Roulette Results - How many spins do casinos use to determine the unbiasness of a roulette wheel?

I know the next spin is independent from the last so how can a casino determine that a wheel has a bias? How many spins do the casinos use to determine this? Can't there be a sleeper or a hot number during an infinite period since each spin is independent regardless of each number having the same odds of coming up?

There's gotta be a number right? 100k, 1 million spins?

If we know that number and say we're 10k short of that number and know that on a single zero roulette table that number 25 has been a real cold number so far, representing only 2.2% versus the 2.7% that is expected. And knowing that there should be no bias to the wheel shouldn't it go on a hot streak for the remaining 10k of spins more so than the 36 to 1 that is on offer if we start flat betting?

The above might be too extreme and exact.

Even if say the casino accepts 2.6% during a finite period to determine if a wheel is bias or not, can't we try to find a cold number that represents 2.4/2.5% and know that it needs to get to 2.6% by a finite period?

Quote: WizardofAus

And knowing that there should be no bias to the wheel shouldn't it go on a hot streak

You just decribed Gamblers Fallacy. Whenever you use
the word 'should' to describe a game of random chances,
you're making the oldest mistake in gambling.

I'm surprised at all of your Roulette questions....... as the WizardofAus you should already know the answers!

Thanks for the replies guys. I understand all the Math's and about the long-term and that anything can happen probably will happen over the test of time but what is the number the casinos use. If they use 100,000 and they find that number 25 only represents 2.4% dorsn't that mean a bias? Can you please read my question and what it really is asking. I know the Wizard does some work for casinos and I'm sure others do as well do hopefully someone can let us know. Cheers.

While I don't know the actual number, the following is a screen shot from TCS John Huxley's website on wheel analysis.
(FYI: TCS is the premier manufacturer of Roulette wheels among other things.)

Admin note: removed image www.djteddybear.com/images/wheel_analysis.JPG

It's impossible to know if that chart represents actual data, or spcific data fed in to make an interesting chart.

It's blurry, but I *think* it says that this is for 2,000 spins - but that still doesn't answer the question.

Quote: WizardofAus

can't we try to find a cold number that represents 2.4/2.5% and know that it needs to get to 2.6% by a finite period?

You are barking up the wrong tree here. If a number is cold, that does not make it "due to come up". If a number has been cold, by the time you have the data for 10 billion spins, say, the number will get back much closer to the expected percentage, sure. But the way that will happen will be by burying the significance of the sequence of spins that showed it cold. The 10 billion spins will just bury that anomaly by larger numbers. The odds that the coldness will be 'corrected' by a period of being hot or no better than the number staying cold, in the short run. Furthermore, if the number never gets hot, the percentage will still 'correct'.

What you should be trying to do is determine if there is a bias in the wheel, if you are serious. That would probably be a sector rather than a number. My gut tells me that you probably need about 10 million spins to start to get fair certainty on that. Also, it is said casinos are 'all over it' when it comes to handling biased wheels these days.

I guess more for curiosity's sake than anything else, I'd love to know how many spins are used to determine the "Hot" and "Cold" numbers as seen on this history display.

Only a fast dealer with few players can get more than one spin a minute. Usually it's more than a minute between spins. I once witnessed a full 7 minutes between spins on a packed table. Therefore, this has to represent a full day's spins, or perhaps even more.


Note that because of the style of lettering, it didn't photograph well. It was much more readable live.

Quote: DJTeddyBear

I guess more for curiosity's sake than anything else, I'd love to know how many spins are used to determine the "Hot" and "Cold" numbers as seen on this history display.

Only a fast dealer with few players can get more than one spin a minute. Usually it's more than a minute between spins. I once witnessed a full 7 minutes between spins on a packed table. Therefore, this has to represent a full day's spins, or perhaps even more.

Note that because of the style of lettering, it didn't photograph well. It was much more readable live.

Cammegh's Billboard Display system defaults to 300 but can be customized.
EFXvLCD Roulette has options for last 100 spins, today,last 30 days and since last reset.
CYCLOPS Looks like for hot 1000 and is says how many times. For cold it says how long it has been

Quote: WizardofAus

I know the next spin is independent from the last so how can a casino determine that a wheel has a bias?

You seem to say that you know each spin is independent but your question implies that there is somehow a "should-happen" event and that you think it should take place on the next spin to "balance out the world" or something and make all things right with the universe.
Well, it might just happen on that next spin.... or the tenth one from now. ... or never.

