Craps Strategy Investment Opportunity?
Quote: MichaelBluejayMy programming language of choice for this type of task is XTalk, because it's really fast to develop with.
put random(6) + random(6) into theThrow
Sorry to rain on the parade, but I would strongly recommend at least doing a basic statistical test on XTalk's "random()" method. Most scripting languages don't pay attention to the subtle implementation details that go along with RNGs. At a minimum, I recommend maybe a 10M trial on whether the distribution of random(6) is sufficiently uniform. If it's not, XTalk isn't a suitable technology for this project. Ideally you'd be using a recognized RNG, like MT19937 or a MWC generator, and then you'd map to your target range [1..6] using a bias-free mapping. But it might be more trouble than its worth to write one in XTalk vs. just using C/C++, Java, or C# with a pre-tested RNG implementation.
Quote: MathExtremistQuote: MichaelBluejayMy programming language of choice for this type of task is XTalk, because it's really fast to develop with.
put random(6) + random(6) into theThrow
Sorry to rain on the parade, but I would strongly recommend at least doing a basic statistical test on XTalk's "random()" method. Most scripting languages don't pay attention to the subtle implementation details that go along with RNGs. At a minimum, I recommend maybe a 10M trial on whether the distribution of random(6) is sufficiently uniform. If it's not, XTalk isn't a suitable technology for this project. Ideally you'd be using a recognized RNG, like MT19937 or a MWC generator, and then you'd map to your target range [1..6] using a bias-free mapping. But it might be more trouble than its worth to write one in XTalk vs. just using C/C++, Java, or C# with a pre-tested RNG implementation.
When I was first programming a craps simulator in "C", I found that the built-in Standard C Library RNG was not producing uniform results, so I bought a set of math functions from "Numerical Recipes" that filled the bill.
Cheers,
Alan Shank
Woodland, CA
I would also perform a chi-squared test on the distribution of all results. If don't know how, send me the results of how often each outcome occurred, and I'll do it.
Also, I thought that if one re-seeds the RNG frequently there would be no need to be concerned about the quality of the random numbers produced.
I have used WinCraps in the past as it is very easy to program RNG changes while running simulations and one can save the rolls created to its own file to be used again and again so re-seeding is not again needed for the same sequences to be produced.
Now,
If the challenge is stuck on the idea that 5 different programmers using the same dice roll file all come out with different results,
then it seems to me the "system" may not be totally understood by all the programmers. Sure errors can be made but simple tests should find the errors in the code, unless the "trees" that are in the "system" are very large and difficult to map.
My solution would to have at least 2 different programmers using the SAME software (WinCrps) or language, running a test sim using the same dice roll numbers and not continuing until both have identical results.
Kind of like a science experiment is conducted.
example 1:
Bet$5 on pass line
When pass line has lost 1 time
Then bet $10 on pass line
etc
simple system simple tree. I have a feeling that 98steps system is not so simple.
example 2:
If x happens
then do this
else do this
If x happens between a and b values
then do this for a if true
then do this for a if false
then do this for b if true
then do this for b if false
If c>5
then do this for a if true
then do this for a if false
then do this for b if true
then do this for b if false
else go to "d"
The more complex the "system" the more complex the tree becomes
Quote: guido111Then if this challenge is not moving forward because of RNG concerns, why not create a 100 billion dice roll file that everyone agrees is random?
Also, I thought that if one re-seeds the RNG frequently there would be no need to be concerned about the quality of the random numbers produced.
I have used WinCraps in the past as it is very easy to program RNG changes while running simulations and one can save the rolls created to its own file to be used again and again so re-seeding is not again needed for the same sequences to be produced.
According to the WinCraps Help file, "The pool of numbers that each RNG (except #3) can produce is so large that it can span any number of games you could hope to play in your lifetime." There is a potential danger in re-seeding between games (sessions), and no need to do so.
Cheers,
Alan Shank
Woodland, CA
Quote: guido111Then if this challenge is not moving forward because of RNG concerns, why not create a 100 billion dice roll file that everyone agrees is random?
Also, I thought that if one re-seeds the RNG frequently there would be no need to be concerned about the quality of the random numbers produced.
I have used WinCraps in the past as it is very easy to program RNG changes while running simulations and one can save the rolls created to its own file to be used again and again so re-seeding is not again needed for the same sequences to be produced.
The Wizards idea seems to be an easy solution to remove the concern about numbers produced by a RNG.
98steps is fine with the idea of re-seeding at the beginning of each session and running 100 7-outs before any wager is made.
Quote: guido111
My solution would to have at least 2 different programmers using the SAME software (WinCrps) or language, running a test sim using the same dice roll numbers and not continuing until both have identical results.
