The Truth, In a Way, The Past Numbers To Have Hit Do Matter
Here's the real deal on past spins influencing future spins. The following information is based on a REAL ROULETTE WHEEL WITH A LIVE DEALER. On these wheels the rotor did NOT alternate spin direction- (every other spin like they do in EU).
The following information does NOT pertain to an RNG based wheel or the random game of roulette. In other words, no electronic machines.
I can tell you the facts, based on my experience, as they pertain to a real wheel. Here they are, like them or not.
1. If a number has hit within the last five spins, it really is slightly more likely to hit on the next spin.
2. Simply playing the last five numbers to have hit will slightly reduce the house edge, especially when the wheel speed is close to the speed of the previous spin. In rare cases, you can get a small edge on some wheels playing this way.
3. The dealer's pick up and release of the ball is not random, like someone on this board had stated. As a matter of fact, it will quickly spike out beyond five standard deviations in a short period of time for most dealers, if you take the time to measure it.
-Keyser
Quote: Keyser
1. If a number has hit within the last five spins, it really is slightly more likely to hit on the next spin.
You do realize that if this effect was actually real, then it would also be cumulative?
In most cases, the effect lasts for only five or so spins. It's because the playing conditions continually change.
Quote: KeyserIt's also perishable in that, it appears to dissipate after 18 to 20 spins.
In most cases, the effect lasts for only five or so spins. It's because the playing conditions continually change.
In that case, the effect is illusory, in that it couldn't possibly be causal.
Quote: KeyserNo, quite real.
Then what is the causal link? What could possibly cause the increased frequency of numbers that have recently hit, but not continue to cause more hits, i.e., wear off, as time progressed?
Quote: Keyser2. Simply playing the last five numbers to have hit will slightly reduce the house edge, especially when the wheel speed is close to the speed of the previous spin. In rare cases, you can get a small edge on some wheels playing this way.
-Keyser
Would you mind clarifying this for me? How slight is the reduction in the HA? How often is it eroded to 0? It strikes me that 2.63% is a BIG HA to overcome, and I'm wondering how often the effect is great enough to actually clear that hurdle.
Also, it seems like what you have analyzed are specific circumstances affecting spins in the past. Is there any way for me to know that I have a small edge on some wheels by playing the last 5 numbers at the time that I'm playing them? When I play BJ and count cards, I know for sure when I have a small edge. Is there a way to achieve that same level of certainty with this Roulette method?
It is also true that wheel bias may cause some numbers to be favored over others. However, a) this effect would be very small (not nearly enough to overcome the house edge) and b) it would be very hard to determine at what point the observed nonstandard results were due to actual wheel bias, and not just random distribution: ANY sample of non-infinite magnitude will show SOME "bias" one way or another, even on a perfectly nonbiased wheel.
These are real wheels.
On real wheels, the physics behind the game matter.
Keep trying though, that's the fun part. Hope springs eternal.
Actually, Croupier (where did he go -- did he ever get back from the States?), a roulette Croupier did a controlled test and he was able to put the ball on one half of the wheel much more often than the other.
While I don't necessarily believe Keyser's claim of 1,000,000+ spins (40 spins / per hour = 25,000 hours = 12 full time years in a casino) with meticulous reasings, I do think (not that I matter) that the physics behind the game does matter and that biased results are possible.
We can make fun of mrjjj's methodology, but if anything, it could only be as good or better than any other betting scheme. If there is bias, it could manifest itself in like results.
Quote: KeyserA dealer could shoot sections, but only if the wheel has a dominant ball drop. It's the wheel that really makes it possible, not the dealer. There is no super human skill involved there.
Please note, I'm not saying that the dealer is the only reason for the effect.
And as I noted, I have employed the help of others to record a great deal of these numbers. I have also been at it far longer than just 12 years. I most certainly do have more than a million spins.
Why collect more than a million spins when a proven formula will give you the result for n spins?
Mere mentioning that you have some numbers, and have done some unspecified analysis on them doesn't make your claim any more believable than the next guy (and by "next guy" I mean mrjjj).
Quote: Keyser1. The data streams won't fit on the screen,
I did not mean paste it into a forum post.
Upload a file to some free service (I can give you a few links if you want), and post a link.