No casino ever shows the "hot" numbers or "cold" numbers on their annunciator panels. They show you the recent history and let you decide if three reds in a row means anything about the next spin.

Its like the Baccarat graph paper the casino distributes and that tiny golf pencil. They provide the little squares but they don't tell you that charting the game and trying discern the results will aid you on your next bet. Sure its been Player for "x" hands. That doesn't mean Banker is "due"... that next hand may still be Player.

Quote: FleaStiff

No casino ever shows the "hot" numbers or "cold" numbers on their annunciator panels. They show you the recent history and let you decide if three reds in a row means anything about the next spin.

Actually some casinos do show you the "hot" and "cold" numbers and they use those specific words.

Actually if you detected a true bias in the wheel due to some physical defect or way the dealers used it, the "cold" number would remain less likely to come up in the future, which is the opposite of what WizardOfAus is speculating. It is amazing how strong a hold the gambler's fallacy has on people's minds. They think they understand it but then it pops up again and again.

There is no magic # of spins that helps you determine if a wheel is biased or not. You can use a hypothesis test and the chi-squared distribution to do the bias test, but the result of the test is only a %.

You could only say with some percentage probability that the wheel was unbiased or not. Then the test itself has a power of the test, which is the probability that is makes an error in the conclusion.

Show us a bunch of spin results, and we can tell you either (a) the wheel is likely biased, or (b) there is not enough evidence to say it is not unbiased. The stronger the result you want, the more spins you need. I know one textbook that says you need a minimum of 5 expected results per outcome, so you need a bare minimum of 5*38 = 190 spins.

It is interesting to note you can't actually PROVE a wheel is unbiased. You can only collect so much data and conclude that there is not enough evidence to say the opposite.

Quote: dwheatley

You could only say with some percentage probability that the wheel was unbiased or not. Then the test itself has a power of the test, which is the probability that is makes an error in the conclusion.
It is interesting to note you can't actually PROVE a wheel is unbiased.

That's a moot point, because there isn't and has never been a single unbiased roulette wheel in the world anyway.

What you can conclude for a given wheel is for instance that the bias is smaller than +0.05 on every number with a confidence of 99%. This means that 99 of 100 wheels that pass the test 99 have a bias of less than 5%, and 1 wheel greater than 5%.

A 5% bias is very significant, I believe casinos maintain much lower values, but even with 5% you are still behind the house edge.

Quote: P90

A 5% bias is very significant, I believe casinos maintain much lower values, but even with 5% you are still behind the house edge.

... except on single-zero wheels.

Yeah, but then I think the bias in practice is kept under 1% with 95% confidence or better, seeing how the casinos track every spin, it's easy to do. And on single-zero wheels, esp. with higher limits, particularly tight.

Every casino that I have been to in Singapore, Malaysia, Australia, New Zealand displays the "hot" and "cold" numbers on their annunciator panels on the Rapid Roulette tables. Rapid Roulette has a live wheel and croupier, but patrons bet on an electronic touch screen.

I cannot recall what they do in Macau as Rapid Roulette was just being introduced last time I was there.

Thanks for the discussion guys. I understand gambler's fallacy and that the next spin is independent or even the next 100, 1000, or 10,000 spins is independent of previous results but no has really answered my question. I've seen casinos fix roulette tables or do some mechanical adjustments sometimes so this may be caused by a mechanical bias. The question is - what is the number that casinos use to do determine it?

Please stop talking about after a billion hands or so because I'm sure the casino probably checks the results before that to determine that every number should come up approx. 2.7% of the time.

It's gotta be a figure closer to 20,000 right? That's once a month assuming 30 spins an hour on a wheel that's open the whole time. Even if it was once a year - that's only 250k spins. A little less than the 10 billion hands.

I love how the everyone who states the gambler's fallacy states it but still states that in the long-run each number 'should' come up equally.

Quote: WizardofAus

Thanks for the discussion guys. I understand gambler's fallacy and that the next spin is independent or even the next 100, 1000, or 10,000 spins is independent of previous results but no has really answered my question. I've seen casinos fix roulette tables or do some mechanical adjustments sometimes so this may be caused by a mechanical bias. The question is - what is the number that casinos use to do determine it?