Kind of like a science experiment is conducted.
Very good idea and that has already been done. A friend of mine has programmed 98 steps independently of me, and his results using the same dice roll file came out with results far from mine. We both used WinCraps, his code was done differently than mine so it takes me time to go thru his code to "see" how he accomplished a "system rule".
I found 4 errors, but he says they are not errors, just different ways of doing the same thing.
WinCraps allows one to "show steps during play" meaning as a roll is made it steps thru each line of code showing what is to be done. Again, another time consuming chore.
In the end, it will be 98steps final decision on how and if to continue with the challenge.
And the "system" is not that difficult to understand or even play at a table, I have done that myself, it is my concern can it be programmed exactly as designed and work always with 100% accuracy in a simulation
Quote: guido111Then if this challenge is not moving forward because of RNG concerns, why not create a 100 billion dice roll file that everyone agrees is random?
Also, I thought that if one re-seeds the RNG frequently there would be no need to be concerned about the quality of the random numbers produced.
Just the opposite - a good RNG is designed to be seeded once and then iterated over and over without reseeding at all (ever). Reseeding the RNG puts it into one of a relatively small number of initial states, usually 2^32, vs. letting it run over a (potentially) much larger cycle.
The point of seeding the RNG is to get it started, but also to enable you to replicate its results later. In other words, the same seed + the same RNG = the same sequence of random draws. That's why it's "pseudo" random.
There's nothing inherently wrong with the idea of a pre-generated roll file, but you'd need on the order of 100B bytes (packing each two-dice combo into one byte). That's about 93Gb. Too big.
Quote: WizardI have to agree with the concerns about the concerns about the random() RNG. What I would recommend going is doing random(1000) and saving the first 10 results. Then keep drawing more, keeping a count, until you get the same 10 again in the same order. You should go to at least a count of 2^32.
I would also perform a chi-squared test on the distribution of all results. If don't know how, send me the results of how often each outcome occurred, and I'll do it.
I'd be concerned that the internal mapping of RNG bits to integers shows modulo bias, too. To test this, do a distribution on random(6) and another on random(8). 8 is an integral power of 2, so the common modulo bias won't be there. If chi-squared on random(6) is worse than chi-squared on random(8), then you'll have to map random(8) to dice rolls properly. You can ignore the modulo bias if you want -- it's typically very small, but it's usually present. I wouldn't recommend knowingly put together a sim that exhibited it, especially with money on the line, and especially without knowing the internals of your RNG in the first place.
Quote: 98stepsThe system re-write in "grandma" is complete. I have sent it to the resident english teacher for review. I hope to submit it to Mr. Bluejay later today.
The writeup is relatively clear, although a couple of terms need to be better defined. I think it should be sufficient for Mr. Bluejay's purposes, though. The methodology is fairly well laid out.
Since the strategy specifies some fairly quick stop-play points, I think it would have to be stipulated that play is assumed to resume again immediately after the stop-play, or reset, points have been reached. Of course, that would be with the base bet amounts, after taking down all previous action. This raises the question of when the stop point is reached, but a bet remains that cannot be taken down, are the "extra" rolls necessary to resolve that bet considered part of the one billion rolls?
Quote: mkl654321The writeup is relatively clear, although a couple of terms need to be better defined. I think it should be sufficient for Mr. Bluejay's purposes, though. The methodology is fairly well laid out.
Since the strategy specifies some fairly quick stop-play points, I think it would have to be stipulated that play is assumed to resume again immediately after the stop-play, or reset, points have been reached. Of course, that would be with the base bet amounts, after taking down all previous action. This raises the question of when the stop point is reached, but a bet remains that cannot be taken down, are the "extra" rolls necessary to resolve that bet considered part of the one billion rolls?
one billion rolls?
No 1 billion rounds.
98steps system is played in "sessions" with different stopping criteria for a session, either win or lose.
So a session can have many rounds and even more rolls.
A round is counted when, on any roll, there is at least 1 or more resolved wagers.
And any contract wagers (pass line and don'pass line bets to be specific) left when a session stop is reached still must be resolved.
and there can be up too 100 7 outs at the start of every new "session" without any wagers being in action
He may at a later date ask for improvements, but that would be his option, not ours to exercise.
We are at a point where Mr Bluejay wants to see and understand the system so it can be programmed.
Quote:Then if this challenge is not moving forward because of RNG concerns...
When the HECK did I ever say that?! When did 98 ever say that? Are you either of the parties involved here? No? Then where the heck did you get the idea that the challenge isn't moving forward?