Quote:and I don't want to go through and edit the names of the casino locatons, etc. Too much work.
If you want to be taken seriously, you are going to have to do some work.
If a single find and replace command in a text editor or command line is "too much work" for you, I don't see how you can expect anyone to believe your claims about a decade of meticulously collecting and analyzing the data.
Quote:2. I can possibly do screen shots, but I'd need the permission of my programmer.
If you have a programmer, just ask him to strip the text from the files, it won't be anything to call "work" for him.
No need to bother with screeshots, they are useless.
Quote:As it is, there's too much sensative information on the screen. I also don't think it's a good idea to show everything on the screen.
What's the point of your posting here then? You are claiming that you have information, but it is too sensitive to be shared. Wel, if you don't want to share it, what is it exactly you intend to do with it then? Are you hoping to sell it here?
Quote:3. I could just post some of the stats, but I don't know if you would accept the results as they are posted.
The "results" without underlying data are worthless. In a way, you have already posted them here, in the form of your unsubstantiated claims. I just can't stop wondering do you really expect that somebody will just take your word?
Would you?
What if I told you, that I have a billion numbers, saying that your analysis is all wrong, and in fact you should be always betting on my birth date, and that gives you a considerable advantage over the house?
Here's a small example below.
30000 trials. (This is just one wheel. Each wheel and sample varies some. Overall, the effect is real.)(Wheel make is a Huxley Mark Series with moveable fret ring). (Date 2007 to 2009)(Right handed)
Betting the last seven numbers to have hit.
Edge reduces to -1.29
Max loss run 37
Max DD 4871
--------------------------
Betting the last five numbers to have hit.
Edge reduces to -1.79
Max loss run 52
Max DD 4209
------------
Betting the last three numbers to have hit.
Edge .02
Max loss run 85
Max DD 1555
-------------
Betting the last number only.
Edge 5.17
Max loss run 85
Max DD 1555
I have explained why it can and does happen and I have provided enough information for you to make an informed decision on your own. If you feel that you need further prove, then perhaps you should go out and collect some spins on your own and test them.
Quote: KeyserI feel that I have provided a satisfactory amount of information given the circumstances.
"the circumstances" seem to be that you're unwilling and unable to learn to use Google docs, and not much more than that.
Quote: KeyserI have explained why it can and does happen and I have provided enough information for you to make an informed decision on your own.
Your reasons were dirt and stickiness on the wheel, and croupier non-randomness. In both cases, there should be no serial correlation (let alone causation) that would make the previous 5 spins special. What does the croupier do on spin #11 that makes spin #8 more likely to repeat than spin #3? How does the dirt know the difference between spin #5 and spin #10? Wouldn't the filth or stickiness make one part of the wheel more likely to hit regardless of what the previous spins were? if 23, 35, 14, 2, and 0 are sticky but 34 just hit, aren't I hurting myself by betting on the 34 just because it's one of the most recent numbers?
*edited typo*
Quote:"Your reasons were dirt and stickiness on the wheel, and croupier non-randomness. In both cases, there should be no serial correlation (let alone causation) that would make the previous 5 spins special."
-Yes, actually there is. It's easy to test, and easily proven.
Quote: Keyser
-It's not always limited to just five spins. I think you will see that I mentioned that above. There are so many variables at work that the effect is not always persistent.
The dealer touching the pocket with wet or oily hands can poison not only the track, but the rotor as well.
" if 23, 35, 14, 2, and 0 are sticky but 34 just hit, aren't I hurting myself by betting on the 34 just because it's one of the most recent numbers?"
-Yes, sometimes you will lose. Your point? I would also not assume that an entire section is a good bet.
That'd be interesting to test based on your data. If we assume your data is sound, and there's no reason to doubt it, then further analysis should be able to determine answers to questions such as:
1) What is the observed probability of a number repeating, if not 1/38? How does that change over time?
2) What is the observed probability of a number and it's N neighbors repeating, if not N/38? How does that change over time?
If you look at N = 1..18 or so, you may find that a certain combination has different EV than others. In other words, assuming there is evidence of bias, and also assuming that the bias persists, what is the optimal way to take advantage of it?
--------------------------------
mrjjj is corrupting y'all. Back to reality!
