It's very difficult to have pure randomes, smoke rising from a cigarette for example. I do know that some encryption algorithms use radio static / white noise on TVs, which can be attributed to things like the waning of Jupiter's magnetic field, sun spots ect. I even read that cloudflare uses webcams and lavalamps to determine the random number seed. I don't know if the national lottery uses the anomaly of a lava lamp to seed the RNG.

Chaos theory - Wikipedia

en.wikipedia.org

 

[Swa]

@Swa If a mechanical draw produced patterned results, would you also claim something about the system is off-balance so as to produce such results? Are you only satisfied with results that appear to you as random?

We can't test a mechanical draw but we can test an algorithm. Well actually we can test a mechanical draw to see if it yield random results and this is something they actually did on the day of the draw discarding machines and ballsets that didn't produce the required randomness.

With a mechanical draw you'll actually notice far less patterns than with an RNG.

 

[cguy]

I am fairly certain the RNG used is sufficiently complex that no analysis is going to show any form of significant bias towards certain sets of numbers.

As for the likelihood of “sequential/pattern appearing numbers” being less likely than “random appearing numbers”, that’s true, but will they still have the frequency of numPatternLooking/numTotalPermutation and numRandomLooking/numTotalPermutations. The first number being non-trivially small.

To actually test if there is some sort of rigging, one would have to see if over the course of many draws, the frequency of patterned numbers is disproportionate from a random draw. As is, there is no evidence of rigging.

 

[Swa]

I am fairly certain the RNG used is sufficiently complex that no analysis is going to show any form of significant bias towards certain sets of numbers.

As for the likelihood of “sequential/pattern appearing numbers” being less likely than “random appearing numbers”, that’s true, but will they still have the frequency of numPatternLooking/numTotalPermutation and numRandomLooking/numTotalPermutations. The first number being non-trivially small.

To actually test if there is some sort of rigging, one would have to see if over the course of many draws, the frequency of patterned numbers is disproportionate from a random draw. As is, there is no evidence of rigging.

You can still test the results it produces for having sufficient entropy, doesn't have to be for actual draws.

As for the rigging, yes that is harder to test unless being caught but I doubt anybody would rig it for a sequential number. More likely the RNG is just not very random.

 

[cguy]

You can still test the results it produces for having sufficient entropy, doesn't have to be for actual draws.

As for the rigging, yes that is harder to test unless being caught but I doubt anybody would rig it for a sequential number. More likely the RNG is just not very random.

Finding a RNG that is shows sufficient entropy isn’t hard. There’s nothing that suggests it’s “more likely the RNG is just not very random”.

 

[MrGray]

As for the likelihood of “sequential/pattern appearing numbers” being less likely than “random appearing numbers”, that’s true...

Not true. The easiest way to appreciate this intuitively is to imagine replacing the numbers with a set of 50 objects from which 6 are randomly selected, i.e. the ordinality implied by using a set of numbered balls is purely incidental and has no bearing on the odds of selecting any specific set of 6 objects.

 

[Swa]

Finding a RNG that is shows sufficient entropy isn’t hard. There’s nothing that suggests it’s “more likely the RNG is just not very random”.

Well none of us knows what the case is with the RNG used. This is simply taking it at their word that it is random.

Not true. The easiest way to appreciate this intuitively is to imagine replacing the numbers with a set of 50 objects from which 6 are randomly selected, i.e. the ordinality implied by using a set of numbered balls is purely incidental and has no bearing on the odds of selecting any specific set of 6 objects.

The same principle still applies. If specific groups of objects keep reappearing or in specific sequences we'll think it isn't due just to chance. Having numbers just makes it easier to recognise.

 

[cguy]

Not true. The easiest way to appreciate this intuitively is to imagine replacing the numbers with a set of 50 objects from which 6 are randomly selected, i.e. the ordinality implied by using a set of numbered balls is purely incidental and has no bearing on the odds of selecting any specific set of 6 objects.

I agree with what you say, after "Not true". The reason that it doesn't refute what I said, is that I think you missed the point of what I said: The set of pattern appearing numbers is smaller than the set of random appearing numbers, so it is less likely to be randomly select out of the former. This can be true while not implying that any individual number has any higher or lower chance of being selected than any other.

 

[cguy]

Well none of us knows what the case is with the RNG used. This is simply taking it at their word that it is random.

They said: "The computerized random number generator used by Ithuba is the same system used by lotteries across the world and is verified and audited."

The same principle still applies. If specific groups of objects keep reappearing or in specific sequences we'll think it isn't due just to chance. Having numbers just makes it easier to recognise.