Next, I think all the concerns about the RNG are way, way overblown. Even a run-of-the-mill RNG is appropriate for this kind of test. An RNG doesn't have to meet a gold standard, it just has to be *good enough*. Even if 98 were trying to game the RNG (which he isn't), that wouldn't be a concern, because my rules specifically disallow that kind of chicanery. And there are easy ways around concerns that a sequence might repeat. An obvious way is to pick and discard a random number of numbers after a random number of rounds. (Slot machines work on a similar principle -- they're constantly picking numbers even when you're not hitting the Spin button, so no possible sequence can be observed.) Oh, you think this isn't fair because discarding trows is unlike real life? Well then, you've never seen a misthrow in a live craps game?
About number distribution bias, I think this is a problem only in theory, not reality. Even with some bias, *in a sim like this* over 1B rounds the difference a regular vs. gold-standard RNG would return would almost certainly be negligible.
But you know what? Before people even raised RNG concerns here, I ran a test where I picked 1B roulette numbers and the first 20 never repeated. I just ran it again picking 4.4 B numbers (to exceed 2^32 numbers, or 4,294,967,296) and it didn't repeat, either. (Which surprised me, I thought the period would be 2^32.) In 1B challenge sim I would easily be picking more than 4.3 B numbers, since I'm picking 2 numbers each roll, and it will take multiple rolls to resolve each bet, and the system can sit out up to 100 resolved Pass Line bets between wagers, but again, I think this *soooo* does not matter.
Finally, I simulated 100 million dice rolls (10x more than suggested) and here were the results:
1: 1667514
2: 1667513
3: 1667511
4: 1667512
5: 1667515
6: 1667513
Is that close enough for ya? :)
It is not about RNGs.
That is not even an issue.
Mr. Bluejay NEEDS the "grandma" version to continue. Then he can $$$, escrow, program, simulation.
Let us not make this thread a "soap opera".
I hope it happens soon
Quote: 7crapsI think it is best for those of us that can see 98steps "system" not to critique any point of the system or offer to change a way a bet is made. That would not be his system then..
Nobody's doing that. There are questions of DEFINITION that need to be resolved, so that the test uses agreed-upon parameters.
I am certain of the outcome of the test, and don't think for a second that the test is even necessary, but if it's going to happen, and especially if there's money ridiing on it, the way it's going to be done should be scrupulously defined by all parties.
Quote: MichaelBluejayAbout number distribution bias, I think this is a problem only in theory, not reality. Even with some bias, *in a sim like this* over 1B rounds the difference a regular vs. gold-standard RNG would return would almost certainly be negligible.
Or to put in another way, what is the probability that even a flawed or biased RNG would produce results aberrant enough to show, after 1 billion trials, a result that significantly deviated from expectation? My guess would be, "essentially zero probability".
Quote: MichaelBluejay
Finally, I simulated 100 million dice rolls (10x more than suggested) and here were the results:
1: 1667514
2: 1667513
3: 1667511
4: 1667512
5: 1667515
6: 1667513
Is that close enough for ya? :)
The issue is that the results are too smooth. Note how the results for each face fall between 1,667,511 and 1,667,515. That is a difference of 4. It is not normal for ten million fair die rolls to be that smooth. The chi-squared statistic on that distribution is 0.000006 with 5 degrees of freedom. The probability of a distribution that smooth is off the charts. Less than 1 in a billion.
Here are the results of my own ten million roll simulation:
| Die | Rolled |
|---|---|
| 1 | 1667573 |
| 2 | 1667877 |
| 3 | 1665406 |
| 4 | 1666397 |
| 5 | 1667232 |
| 6 | 1665515 |
| Total | 10000000 |
Note the difference between maximum and minimum observations is 2,471. This distribution has a chi-squared statistic of 3.356571 with 5 degrees of freedom. The p value is 0.645192. That means the probability of a truly random simulation of ten million die rolls would have a 64.5% chance of having more variance, and 35.5% of having less variance. So I submit that this is an example of what a fair test of ten million rolls would look like.
Now, maybe your RNG is still good enough to test this system. I would defer to MathExtremist on that.
98 just sent me the description of the betting system, and it's exceptionally easy to understand. I'm ready to execute a contract. As promised, I'll post my proposed contract wording below to get input from the community (and especially from 98steps) before I submit it to an attorney for review. I tried to make it as generic as possible, and put the exceptions to the generic version all the way at the end, so I can use this contract again in the future if need be.
Incidentally, the betting system is extraordinarily complex. I'm skeptical that the average person could actually execute it in a live casino. But I'll concede that point because it doesn't make the system any likelier to win, in my opinion.
Quote:(1) Parties. This contract is between Michael Bluejay and Christiaan Krebs.