The question actually has been answered. There is no hard number such that more spins than that is automatically considered "long run". Instead, the number of spins is a parameter in a certain formula. Another parameter could be, say, the number of times a given number hits. You plug those two parameters into the formula, and you get a result like "the wheel is unbiased with 99.5% certainty".
The higher the number of spins, the better certainty you get in your answer. For a rough estimate, just a few hundred of spins may be enough. For a more or less "conclusive proof", a few thousand may be required, maybe even a few hundred thousand, depending on the basic hypothesis you are trying to prove.

But what's your point anyway? For the sake of argument, let's say the number you are looking for is 1000 spins. Now what?

Quote:

Please stop talking about after a billion hands or so because I'm sure the casino probably checks the results before that to determine that every number should come up approx. 2.7% of the time.

Sure, they check the results constantly. If something seems suspicious to them, they'll replace or fix the wheel. How does it help you though?

Quote:

I love how the everyone who states the gambler's fallacy states it but still states that in the long-run each number 'should' come up equally.

The trick is, there is no way to tell when exactly the "long run" will begin. What's certain is that it is definitely not going to happen on your next spin.

There is no "long run" or "short run".

There is just the run, which is completely random, and you can cut it up into segments of arbitrary length, which tend to become more irregular as they get shorter, but not necessarily more regular as they get longer.

Thanks for the in depth reply Wease.

Ok. So there's a chance it could be just some formula they use instead of merely assuming they are looking for 2.6% or 2.7% for each number over a given period.

If it was 1000 and using the 2.7% logic then there'd be way too much biasness. I'm assuming ut's at least 20,000 if they were to use my logic. I guess it would be done formula base instead.

My original point was, say it was 10,000 spins or once a month they check it. Assuming the wheel is only open half a day. After 9,000 spins there could be a few numbers well below 2.7% or whatever parameter the casino sets. You would then flat bet that number for the remaining 1,000 spins and assuming the wheel to be unbiased it would get closer to the 2.7% or at least better than the 1 in 36 spins.

But Wease you're probably right that they use a lot of other mathematical formulas/tests to determine the unbiasness of a wheel.

Cheers.

BTW- The wheel at my local casino shows hot and cold numbers for the last 250 spins. Even shows how many times they've hit.

I've seen 17 times for a hot number and actually once seen 0 times for a cold number. That particular cold number did end up coming after 304 spins.

It was one of those virtual tables that produces 100 spins an hour.

Quote: WizardofAus

Please stop talking about after a billion hands or so because I'm sure the casino probably checks the results before that to determine that every number should come up approx. 2.7% of the time.

As an engineering question I would say zero. Maybe they measure it during the play, and report out the data at the end of the year, but it is too gross of a test to spot a bias in the wheel. I would suspect you would have sensors to measure decay rate as the wheel slows down. If you have enough data, you can sense a bump in the decay rate, and see if the ball bearing is out of line.

It was a common question when we were testing sonar transducers. We used to test them to 100 million cycles, and we cut it back to 50 million cycles. Now in operational use, one might go through 50 million cycles in less than a month. People would always ask what kind of statistics we used to determine the number of cycles. The answer was always we didn't use any statistics, but we knew that most units that survived initial use, would often last for years without problems. It's just that 100 million cycles simply was too expensive.

This topic was covered by 'statman' (a banned member) in a few of his threads. He deleted most of his.
Here is some reading on the subject

Roulette Bias Exposed

and
From Author(s): S. N. Ethier
Testing for Favorable Numbers on a Roulette Wheel

and from S.N. Ethier
Doctrine of Chances
2010
Roulette Chapter:
"So far, we have assumed that the roulette wheel achieves perfect randomness,
that is, we have assumed that each of the 36 + z numbers is equally likely.
There is an abundance of anecdotal evidence that this is not always the case
(see the chapter notes).
The question of how to detect and exploit biased
roulette wheels is an interesting one that we attempt to address in this section.

Suppose we have “clocked” a roulette wheel for n coups, and the most
frequent number has occurred a proportion 1/25 of the time.
Is it a biased wheel?
Although we can never be absolutely certain, we can nevertheless
quantify our degree of confidence. It will of course depend on n."

page 472 continues
"Using r = 37 or r = 38 gives a test for determining whether the wheel
is biased, based on the frequency of the most frequent number.
Actually,
we are not concerned simply with whether the wheel is biased.
The relevant
question is whether there is a favorable number, that is, one that results in
a superfair bet; a necessary and sufficient condition is that its probability p
satisfy 36p − 1 > 0, or p > 1/36."

Interesting math and reading.
This is an expensive college text book. maybe can find a Google preview.