Moving from -5.26% to the (-2.8 and -3.4% for five numbers) has it well beyond five standard deviations after a million trials. Dropping the house edge from -5.25% to just -2.7% would be close to 10 standard deviations, so just guessing, the five number is probably close to seven or nine standard deviations above the norm.
Quote: KeyserNo, we're not comparing apples to apples here and it's beyond the normal variance.
Moving from -5.26% to the (-2.8 and -3.4% for five numbers) has it well beyond five standard deviations after a million trials. Dropping the house edge from -5.25% to just -2.7% would be close to 10 standard deviations, so just guessing, the five number is probably close to seven or nine standard deviations above the norm.
But if that's true, you should be able to move from -2.7% to zero or very close to it. Are you suggesting that the same dust/sweat factors you observed don't appear on a 37-spot wheel?
Quote: KeyserI feel that I have provided a satisfactory amount of information given the circumstances
I have explained why it can and does happen and I have provided enough information for you to make an informed decision on your own. If you feel that you need further prove, then perhaps you should go out and collect some spins on your own and test them.
Well, if this is so, then you should drop everything, and head to the nearest casino right now, and bet there on 10 and 22, because I have provided just as much information to you that betting my birthday (10/22) gives you a large advantage over the house. If you feel that you have provided enough information, then I have provided enough to you as well (because I have provided as much), so what are you waiting for?
If you feel that you need further proof, just go out and collect some spins on your own!
Quote: KeyserDid you read the part where I said that I tracked several hundred different wheels?
Maybe there was a reason for that. :)
Well now, just a minute. If the bias is on real wheels, in real world conditions, and you then used several hundred different wheels, then you didn't actually explore the manufacturing/real world bias of any one wheel for 1,000,000 trials, did you? Assuming that all wheels got the same number of trials, and that "several hundred" means a minimum of 200, then no one wheel got more than 5000 trials.
I don't doubt your sincerity, or your results, but I disagree with your interpretation of your results. I think you are seeing patterns where there are none.
I am not advocating this use of this method as a means by which you can win back your farm, your tractor, your wife, or your dog.
You imply that every roulette wheel you've ever played at is defective. Is that what you believe?Quote: KeyserMy point is that past numbers that have hit on a real roulette wheel dealt by a live dealer do have a slight effect on the numbers that have yet to hit. The cause is related the the nonrandom effects of a defective gaming device being influenced intentionally or unintentionally by the dealer and other conditions at the table.
Quote: KeyserI am not advocating this use of this method as a means by which you can win back your farm, your tractor, your wife, or your dog.
LOL!! I concur. Nowhere on this post is Keyser saying he can buy 1/2 a restaurant with his findings!!
Quote: KeyserMy point is that past numbers that have hit on a real roulette wheel dealt by a live dealer do have a slight effect on the numbers that have yet to hit. The cause is related the the nonrandom effects of a defective gaming device being influenced intentionally or unintentionally by the dealer and other conditions at the table.
As best I can tell, that is not your point at all. Your point seems to be that past numbers can be indicative of a wheel bias that can temporarily predict results in the short term future... not that the numbers themselves have any influence. That is a very different point.
I contend that any bias, if it is as delicate and nuanced and ephemeral as you suggest (the oil on the dealer's fingers, for example) is infinitesimal and likely to have vanished (or changed) by the time it is discovered, if it is large enough to be discovered at all.
I'm not talking about showing results that you've already got; I'm talking about doing it in the future.
Quote: MoscaAs best I can tell, that is not your point at all. Your point seems to be that past numbers can be indicative of a wheel bias that can temporarily predict results in the short term future... not that the numbers themselves have any influence. That is a very different point.
I contend that any bias, if it is as delicate and nuanced and ephemeral as you suggest (the oil on the dealer's fingers, for example) is infinitesimal and likely to have vanished (or changed) by the time it is discovered, if it is large enough to be discovered at all.
The numbers hitting are just a mere reflection of the transient state of the gaming device and dealing proceedure.
Yes, you're correct when you state that it is evanescent at times, however, it also does not always dissipate as quickly as you might think. Regardless of whether it lasts for one spin or 10000 spins, it's still easy enough to measure.