(2) The Wager. Krebs will supply Bluejay with the rules for a gambling betting system (with limits described below), and Bluejay will program a computer simulation to test it. If the system shows a profit after 1 billion simulated rounds, Bluejay will pay Krebs $10,000. If it does not, Krebs will pay Bluejay $1000. A round is any roll or hand in which at least one wager is resolved.
(3) Money. Each party will deposit his funds with a mutually-agreeable third party who will also act as judge. The judge and escrow agent is Michael Shackleford ("The Wizard of Odds").
(4) System Rules. The betting system must:
. . . . (a) Be based on a game of craps, roulette, or baccarat, as the player (not the casino), using common (not exceptional) Las Vegas rules. In craps, free odds allowed are 3-4-5X and commissions on Buy and Lay bets are paid only on wins. In roulette, single-zero is allowed.
. . . . (b) Utilize a betting spread of no more than 1 to 500. (The largest bet can be no more than 500 times the smallest bet.)
. . . . (c) Place a bet at least once every 100 rounds. While "sitting out" and not betting, a sitting-out round is the challenger's choice of:
................(1) 100 completely resolved Pass Line wagers, 100 seven-outs, or up to 5 consecutive successful Pass Line or Don't Pass wagers (Craps).
................(2) 100 spins, 50 reds, 50 blacks, 5 consecutive reds, or 5 consecutive blacks (Roulette).
................(3) 100 hands, 50 Banker results, 50 Player results, 5 consecutive Banker results, or 5 consecutive Player results (Baccarat).
. . . . (d) Be playable by an average person in a typical casino, with no special equipment required other than a pen and paper.
. . . . (e) Be a bona-fide betting system where bet timing and/or amount is based on previous wins or losses. Any attempts to exploit the computer's RNG (random number generator) or a loophole in these rules, rather than trying to prove the viability of a betting system for use in a casino, is specifically disallowed.
. . . . (f) Be explained clearly.
(5) Forfeiture. Once the contract has been signed, any party who backs out will forfeit. Krebs forfeits if he does not submit a betting system as described above to Bluejay within one week of the signing of this contract. Bluejay forfeits if he does not supply Krebs with the results of the simulation within one week of receiving the system from Krebs.
(6) Bluejay's promises. Bluejay asserts that:
. . . . (a) He has not yet programmed the simulation, so he is not privy to the simulation's results.
. . . . (b) If the system proves to be a winner, he promises not to use the system for profit.
. . . . (c) If the system proves to be a winner, he will publish a prominent statement of such on the front page of Easy.Vegas, for a period of at least one year from the date of the results, along with a prominent link to the sale website if the system is for sale on the web.
. . . . (d) He will not disclose Krebs' name, contact information, the specific rules of the system, or statistics about the simulation other than the final results, without the challenger's permission. He may, however, at his option:
.................(1) Name the system and the seller if the system is available for sale to the public, and link to any website where the system is for sale.
.................(2) Describe the system in general terms. (e.g., "This is a standard positive-progression system," "This system is based on betting on red after a certain pattern of red/black hits has been established," etc.)
.................(3) Publish the final results of the simulation (win/lose, percent of wagered money lost).
(7) Recourse for Krebs. Prior to running the complete simulation, Bluejay will provide the source code and sample output to Krebs, and Krebs will have one week to dispute the validity of the code (to allow time for his own experts to verify the source code). If Krebs does not explicitly dispute the code, Bluejay will run the full test and the judge will decide its validity. If Krebs disputes the code, he must immediately point out why he thinks the source code is flawed. If Bluejay agrees that the code is in error then he will fix the code and provide the new code and sample output to Krebs. The one-week acceptance/repair period will repeat as many times as Krebs disputes and Bluejay agrees that the code is in error. At any point that Bluejay disagrees that the code is in error, Krebs' initial recourse shall be to appeal to the judge. If the judge believes the code is valid, Krebs may submit case to a professional arbiter, agreeable to both parties, with the losing party in the arbitration paying the arbiter's costs.
(8) Recourse for Bluejay. If Bluejay feels that the system submitted by Krebs does not meet the System Rules specified here, he may appeal to the judge to declare a forfeit. If the judge declines to forfeit, Bluejay may submit the case to a professional arbiter, agreeable to both parties, with the losing party paying the arbiter's costs.
(9) Amendments.
. . . . (a) Bluejay agrees that the system shared by Krebs on Oct. 2, 2010 satisfies the System Rules conditions.
[My notes to change for the generic version:
(1) Limit to 50 seven-outs.
(2) Provide source code after.]