One can actually measure the likelihood of a given sequence of lottery draws not coming from a uniform random distribution. If this likelihood is too high it would imply rigging or algorithmic problems. I haven't seen any evidence of this provided anywhere.

 

[Ninja'd]

Those are actually not strange. Any random sequence of this type will produce sequential numbers or nearly sequential numbers half of the time. So out of thousands of draws you will get a few that look like that.

And somehow 33 people managed to get it correct.

 

[Geoff.D]

When you change the wrong Random(20000000) variable to Random(20)

Wait, how many won?

20 people won

 

[cguy]

And somehow 33 people managed to get it correct.

Numerical sequences are more likely to be selected by humans as are combinations with 1 and 9 in them due to birth year, and 1-12 due to birth month, etc.

Assuming the earlier mentioned calculation of 5 per number, having 4x that on a human selected number seems plausible.

 

[Geoff.D]

In

It's interesting that people like to pick patterns like this. I had no idea.

2003 yes it was also heavily debated at the time. Fascinating to see how many people don't actually understand the statistical principles behind all of this though --- the same lot that try and make sweeping statements about the Covid 19 numbers .....

 

[Geoff.D]

Not a math whizzkid, but 59.62 million people in SA don't play lotto/powerball. I'd wager maybe 2 million players actually bought tickets.

At R 5 per line so the total number of lines played was 5 160 733 lines. Now, most people play more than one line.
If each person played 6 lines then the total number of players would be 860 128 thereabouts. There are stats somewhere about what the median number of players per draw is and how many lines those players on average play per draw. Further, there some that play huge amounts per draw. At our local PnP the manager was talking about an individual that comes in twice a week and spends about R 200 playing the Powerball per draw. So the numbers of unique players is quite low given the size of the population. And then you have to also be over 18 to play.

Then the math wizard does not know what the odds are for the Powerball either. According to the website ( I have not checked these numbers), the odds of drawing 5 plus 1 in the SA Powerball is 1 in 42 375 200.

 

[j4ck455]

Guys, I analysed all the lotto results from the implementation of the RNG and I cracked the code.
For R100k, I'll sell you next week's winning numbers.

Same here, just "hacked" some parameters in the url and I got 11 12 13 14 15 16.

 

[newklear]

At R 5 per line so the total number of lines played was 5 160 733 lines. Now, most people play more than one line.
If each person played 6 lines then the total number of players would be 860 128 thereabouts. There are stats somewhere about what the median number of players per draw is and how many lines those players on average play per draw. Further, there some that play huge amounts per draw. At our local PnP the manager was talking about an individual that comes in twice a week and spends about R 200 playing the Powerball per draw. So the numbers of unique players is quite low given the size of the population. And then you have to also be over 18 to play.

Then the math wizard does not know what the odds are for the Powerball either. According to the website ( I have not checked these numbers), the odds of drawing 5 plus 1 in the SA Powerball is 1 in 42 375 200.

It's going to take awhile to find out, out of just one random seeded instance:

import random
import time

matchlist = [5, 6, 7, 8, 9, 'Bonus: ', 10]
count = 0
notmatch = 0
start = time.time()
while notmatch == 0:
    randlist = random.sample(range(1,51),k=5) + ["Bonus: ", random.choice(range(20))+1]
    count += 1
    if randlist == matchlist:
        notmatch = 1
        print('The odds are 1 in', "{:,d}".format(count), 'to match with', matchlist)
        print('That took:', round(time.time() - start, 2), 'seconds')

 

[Geoff.D]

Guys, I analysed all the lotto results from the implementation of the RNG and I cracked the code.
For R100k, I'll sell you next week's winning numbers.

Sure I will pay you AFTER I win and you must float me a loan to play.

 

[Geoff.D]

People play the lotto the same way they select passwords and pin numbers -----

The winners, in this case, were probably people that chose their own numbers - not the quickpick players.

 

[Swa]

They said: "The computerized random number generator used by Ithuba is the same system used by lotteries across the world and is verified and audited."


One can actually measure the likelihood of a given sequence of lottery draws not coming from a uniform random distribution. If this likelihood is too high it would imply rigging or algorithmic problems. I haven't seen any evidence of this provided anywhere.

That could mean a number of things. Is it the same Raspberry Pi but a different program? Is it the same program but on another computer that could affect the results? Have people actually noticed clear patterns before that were just dismissed?

Then verified and audited doesn't mean much. Previous rigged lotteries were also audited and the results verified but nobody thought about checking the machines.