The casino uses statistics to track whether a wheel is biased or not. It can use a constant chi-squared analysis that determines whether a wheel is biased or not. Basically, to tell whether a set of results is biased or not, you take the sum of squares.

Chi Squared = Sum of all [(observed - expected)^2] / expected.

For example, if you have a sample of 380 spins (00 wheel), with the following distribution:

number of occurences / numbers having that occurrence
6 - 1
7 - 3
8 - 5
9 - 7
10 - 8
11 - 5
12 - 3
13 - 4
14 - 2

The expected value is 10. Chi squared is (1.6 + 3 x .9 + 5 x .4 + 7 x .1 + 8 x 0 + 5 x .1 + 3 x .4 + 4 x .9 + 2 x 1.6) = 15.5.

There are 37 degrees of freedom (n-1). You plug this is and get a Q (probability) of .9995. The wheel is not biased.

For a wheel to be biased you would be looking at a Q value of .05 or less. For that to occur Chi Squared with 37 degrees of freedom would need to be about 52. For example, a result of a number occurring 30 times in 380 spins would raise a flag (as the contribution of that to the chi-squared alone would be 40).

I am not sure what sample size I would use but if I was a casino operator, I'd be looking at multiple sampling intervals of spins to look for both short term and long term bias.

Quote:

It was one of those virtual tables that produces 100 spins an hour.

Unless I'm mistaken, that's a table they are just going to trust. It shouldn't have any bias, but could have the best flaw of all, predictable numbers. Presumably there is nothing simple about how the numbers are generated, though. You'd have to crack the algorithm used by the RNG, a pretty tall order. Nonetheless, some funny things have happened, such as an RNG that starts over from the beginning because the casino turned off the machine every night.

Quote:

can't we try to find a cold number that represents 2.4/2.5% and know that it needs to get to 2.6% by a finite period?

I still say this statement betrayed some fallacious thinking, in spite of your protests.

Quote: WizardofAus

My original point was, say it was 10,000 spins or once a month they check it. Assuming the wheel is only open half a day. After 9,000 spins there could be a few numbers well below 2.7% or whatever parameter the casino sets. You would then flat bet that number for the remaining 1,000 spins and assuming the wheel to be unbiased it would get closer to the 2.7% or at least better than the 1 in 36 spins.

That's really the most blatant form of Gambler's Fallacy.
No offense meant, but you'd expect someone picking such a rather ambitious username to be better with the basics.

Let's put apply it in a different scenario.
Given: Over the past 100 years it has been determined that a male scores on the average 3 times a year. In the first 352 days of 2011, Butthead never scored.
Question: Based on this data alone, is Butthead more likely to score in the remaining 3 days of year 2011 or in the first 3 days of year 2012?

Quote: WizardofAus

Ok. So there's a chance it could be just some formula they use instead of merely assuming they are looking for 2.6% or 2.7% for each number over a given period.

If it was 1000 and using the 2.7% logic then there'd be way too much biasness.

Too much biasness? Because of the small sample?

What if you had a sample of 100 spins, and in that 100 spins one number hit 4 times, while the pocket next to it hit only once. What does that tell you about those two wheel pockets? Did you say "That there's a bias towards one and away from the other" ?

(Correct answer: NOTHING!)

Now multiply both numbers by 10 for your 1,000 spin sample.



The sample size does not affect whether or not there is a bias. It just affects the variance.

Quote: WizardofAus

My original point was, say it was 10,000 spins or once a month they check it. Assuming the wheel is only open half a day. After 9,000 spins there could be a few numbers well below 2.7% or whatever parameter the casino sets. You would then flat bet that number for the remaining 1,000 spins and assuming the wheel to be unbiased it would get closer to the 2.7% or at least better than the 1 in 36 spins.

So you think that a frequency of a particular number hitting should depend on the casino policies regarding monitoring their results?
What do you imagine the mechanism of such dependency should be? I mean, how does the wheel "know" that it is about to be checked for bias and needs to make up for the "cold" numbers or risk getting replaced?

Quote: WizardofAus

My original point was, say it was 10,000 spins or once a month they check it. Assuming the wheel is only open half a day. After 9,000 spins there could be a few numbers well below 2.7% or whatever parameter the casino sets. You would then flat bet that number for the remaining 1,000 spins and assuming the wheel to be unbiased it would get closer to the 2.7% or at least better than the 1 in 36 spins.

I will quote this reply as well, because this is nicely summarizes the problem with this idea. An unbiased wheel will not cause one number to come up more often to make up for the number not coming up in the past! That is, frankly, really silly.