In case of a dispute, it also seems unfair to unilaterally put the burden of payment to the arbiter on Krebs. The cost of the arbiter should be paid by the LOSER of the dispute (once again, it would almost certainly be Krebs, but that is not relevant).
The possibility also exists of Krebs simply surrendering at that point, due to the cost of finding a suitable arbiter (who would, after all, have to be both a computer programmmer AND conversant with contract law, thus, expensive), which would resolve the wager but not the underlying dispute. This would not be an ideal outcome.
The contract otherwise seems very well thought out. I am just troubled by the fact that the whole process seems to shift the burden of proof to the correctness of the source code used to run the simulation and away from the correctness of the system.
Quote: MichaelBluejay
(9) Amendments.
. . . . (a) Bluejay agrees that the system shared by Krebs on Oct. 2, 2010 satisfies the System Rules conditions.
. . . . (b) Krebs agrees that if the system fails, Bluejay may publish the rules for the system.
My, my.
98steps if you are looking for any advice.
(b) should not be agreeable to you.
You may fail the challenge but would like the results to be for you only. It still may be valuable to you for what ever purpose.
Reason:1 Billion rounds is unfair, because no one EVER could actually accomplish it.
Bluejay wants you to meet rules that can actually be done in a casino but yet can never be done by someone in their lifetime.
It would take you well over 3000 years to actually play 1 Billion rounds if your bet was the pass line bet playing 24/7. impossible, could never happen in this universe.
It would take you well over 1000 years to actually play 1 Billion rolls if your bet was the field bet playing 24/7.
I vote to keep the rules to yourself.
Quote: mkl654321
The contract otherwise seems very well thought out. I am just troubled by the fact that the whole process seems to shift the burden of proof to the correctness of the source code used to run the simulation and away from the correctness of the system.
I having not seen 98 system, I know Wincraps and so does 98 steps.
I have seen the code that Bluejay would use, and I could not even come close to seeing if there was an error.
7Craps mentions in an earlier post his same concern. And even states he and another programmed the system in Wincraps and using the same dice roll file, came up with different results.
Now I really want to see this system!
98steps, tell me when your book will be out!
Quote: mkl654321As much as I hate to acknowledge it, this places an undue burden on Krebs. ... Krebs will not accept the results, arguing that your coding methodology is flawed---but how could he possibly detect a specific flaw, even if there was one? I know enough about programming to know that quite frequently, even an expert in the same programming language cannot unravel another's code. Yet, Krebs would have to be able to detect the putative error, while lacking the expertise to do so.
Hello, having a recourse *available* to you is not a BURDEN! No action is *required* of Krebs, thus no burden is imposed. If he wants to *exercise* the *option* to dispute which I *provided*, then the only fair way to him to object is to show me what I did wrong. Of course being a non-programmer, he'll have to get a programmer to review the code for him to find any errors. You might consider that a burden, but it's absolutely not an UNDUE burden.
Because if it's really "undue", then what kind of recourse am I supposed to offer him instead? That h can just make a blanket objection to the results without specifying what's wrong with my code? Not-uh, no dice.
Quote: mkl654321In case of a dispute, it also seems unfair to unilaterally put the burden of payment to the arbiter on Krebs.
Your opinion. If the judge agrees that I've won, then why should *I* have to pay for Krebs to challenge that ruling? I won't.
And likewise, you'll see that if *I* request a forfeit and the judge declines, then *I* have to pay for the arbiter. Whoever disagrees with the judge pays the arbiter. Fair's fair.
Quote: guido11198steps if you are looking for any advice. (b) should not be agreeable to you. You may fail the challenge but [may] like the results to be for you only.
This is certainly a negotiable point, which is why I put it in the Amendments rather than the general contract. Given the extreme interest in 98's system, I'm hoping he will allow the system to be shared if it proves to be worthless. But that's his call, and if wants to keep it secret no matter what the results then I'll remove that amendment if he asks me to.
By the way, I'm making some minor wording changes to the posted contract.
Quote: MichaelBluejay
Quote: guido11198steps if you are looking for any advice. (b) should not be agreeable to you. You may fail the challenge but [may] like the results to be for you only.
This is certainly a negotiable point, which is why I put it in the Amendments rather than the general contract. Given the extreme interest in 98's system, I'm hoping he will allow the system to be shared if it proves to be worthless.
worthless to you maybe.
Not worthless to 98steps.
But if it ends up failing after 20 million rounds it is of great value to 98 since that is more likely to be accomplished in ones long lifetime.
edit:3 million dice rolls seems to be a good lifetime number that any person could realistically attain in their life.
one mans junk is another mans treasure.