-Keyser
Quote: MoscaYou know, this shouldn't be hard to prove. Get some folks together, name a time and a place, spin a wheel and pick winners. If you can do that to a degree exceeding statistical randomness, a number of times in a row, maybe you have something.
Mosca,
Yes, it shouldn't take to long to measure a change from -5.26% to -2.9%. We should be able to test this over a breif lunch break by spinning a quick 80k to 100k spins, while enjoying some chips and salsa. :)
Actually, reducing the house edge from 5.26% to 2.7% (Betting the last five numbers) within 100k spins would only be about 3.3 standard deviations.
Within a sample of that size a normal random fluctuation could potentially wash out the effect just long enough that we may need an additional 80k spins. This means we may need to play through dinner as well. :)
-Keyser
Again, understand that I'm not doubting your sincerity, nor your data, just your interpretation of it. Even if the effect is real (and I make no concession nor denial of this), I contend that it is too slight to be useful in real time, but only in retrospect.
Quote: MoscaIf you need that many spins to find it and take advantage of it, then it is useless. You said that the past five numbers are all that is needed. So you should be able to show it by batting the past five numbers.
I don't suggest it as a way to get an edge at roulette.
By the way, I'm glad someone brought up The Newtonian Casino; I read that long ago, it was published in the US as The Eudaemonic Pie. I loaned it out and never saw it again, it was out of print and now it's back. In college I knew someone who knew the actors involved in that, but it was long ago and during a smoke haze, all I remember is knowing about it. Then the book came out about 10 years later. I remember it as a good read.
Quote: KeyserI feel that I have provided a satisfactory amount of information given the circumstances.
Here's a small example below.
30000 trials. (This is just one wheel. Each wheel and sample varies some. Overall, the effect is real.)(Wheel make is a Huxley Mark Series with moveable fret ring). (Date 2007 to 2009)(Right handed)
Betting the last seven numbers to have hit.
Edge reduces to -1.29
Max loss run 37
Max DD 4871
--------------------------
Betting the last five numbers to have hit.
Edge reduces to -1.79
Max loss run 52
Max DD 4209
------------
Betting the last three numbers to have hit.
Edge .02
Max loss run 85
Max DD 1555
-------------
Betting the last number only.
Edge 5.17
Max loss run 85
Max DD 1555
I have explained why it can and does happen and I have provided enough information for you to make an informed decision on your own. If you feel that you need further prove, then perhaps you should go out and collect some spins on your own and test them.
Assuming your data is accurate, how do we know you didn't just take the best 30,000 results to base your claim? I'm curious about the other 970,000. Oh, but they are too "sensitive". Perhaps there is another person out there with a million spins to verify this?
Quote:I've been thinking about this more. I just don't buy it. You say that it's this little thing, or that; but if you can't duplicate it, then how do you know? You said the oil on the croupier's fingers; can you show a difference between spins with oily fingers and dry fingers? Until you can measure it, quantify it, replicate it, and use it to make predictions, it isn't there. All you have is data and a hypothesis to explain it. It may be correct, but you've stopped short of proving your point.
I didn't say that you or I can't duplicate it. I have measured it, I have quantified it, and I have replicated it. As far as using it to make predictions, here you go: Bet the last five numbers and use a rolling horizon.
Good luck. :)
-Keyser
Quote: KeyserBet the last five numbers and use a rolling horizon.
Keyser
a rolling horizon?
English is not my first language.
I thought I could read and understand it well,(except for technical scientific stuff) boy am I so wrong. This thread has my head spinning. I need to find someone to translate to me.
If one bets the last 5 numbers on a 00 wheel the probability of the very next spin to be one of the 5 numbers is (and it is an average)12.4834%. for a single 0 wheel: 12.8025%
One must calculate all the possible ways the last 5 spins can be distributed then take a weighted average.
An easier way is to make a cumulative frequency table using p=1/38 or p=1/37. your choice.
formula: 1-((1-p)^n) where n= # of last numbers spun.(did I explain that correctly?)
example "00 wheel": 1-((37/38)^5(last 5 numbers))= 0.124833597 or 12.4834%.
last 6 numbers: "00Wheel"=1-((37/38)^6)=0.147864291 or 14.7864% / "0Wheel"=1-((36/37)^6)=0.151592185 or 15.1592%
(This is the median)last 26 numbers: 00Wheel0=50.0114% / 0Wheel=50.0114%
last 42 numbers: 00Wheel0=67.3742% / 0Wheel=68.3603%
last 87 numbers: 00Wheel0=90.1740% / 0Wheel=90.7792%
Quote: KeyserGood luck. :)
-Keyser
Thank you.