'Unbiased' wheels will occasionally have results that look biased, especially over the 'short-term'. That's why we can't prove they are unbiased, only that they probably are. Some periods of observations will have results that look biased, and we could even mistakenly conclude that an unbiased wheel is biased.

If you believe the wheel is UNBIASED, there is no reason to bet any number (due or not due), you will lose in the long-term.

If you believe the wheel is BIASED, you should be betting the numbers that come up more frequently!

So, with that said, wouldn't the best gambling system for roulette be to always play the most frequent numbers?

The probability that a wheel is biased > 0. Bias means that a number is more probable to land in the same spot. Therefore, by betting the same numbers, your house advantage will be reduced by the bias of the wheel, if any.

I love the passion on this forum. I'm glad I found it and I'm sure I'm gonna be saving a lot money because of it. Thanks all. You guys must really hate casinos. Can I ask of all those who have replied to this thread if you do play in casino's or gamble on anything with a house edge - if so - why and is it with a system/strategy?

Gambler's fallacy is quite a great concept - even though the roulette wheel doesn't have a memory it does have a degree of certainty of for every singly possible future situation. I think when someone comes up with a system or strategy - it's not to beat the roulette wheel or casino in the long run - it can't be done - it's just to beat it for a couple of hours or so when that person enters a casino to make a bet.

Here's something t get the argument continuing - Video Poker is a game that can sort of be beaten - with a large bankroll you could see a profit in the long-run that will subsidise an income - but it's the same as roulette in one sense over 2 hours - you sit down, you play - you might get the really bad end of the stick for the 2 hours. You don't know what's gonna happen the next hand or the next 2 hours, you might know what's gonna happen in the long-run but do many people have the bankroll needed to find out the long-run that is profitable in one human lifetime?

You sit down at a single-zero roulette table or any house game with a very small house edge and you come up with a system/strategy that has a degree of certainty of 99% over your 2 hour session - is that possible?

If I gave you the below parameters:

Roulette table that spins 100 spins an hour - minimum bet 50c - maximum inside bet $100 - maximum outside bet $500. Is there a strategy that could earn a person $40 an hour? I posted a thread in systems forum that my mate & I have started playing on a virtual table that allows the above.

I used to play or sorts of roulette systems 10 years ago but gave it up, most made money but eventually failed - wisely - and started playing poker - but one of my poker buddies introduced me to these new roulette tables - the speed and limits have re-invigorated my desire to look at systems/strategies that may actually work on these systems in the "short-term" - it's a single zero - so the edge is 2.7% which with in the short-run could be overcome with a solid strategy...let the debate continue...

Quote: WizardofAus

it's a single zero - so the edge is 2.7% which with in the short-run could be overcome with a solid strategy....

Why do you think that?

Why do you imply otherwise? Are you say that roulette strategy/system cannot overcome the 2.7% over 2 hour session/200 spins?

Any system or strategy can get lucky, but to say that a specific system or strategy WILL be successful is wrong.

It's quite funny when people say you were lucky to win at Roulette. I beg to differ... If you had a system similar to my friend's one that I posted in the Betting Section where he uses a 150
Spin Betting Progression and has odds of losing of 1 in 61 betting progressions and if he were to play 8 betting progressions in a 2 hour session... I think he'd be very Unlucky to lose. He'd be right to win for the 2 hrs. We're talking 2 hours not an infinite amount of time. I know sessions (in this case 200 spins) are arbitrary in these mathematical discussions but no human will ever play close to a billion hands which is good test of time for these strategies/systems. A human who played 200 spins a day, seven days a week for 50 years would only see/play 3.6 million hands. Is there a system that might last that period of time...I'm not sure.

Test of time is a funny test...has anyone who has ever tested a system/strategy ever lost on the first progression that they ever played? how about if a person came up with 10 different progressional systems/strategies and played all 10 in a casino each day. After each ONE progression, rip it up and never use it again. There's a strategy to think about. I've noticed everyone who stands behind the gambler's fallacy always states that every system will eventually go bust. Let's not allow it to get to that point. Just one progression. I'm sure if I had a BR of $5000 and took into a casino 10 different progressions that gave me a degree of certainty of winning of 98/99% (similar to my friend's one that I posted in the betting system section) then I would consider myself very UNLUCKY to have lost that SINGLE FIRST progression for each strategy/system when I had a 1 in 50 chance of losing. I would not be considering myself LUCKY at all for winning those 10 single different progressions.

