TurboGenius When does it lose ?

TG asked, when does it lose? I think this pertained to numbers catching up.

After all the testing and discussion it seems the answer is, sure it can win, but eventually it loses somewhat, and possibly sometimes more often than not.

So is there an edge to be gleaned from numbers catching up?

I think the jury is still out, and only the lawyers are making any money.

 

Is there a way to test it properly? I don't know you much, are you yet another riddler, or do you openly talk about your bet selection?
OMG now we have LGBTQ+BS, you cannot assume before saying it, you might get sued, lol.

 

You actually don't know shit about math. I thought it's an old fight and I never jumped into it, but, omg, man you are off the court. Let's talk physics.

Entropy, microstates, and the 2nd law of thermodynamics. It's the best way to study randomness.

The movements of gas particles (microstates) in a room (the whole gas system as a macrostate) are totally random. Their movement determines the properties of the macrostate (the whole gas system). These movements comply with the 2nd law of thermodynamics, so the system is losing entropy over time (losing heat, in this case). Losing entropy means the randomness of microstates is decreasing, so you can identify the location of each particle with fewer questions.

There are infinite random orders of gas particles with a specific level of entropy that defines the current properties of the macrostate. Think of every possible state of particles... One of them is all of the particles on one side of the room leaving half of the room totally empty. This is technically possible but won't happen in reality. It's exactly like the stream of 1s you posted. It is technically possible, but it statistically won't happen ever in our world.

Your crazy stream of random 1s lacks many factors of randomness (it looks like all the gas particles in one half of the room). One of the factors that are missing in your made-up random stream of ones is the normal distribution of repeaters/singles. Also, you can easily say it's not random because you have enough numbers to write down a Shannon entropy equation for it and see that the entropy of your wheel is 1. It means only 1 question is enough to guess the outcome.

- Is the outcome 1?
- Yes.

This is the definition of a not random wheel. You can throw it away after reading this.

You think the most important property of a random outcome is being unique. You are wrong. Random outcomes have statistical properties as well. That's why you chose a stream of 1s to make a point. You yourself knew it was not random. You just tried to imply that it's as unique as a really random one. You are right, but that doesn't make any sense in this context.

Repeaters, singles, random streams, you might say it's all statistics and randomness doesn't need to comply with stats. Then you might shit bricks if you find out that the 2nd law of thermodynamics is a statistical law, meaning that it technically can be violated, but don't bet on it, you won't win in billions of billions of billions of years.

 

I’m sorry everyone, but I hate to say it but dr sir is right. the chance of the sequence of 1’s is exactly the same!

Lets pretend the chance of getting al those 1’s is 1:1000 and same for the random stream.

Turbo and others are saying the stream of 1’s will never happen because it is not random???

The point is it SHOULD happen every 1000 spins on average or it would NOT be random!

Lets ask a different question:

Whats is the dropoff @TurboGenius ?

Meaning we start with a string of

2 x 1 what is the chance of happening?
3 x 1 “ “
4 x 1 “ “
5 x 1 “ “

etc etc

So turbo where do you say you never will see this string of 1’s?

10 times 20 100??? There should be a point (according to you) that it would not be random anymore

 

The point is that if your bet selection is 18,10,7,6,13,28... you win just as often on average as when betting 1,1,1,1,1,1... and also just as often if betting on each of the remaining (1/37)^100 - 2 sequences. While it's true that the distributions are not all equal, for any "common" distribution (an example being the first sequence), you won't win any more often when using it as your bet selection, so to all intents and purposes it's as rare as the second. To believe otherwise is like thinking that if you pick a sequence of lotto numbers like 3,7,18,29,38,42, then you're more likely to win than if you pick 1,2,3,4,5,6, because the first sequence looks more random. In reality they are equally likely. There are no "rare" sequences, and that means targeting sleepers is ineffective.

 

Shank, best to ignore Spike. For years he claimed his win-rate on the ECs was 72% (00 wheel!), and now he's saying it's 85% because he's improved with practice. It's in the same category as Gizmo's "reading randomness" twaddle, and there's no way to test it, so it might as well be precognition.