I need all the luck I can get!
Maybe you publish a book about all this in the very near future?
Quote: KeyserI didn't say that you or I can't duplicate it. I have measured it, I have quantified it, and I have replicated it. As far as using it to make predictions, here you go: Bet the last five numbers and use a rolling horizon.
Good luck. :)
-Keyser
:) Oh jeez, no; not me, with my own money! I don't agree with you, and it's not high enough on my radar for me to care enough about to pay for an answer; right now it's just an interesting discussion on the internet.
Regarding the hypothesis, and the story in The Eudaemonic Pie, any cell phone has far more than enough computing power work the numbers. All that's needed is an app. Casinos would have to ban cell phones... see how far that would get. But I think I recall reading that they took steps to randomize the spins more. But again, that was 25 years ago; something I filed as "solved" without remembering the details.
So, while I applaud your hypothesis and research, I still don't find it complete.
So I guess my point is that the idea that there are biased wheels out there is not new.
Quote: MoscaBy the way, I'm glad someone brought up The Newtonian Casino; I read that long ago, it was published in the US as The Eudaemonic Pie.
Are you referencing something similar here?
Quote: Moscaany cell phone has far more than enough computing power work the numbers. All that's needed is an app. Casinos would have to ban cell phones... see how far that would get.
No, the law is already there. Using a wheel-clocking app on a cell phone makes it an illegal "device for calculating probabilities":
Quote: NRS 465.075Use of device for calculating probabilities. It is unlawful for any person at a licensed gaming establishment to use, or possess with the intent to use, any device to assist:
1. In projecting the outcome of the game;
2. In keeping track of the cards played;
3. In analyzing the probability of the occurrence of an event relating to the game; or
4. In analyzing the strategy for playing or betting to be used in the game,
Ê except as permitted by the Commission.
Moreover, it's a felony to use the app and apparently even to write it:
Quote: NRS 465.088Penalties for violation of NRS 465.070 to 465.085, inclusive.
1. A person who violates any provision of NRS 465.070 to 465.085, inclusive, is guilty of a category B felony and shall be punished:
(a) For the first offense, by imprisonment in the state prison for a minimum term of not less than 1 year and a maximum term of not more than 6 years, or by a fine of not more than $10,000, or by both fine and imprisonment.
(b) For a second or subsequent violation of any of these provisions, by imprisonment in the state prison for a minimum term of not less than 1 year and a maximum term of not more than 6 years, and may be further punished by a fine of not more than $10,000. The court shall not suspend a sentence of imprisonment imposed pursuant to this paragraph, or grant probation to the person convicted.
2. A person who attempts, or two or more persons who conspire, to violate any provision of NRS 465.070 to 465.085, inclusive, each is guilty of a category B felony and shall be punished by imposing the penalty provided in subsection 1 for the completed crime, whether or not he or she personally played any gambling game or used any prohibited device.
Quote: KeyserWhy on earth would you need to use a device at the table?
FYI, You are, however, allowed to use a pen and paper at the table, and you can enter your data into your home computer without breaking any laws. Just betting the last number or last five numbers to have hit doesn't require a computer.
Just betting the last five numbers won't beat the game, either. Clocking the wheel will, but basically nobody can do it without electronics. With electronics, as Eudaemonic Pie demonstrated, it's very feasible to beat the roulette wheel.
Quote: KeyserJust betting the last number or last five numbers to have hit doesn't require a computer.
It doesn't require anything. The casino keeps track for you on the electronic board at every table.
Quote: MathExtremistJust betting the last five numbers won't beat the game, either. Clocking the wheel will, but basically nobody can do it without electronics. With electronics, as Eudaemonic Pie demonstrated, it's very feasible to beat the roulette wheel.
MathExtremeist,
Let me explain the phrase, "Clocking the Wheel". It really has two meanings.