Just one Human Lifetime... not 330 lifetimes... Just one.

So if for those 10 progressions you won $200....after 25 sessions (only 50,000 spins not 1 billion) you make your BR capital and then are free rolling for the next 50,000 spins.

Come on... I'm sure if all the mathematical minded people on this forum put their minds together I'm sure we could come up with hundreds of different system/strategies that fit within a $5,000 BR and had at least a 98% DC that we could play just once.. Just once!

Here's a challenge for a team of 10 'mathematically challenged' as the Wizard puts it. 10 of us put in $500 each. Each day we come up with a system/strategy that could be run at say BODOG within their limits and then post those 10 strategies and their results in a thread in the BETTING System and the result each day.

It would be quite interesting to see how we go.

Quote: WizardofAus

Come on... I'm sure if all the mathematical minded people on this forum put their minds together I'm sure we could come up with hundreds of different system/strategies that fit within a $5,000 BR and had at least a 98% DC that we could play just once.. Just once!

What would it prove? That you got lucky and won
once? Whats the point.

No. That for that one occasion we didn't get UNLUCKY!!!

So we come up with 10 progressions each day and use it at BODOG and make $200 each session.

For every 250 systems/strategies that we come up with we win our BR.

So it doesn't prove that we are lucky it proves that a human may not get Unlucky if they had a strategy that went against the long-term. In this case beyond just one progression.

If I said to you - as a living ($200 a day net), each day I want you to come up with 10 new roulette strategies/systems that had a DC of at least 98% (similar to the 150 betting progression one that is on my other thread) and had an initial BR of $5000 and table limits of 50c minimum and $100 maximum insides, $500 maximums outside... Would the average person do it?

That's the whole point of getting lucky or not UNLUCKY just once!

Quote: WizardofAus

So we come up with 10 progressions each day and use it at BODOG and make $200 each session.

So you don't want to play just once. Look, I'm
trying to not say a lot of things here, but you
really don't understand that a progression is
just a bandaid that postpones the inevitable
loss. It doesn't beat or fool a game where the
edge is permanently against you. What you're
suggesting has been done already, probably
thousands of times in the last 200 years. It
doesn't work because it can't work. If you
don't have the edge, you can't win consistantly.

I know you're trying to say it's about the long-run EvenBob.

It is just once - variations of probably some progression that is very similar but you only play it once. Then you jump onto the next one. It's like entering a casino, playing the dreaded Martingale just once then playing Labouchere just once. But in this case you have a BR of $5k, single-zero, limits that allow for long progressions and high DC.

So that's the job... 10 different progressions per day. BTW- Each betting spin needs to net $1 as per the progression my friend came up with that I posted on the other thread.

If I may back up to the point of casino maintenance, it's done periodically. Maybe its every quarter during the off season on a dead table, mayby monthly on the full time wheel, maybe several times in between due to random events like an equipment audit or spillage. As one of the few "dark siders" here, I don't know and couldn't tell you exactly. And if I'm here 5 days a week and deal with this stuff and still don't know for sure, it's doubtful an outsider could somehow know.

As far as "detecting bias", I'm not entirely sure there's a big to-do about it. (I actually don't think they'd care if the thing sat sideways and grinded to a halt, if no one's killing them on it, meh, who cares.) How many spins happen over a month long, 24/7 period? Enough to make a determination? My gut says no, and then they go and take the thing apart and replace it with the reconditioned wheel from out back. Now you have another whole month to start over from scratch. But hey! that first wheel comes back out, right? Indeed it does, but so?

I mean, say we have the dream, a roulette wheel whose fret around, say, the #4 is a complete loose sponge. It just caressess the ball as it hits and gives #4 a decent bias. How do you notice that? Check the last 20 on the tree? Nope, not enough. Check the last 200 or even 2,000 on the program? Nope, not enough. Sit there every waking moment taking notes? You're allowed to, but who's going to? And even if someone did, are they also going to keep track of when that wheel comes off and notice when it comes back on? How much money would one lose before they noticed? And how long would it take for someone to notice that, hey, Steve the Tech done went and fixed that fret back in Octember?

More power to you, but I just can't see this happening. Seriously, if I had 10 wheels in a room where one was purposely biased, all manned by a croup and played and replaced per regulations, how many man-hours would it take to definitively find the bunk wheel? Now take that man-hour number, and imagine doing all that work for the extremely remote chance of finding one in the real world. I'd have to have some seriously deep passion, seriously deep pockets, unlimited free time and one hell of a table max to bother with such a hunt.