98steps seems to have his head screwed on straight. He does not cuss out any one or even show anger in his posts. I think he knows what he wants to do.
let the show begin!
Quote: MichaelBluejayHello, having a recourse *available* to you is not a BURDEN! No action is *required* of Krebs, thus no burden is imposed. If he wants to *exercise* the *option* to dispute which I *provided*, then the only fair way to him to object is to show me what I did wrong. Of course being a non-programmer, he'll have to get a programmer to review the code for him to find any errors. You might consider that a burden, but it's absolutely not an UNDUE burden.
Because if it's really "undue", then what kind of recourse am I supposed to offer him instead? That h can just make a blanket objection to the results without specifying what's wrong with my code? Not-uh, no dice.
Your opinion. If the judge agrees that I've won, then why should *I* have to pay for Krebs to challenge that ruling? I won't.
Whoa, Nelly. Unwarranted vehemence. You asked for opinions on the contract, I gave you some. Get out of my throat--you're choking me.
I'm simply refrring to the inevitable outcome. Your simulation will show the system to be a loser. Krebs will challenge it. This will happen surely as the night doth follow the day.
There is an asymmetry of information here. You KNOW whether or not your simulation is valid; Krebs has no idea one way or another, and he won't know even after he sees the source code. Therefore the burden of proof lies on you to demonstrate that said source code is valid. I know that if I had $1000 at stake, I'd want to see such validation.
Since you seem to have missed it, I'll say it again: the LOSER would pay for arbitration. If you thought you were right, then what would be your objection to this? I said that you would only have to pay for the arbitration if you were WRONG.
You may feel that you're working from the high ground here, but if you truly want this to go through, you'll have to level the playing field. As written, Krebs probably won't sign this contract.
Quote: mkl654321Whoa, Nelly. Unwarranted vehemence. You asked for opinions on the contract, I gave you some. Get out of my throat--you're choking me.
I'm simply refrring to the inevitable outcome. Your simulation will show the system to be a loser. Krebs will challenge it. This will happen surely as the night doth follow the day.
mlk makes a great point.
and you did ask for opinions.
reality is reality.
I think 98 will take the challenge.
He has so much to gain even if he could show a profit after 3 million dice rolls.
mlk, you willing to take bets? LOL!
Quote: guido111
Reason:1 Billion rounds is unfair, because no one EVER could actually accomplish it.
Bluejay wants you to meet rules that can actually be done in a casino but yet can never be done by someone in their lifetime.
It would take you well over 3000 years to actually play 1 Billion rounds if your bet was the pass line bet playing 24/7. impossible, could never happen in this universe.
It would take you well over 1000 years to actually play 1 Billion rolls if your bet was the field bet playing 24/7.
The point of the one billion rounds is not that anyone could actually play that much; the issue is whether any system can provide a positive EXPECTATION, and the billion rounds are really overkill, but if a system has a positive expectation, it will show up in a billion rounds or many FEWER.
In my experience with WinCraps, 10,000 sessions of 200 rolls bring out the rough expected results, but there are certainly differences from theoretical expectation; for example, some runs of that size will show an HA for line bets of 1.3%, some 1.5%. In the case of 98 Steps' system there seem to be enough bets with HA and/or vig involved that I would be quite surprised to show a net gain even over 10,000 sessions, but we'll see - maybe.
Cheers,
Alan Shank
Woodland, CA
mkl64321, you talk about this alleged burden, but you don't suggest an alternative...even though I asked. I'm giving 98 a way to challenge the results. You think that's burdensome because he's not competent to challenge my code. Okay, so in your universe, how exactly *would* 98 challenge the results?
As for this...
Quote:You KNOW whether or not your simulation is valid; Krebs has no idea one way or another, and he won't know even after he sees the source code. Therefore the burden of proof lies on you to demonstrate that said source code is valid.
No, it doesn't. The judge decides whether the test was conducted fairly or not. 98 has the option of challenging that ruling if he wants. That he might not *personally* understand the code is not my fault. Nor my problem.
Quote:Since you seem to have missed it, I'll say it again: the LOSER would pay for arbitration. If you thought you were right, then what would be your objection to this? I said that you would only have to pay for the arbitration if you were WRONG.
Oh, I thought you meant the loser of the wager, not the loser of the arbitration. Okay, this is a really good suggestion. I'll edit the contract.
Quote: mkl654321Your simulation will show the system to be a loser. Krebs will challenge it. This will happen surely as the night doth follow the day.
The one guy who accepted my betting system challenge didn't argue when I told him he lost. Not for one second. Here is the whole story.
Thanks all for this most entertaining thread..