 

With all due respect, you are wrong. They are not equally probable. To understand the difference, you need a complete understanding of the probability of a sequence of 37 spins, not only with 36 repeats, but also it's specifically 1 that is repeating 36 times. No RNG can ever generate this sequence, only a human being with wrong intentions would call this a random sequence.

You just need a normal distribution graph of the outcomes of 37-spin cycles to understand this.



Lol.

 

So you agree with me then. Both sequences and in fact any sequence of length n has the same probability.

But that ignores the probability distribution. Waiting for a run of 100 1's or any specific sequence will see us through to the decay of all matter in the universe.

My point is that no-one bets like that. It's a senseless mute point as it's an infinitesimally small probability to succeed and ignores the probability distribution.

When we bet, we bet on a probability distribution, not on a long specific sequence that has a probability that is smaller than the tiniest thing you can even measure in the universe or waiting for that specific outcome to occur would take longer than the decay of all protons and matter in the universe.

 

Can you explain how you can get normal distribution on roulette number sequences?
If is as you claim, then also sequence 18-10-7-6 must-have higher probability than 1-1-1-1 ...
Can you show how you calculate these probabilities and where is the difference?

 

Totally agree with you! The chance it would happen is ridiculously small. But the point is turbo and shank say that the stream of 1 is different and not random and will never happen which is simply not true!

 

With all due respect could you show me the math that they are not equal

 

That's not my claim. Do it for yourself (I have better things to do).

Calculate the probability of seeing 37 spins with 36 repeats and 36 unhit numbers and the probability of seeing 37 spins with 12 repeats and 12 unhit numbers.

 

I just did. You just missed it. More info: https://letmegooglethat.com/?q=normal+distribution+of+random+variable

 

It's as true as the 2nd law of thermodynamics. Both can be violated. You will not be there when it happens if it ever does.

 

God no, the whole concept is ridiculous. You make a decision of where to bet on a spin by spin basis. How do you test that. You are coming at roulette from the totally wrong direction. You cannot test random outcomes. You would have better luck nailing Jell-O to the wall. If you do not understand that give up on roulette and find something else.

 

If you cannot describe your logic (so that it can be tested), there is probably no logic at all. Maybe it's something like precognition. Lots of claims, no actual way to validate.

 

If you are looking at outcomes as they appear one by one, and basing your bet selection on that, on what do you base your test. Testing would be impossible. The ultimate test is do you win more than you lose. I know this is a radical concept, but it is the only test that is valid. Testing random outcomes is a ridiculous waste of time.

 

Well done Shank. You are correct. Thumbs up.

The difference is, one is the probability of an event and the other is a distribution probability.

SirNoOne list of 100spins refers to the distribution probability and not the probability of an event.

 

Well in this case because we know the probability, and that it's a binomal distribution, the binomal distribution function is the most accurate. The normal distribution will approximate the binomal distribution with sufficient spins.

How many spins is sufficient?

You want both p*n and q*n to be 10 or greater, where q = p-1 and p is the probability of the event. 10 is standard practice but 5 is often good enough.

So for an even chance bet, p is 1/2 and q = 1-p = 1/2 also, and n would need to be 10/(1/2) = 10*2 = 20. A more precise calculation would be 10/(18/37) = 10*37/18 = 370/18 = 20.55 but 20 spins is satisfactory.

For a single bet 1/37, n would need to be the greater of 5/(1/37) = 10*37 = 370 and 10/(36/37) = 370/36.

So 370 spins is about right for the normal distribution to very closely approximate the binomal distribution but after 185 spins it's close enough.

A Binomial distribution is discrete, meaning it has an integer (countable) number of outcomes, like 7 or 9 but not a fraction like 4.3.

A normal distribution is continuous and often way easier to work with as the binomal calculation, the numbers get massive very quickly and overwhelm many computational limits due to the evil factorial (!) calculations required.

 

Thanks for that link, and what should i fill in here?

With all due respect but the probability of both strings are exactly the same! If not then please show me the drop off point, meaning at what max sequence of numbers are they different.

Here is funny anecdote for you. Not so long a go the guys at apple had to "fix" the shuffle function for the music player. Why, cause a lot of people complained that the shuffle function was not random enough! Sometimes they got doubles triples and quadruple's of the same song.

This is exactly how random behaves. So what was the fix? They made it less random so people wouldnt complain lol