Quote: WizardofAus

So that's the job... 10 different progressions per day. BTW- Each betting spin needs to net $1 as per the progression my friend came up with that I posted on the other thread.

That's not a "job" anyone here would be willing to work on, except perhaps for money. But you've really got to get over the delusion that some combination of bets can beat the house. If you admit, as presumably you do, that no single bet can beat the house, then you should examine why any combination of those could. It's like saying "I can't go forward by taking one step back, and I can't go forward by taking two steps back, but if I mix going backwards by one and two steps then I'll be going forward." It just ain't so.

Thanks for the responses guys.

Maths - I understand the maths but we're talking short-term blocks that are a fraction of the time period you're referring to. In the long-run the house edge is the house edge and for every $1000 I place I should lose $27....in the long-run but how about if you never get to that point in the arbitrary time you're referring to?

If that job was an hour a day making the systems and then an hour or two more playing them then the rest of your day playing golf or whatever passion in life you have wouldn't that be a job you would want.

I guess that's what I'm trying to determine on this virtual wheel that my local casino is offering.

If I can do it part-time for two hours a day - if it's one system, 10 different systems for 10 progressions or if it's a 100 systems that I just randomly rotate through can it work? Can you miss the unlucky short-term rotation that the each system must miss - 1 in every 50 progressions. So after playing it 50 times you should lose but what if you miss that one time that you were meant to lose for an entire lifetime? Probably not but will you miss it as little times in a lifetime that makes you a profit in a lifetime?

How about if it was 1 in 100 - I'd the DC was 99%.

Let's all remember that while no system can beat the house edge, some systems *do* increase the short-term win, at the expense of suffering a huge loss when you do lose. For example, playing the Martingale on a $5 Roulette table with a starting bankroll of $1000 gives you an 82% chance of winning a one-hour session, compared to only a 46% chance of winning if you flat bet $5 every time. You'll win most of your sessions, but when you do lose, the loss will be catastrophic. Taking this further, some systems or scenarios could give you an even greater chance of winning a one-hour session, which could make it attractive to those who want a higher chance of winning. The problem is that even if the chances of winning a one-hour session are 95%, that means that you'll expect to go completely bust once every 20 sessions (i.e., once every 20 hours). So forget about the long-term; even a 95% system wouldn't normally last even a week of full-time play.

Thanks for responding MB.

What you've written is essentially what I've written but is a lot different in terms of numbers - 95% is a lot different to 98/99% - you could foresee a very long streak of wins on a system of 99% compared to a 95% system. with odds of 1 in 100 - there's a lot bigger chance that the streaks could be massive - Even with a 37 number roulette table where the odds are 1 in 37 - people have noticed sleepers that run 500 spins when a test of 1 million has been run - imagine something a roulette wheel with 100 numbers a sleeper could go a lifetime - and maybe a human might get lucky and this system could go a lifetime.

Just food for thought.

Thanks once again for your response - hopefully someone can help with the testing of this system.

Quote: WizardofAus

- 95% is a lot different to 98/99% - you could foresee a very long streak of wins on a system of 99% compared to a 95% system.

In every system, loss probability is in direct proportion to win/loss ratio, given the same system efficiency (maximum is presumed for below calculations).

This means that where a 95% system has a win/loss ratio of 0.5^log[20/38]0.05=0.04 or 1:25, i.e. for your 19 wins of $10 each you'll lose $250 when you lose, a 99% system has a win/loss ratio of 0.5^log[20/38]0.01=0.007 or 1:144, i.e. for your 99 wins of $10 each you'll lose $1,440 when you lose, and you will lose.

Quote: WizardofAus

- and maybe a human might get lucky and this system could go a lifetime.

Or a human might get lucky and hit a lottery jackpot. In fact, lotteries sometimes turn into a +EV game, when the pot is greater than ticket price * odds.

Not only that, but humans actually have gotten lucky and won lottery jackpots. It happens every year, every week, every day even if you count all the numerous lotteries.

Thanks P90. Appreciate the response.

If you read this thread and the other one that I created you know I've already suggested what a bad bet this is in the long run...over a billion hands or over the test of time.

But for a 1 progression or for 2 hours it might be a good bet.

If I gave you a wheel that has 100 numbers and gave you the odds of 101 to 1 so giving you an edge but the only bet you could make is one single bet on one number each day for a whole year is there a chance that you will miss that one number chosen every day for a whole year? I think there's a big chance that you might miss.