Quote: SOOPOOThe contract is fair as written. I would sign it if i was 98steps. Having read the many posts back and forth I fully expect 98steps to be a gracious loser. Of note, when I use the word loser I just mean that his system will fail the trial.
Because 1 billion rounds is an overkill as stated by goatcabin earlier, something that is not realistic. No one can play 1 billion rounds of any casino game in a real casino in a lifetime. But some could see 2 to 3 million dice rolls in a lifetime.
65 years (X) 12 trips to Vegas per (X) 32 hrs at a craps table (X) 120 rolls per hour = almost 3 million dice rolls seen.
So, then a better question is can his system, in 1 million challenges, how many challenges will show a "win" after 3 million dice rolls.
But then again Mr. Bluejay's rules state and unlimited bankroll can be used.
Do not think we can find that to hold true for any gambler.
Quote: SOOPOOOnce you have his system, will you be able to provide us with the chances of his system being ahead at, say, 1000, 100000, .....1 billion rolls?
Thanks all for this most entertaining thread..
It is not up to Mr. MB to determine that unless 98steps agrees to it.
I think Mr. Bluejay states in the contract:
He will not disclose Krebs' name, contact information, or the specific rules of the system without the challenger's permission. He may, however, name the system and the seller if the system is available for sale to the public, and may describe the system in general terms. (e.g., "This is a standard positive-progression system," "This system is based on betting on red after a certain pattern of red/black hits has been established," etc.)
I have offered my opinion to 98steps about not posting the rules or even the specifics of the results unless he wants to.
Mr. Krebs knows what he is doing.
Lets say, for example:
The results show for every 20 years of realistic play, his system shows a 65% win rate, wth a total net bankroll increase.
Some may see value with those numbers, others may not. Some may believe the results yet others will not.
Just because one person says something is "worthless" does not mean it "is" worthless to everyone.
98steps system, based off of the D'Alembert betting system and variations of it is nothing new to the world.
Member "goatcabin" for example, and I am not picking on him, only that he has shown many good simulations here at the forums,
has posted results of different styles of play in the game of Craps. NONE of his results shows a gambler can win 1 billion rounds.
But it does show, that sessions can be won for a specific amount of time.
Alan would agree that it is possible for someone to play for 70 years and die showing a positive net outcome from gambling.
But that should be the subject of another thread.
Quote: goatcabinThe point of the one billion rounds is not that anyone could actually play that much; the issue is whether any system can provide a positive EXPECTATION, and the billion rounds are really overkill, but if a system has a positive expectation, it will show up in a billion rounds or many FEWER.
In my experience with WinCraps, 10,000 sessions of 200 rolls bring out the rough expected results, but there are certainly differences from theoretical expectation; for example, some runs of that size will show an HA for line bets of 1.3%, some 1.5%. In the case of 98 Steps' system there seem to be enough bets with HA and/or vig involved that I would be quite surprised to show a net gain even over 10,000 sessions, but we'll see - maybe.
Cheers,
Alan Shank
Woodland, CA
Points well taken.
Then who sets the rules as to what is "worthless"?
After 500,000 rolls or 1 million rolls, 2,3 4 million is a more realistic rule? Less worthless?
But the rule using an unlimited bankroll, now that is not realistic. I say that rule of the challenge is "worthless" because any results from it would not be realistic.
Hey it is football time!
Quote: WizardFor those who think a billion bets is unrealistic, think of a billion bet simulation as the average results of a 10,000 people using a system for 100,000 bets.
Nice observation. I had not thought about it in that sense.
Then would 1 billion people using a system for 1 bet also be the same thing?
My point only was that NO single person could ever see 1 billion rounds in a casino in a lifetime.
I would make that a challenge for those that disagree with me.
Quote: 7craps
But it does show, that sessions can be won for a specific amount of time.
Alan would agree that it is possible for someone to play for 70 years and die showing a positive net outcome from gambling.
But that should be the subject of another thread.
That is a good idea for a new thread.
After football today, which is "WAY MORE IMPORTANT", I will start one about the chances of gamblers who play for a lifetime and their probabilities at death to have a net win from gambling.
I like things that are more realistic.
NEW THREAD HERE
1B resolutions is the number decided to eliminate any short term variance, which is what the challenge is about. Thus it is appropriate.
Now, from all the chatter, it sounds like 98steps may have a method of placing bets that may result in a distribution of more positive outcomes, one that may work the majority of the time in a lesser number of resolutions. This of course could be of great interest to the rest of us, but that is a separate matter from the challenge itself.
I don't see how it is possible for 98steps to wins the challenge, but, I am very curious to see what the method being employed is.