Unless you revert to the Gambler's fallacy and pick the same number everyday. But I'm sure you don't believe in past results influence the next spin.

So the best way to check the above challenge is to have two separate RNGs and getting them to create 365 numbers in a row and seeing if any two numbers occur at the same time. For a complete test of time over a billion hands there could be huge stretches
of 1,000 or even 10,000 spins until the two might hit together even though it's 100 to 1 chance.

I already advised that I understand why the maths say the next spin is independent of history...a ball doesn't have a memory but I think a RNG does have a memory due to the input that the odds are for 37 numbers of any number coming up is 36 to 1. It may not come up in the next spin but an RNG will surely require a number to come up within the next 1,000 spins or even shorter most occurrence. I think someone did a 1 million spin test and they noticed sleepers for 500+ spins but they highly thought it would BR doubtful for a number to sleep for much longer than that.

So an RNG has some sort of memory because we've programmed the 36 to 1 parameter into them even though within 500 spins almost anything can happen but not everything!

Quote: WizardofAus

If you read this thread and the other one that I created you know I've already suggested what a bad bet this is in the long run...over a billion hands or over the test of time.
But for a 1 progression or for 2 hours it might be a good bet.

A bet that is bad in the 'long run' is just as bad in the 'short run'.

Short and long runs only exist in sports. In gambling there is just one infinite run of outcomes. By selecting a shorter section you are less likely to encounter a canyon wall drop. You are also less likely to encounter a mountain.

You probably have at some point played paintball, airsoft, lasertag, or participated in a exercise with blanks or Miles.

What would you say if I load 1 live round for each 999 blanks I'll be firing at you, and give you a dollar for every round I spend? Being a decent marksman, there is little doubt I'd kill you in the long run... but hey, if I only take four 5-round mags with me, isn't it just a free twenty in the short run?

Quote: WizardofAus

...a ball doesn't have a memory but I think a RNG does have a memory due to the input that the odds are for 37 numbers of any number coming up is 36 to 1. It may not come up in the next spin but an RNG will surely require a number to come up within the next 1,000 spins or even shorter most occurrence.

If it has memory, it's not a RNG.
It will never be certified for use as a RNG, never be included in any modern RNG library, and will never be used by any casino worth its felt.

Quote: WizardofAus

doubtful for a number to sleep for much longer than that.

I slept last night. But as for either of my numbers, I don't think they did.

P90 - very interesting analogy but for it to be like roulette you'd need to be picking that one live bullet from a box of 1000 bullets. If I gave you one go each day at it. There'll a chance that even if you lived for another 100 years there'd be a big chance that you might not ever pick the right bullet. Remember the one that are blanks that you pick get replaced into the box each day and shaken around randomly again.

Quote: WizardofAus

If I gave you one go each day at it.
Remember the one that are blanks that you pick get replaced into the box each day and shaken around randomly again.

Well, I never said it has to be one-on-one. That would be kinda pointless. So rounds will get used up, and, yes, replenished at the same 999 to 1 ratio of blanks to live, continuously. We have a deal then, if I'm hearing you right.

While at that, did you know American boys you-know-where spend over 250,000 cartridges for a kill, most of that actual combat expenditure? That's 80 times better odds than I gave you. The other side probably spends more further (they just suck at bookkeeping), so your odds on the blue team are even better - and that's before you count armor plates in. Live the high life, join the forces for a few short runs.

P90 - We have a deal.

So each day for the rest of your life you choose one number between 1-999. We then spin the wheel and see if you're right. If you're right I will give you $1000. You have a house edge there. If your number doesn't match I get $1 from you.

We can start today.

I live in Australia though do let's settle accounts after every month.

You choose a number we spin the 999 number wheel and we see how you go.

How old are you? Trying to see how many goes you might have at it.

You might get lucky straight away but I have a 99.9% DC that you'll be wrong.

I'll take my chances.

Deal?

Quote: WizardofAus

P90 - We have a deal.
So each day for the rest of your life you choose one number between 1-999. We then spin the wheel and see if you're right. If you're right I will give you $1000. You have a house edge there. If your number doesn't match I get $1 from you.

Well, almost there, but it sounds like you might have misunderstood my proposal somewhat. It's more like you take a reliable life insurance policy for $400,000 in my name, we set up a randomizing scenario, and if the smallish element of chance happens to get chambered at the right time and catch up with you, we proceed to collecting.