I had coded a method in WinCraps that placed $660 across after the point was established, and then when any one of the numbers hit, take the bets down to the table minimum and play a take, press strategy from then on. This method ended up with a positive skew on the outcomes, whereas most of the sessions ended in the positive. After all, you will be ahead money every roll any 4,5,6,8,9,10 rolls before a 7. Of course PSO costs you $660. The drawback was it required around $8,000 BR to keep from busting out an enormous number of times.
None of this really matters though. My main reason for posting was this thread deserves to be the top thread of all time, so I am posting in hopes of being able to help displace that 1.414 hoax thread!
Quote: guido111Nice observation. I had not thought about it in that sense.
Then would 1 billion people using a system for 1 bet also be the same thing?
My point only was that NO single person could ever see 1 billion rounds in a casino in a lifetime.
I would make that a challenge for those that disagree with me.
To avoid derailing this thread, let's table this until your post-football thread.
Quote: MichaelBluejay
mkl64321, you talk about this alleged burden, but you don't suggest an alternative...even though I asked. I'm giving 98 a way to challenge the results. You think that's burdensome because he's not competent to challenge my code. Okay, so in your universe, how exactly *would* 98 challenge the results?
I must inhabit the same universe as you do; otherwise, we wouldn't be able to have this conversation in the first place.
I didn't suggest an alternative because without radically altering the terms of the contract, there IS no alternative. It's not unlike one person standing up and making a speech in Swahili (however powerfully logical and persuasive), with the correctness of the speech to decide a bet. Of COURSE, I realize the inherent difficulty here, but it's a conflict that is naturally brought about when science confronts belief. This, in turn, is brought about by illogic:
1. 98steps says he has a winning craps system.
2. The posters on this board, the Wizard, 400 years of mathematics, and the bitter experience of millions of gamblers say that no such thing can possibly exist.
3. 98steps then accepts a challenge to prove or disprove the validity of his system--VIA THE SCIENTIFIC METHOD.
So if he didn't believe the science before, why would he believe it after the challenge showed the science to be correct? After all, strictly speaking, the simulation is unnecessary--the math already proves what we all know to be evident.
In any case, a neutral third party arbitrating this whole mess would certainly ask for an independent evaluation of your source code. Since 98steps will almost certainly challenge your results, you would probably save time--and get your money faster--if someone else verified your code. I don't know if the Wiz is conversant enough to do that, but he would be a logical choice.
Is someome present when mrbluejay runs his simulation. Is it possible this system, by some miracle of miracles, produces a profit after the first try, but then it gets run again and loses? Then the 2nd run is what is used as proof that it failed? I would think someone from both sides would have to see the running of the similation???
Quote: 98stepsI have developed a craps strategy that utilizes a combination of 0% wagers and systematic betting progressions. Thru fairly extensive testing it is trending at 95% effectiveness, winning between 500-800 (average win 700) on 19 out of 20 sessions, losing +/-4500 on the one session. I have tested my theories at home as well as on live tables.
If a system is proven to be a consistant winner, is there anyone out there that would consider financing its operation?
Quite an interesting thought to open a thread here.
Yes, I would.
I would need to see data that you have gathered first.
Then I need to know your meaning of the phrase "consistent winner" and for how long it could be a winner.
Then I would need to see your system and test it myself so I would be satisfied with it's risks.
But after all that, why would I need you?
I'd be off quietly grinding out a living and making many shake their heads at me at what I'm doing.
I have lived most of my life in casinos, spent most of my life dealing Craps, and I could honestly say I have seen it all. Us dice dealers know what I am talking about.
I will wait to read it in your book.
But as for the Craps challenge, after reading this very long thread, I do not think it will happen, but if it does I already know what the end result will be, as many others know aslo.
Quote: WizardThe one guy who accepted my betting system challenge didn't argue when I told him he lost. Not for one second. Here is the whole story.
The first and the final paragraphs of the story imply that the behavior of that person was highly unusual, even unique. You state that every other person you've dealt with in a situation like this has tried to weasel out of it.
98steps has passed through the initial stages of the gauntlet--he has tendered the actual system and has tentatively agreed to the contract--but it's a long way from contemplation to consummation, as I learned on high school dates. I would be gratified to see this challenge actually come to fruition, though.
Quote: Zcore13Question...
Is someome present when mrbluejay runs his simulation. Is it possible this system, by some miracle of miracles, produces a profit after the first try, but then it gets run again and loses? Then the 2nd run is what is used as proof that it failed? I would think someone from both sides would have to see the running of the similation???
I would imagine that's the reason for the 1 billion "rounds": to reduce to essentially nil the probability that the simulation will produce an aberrant result.
I think MBJ is willing to lay 1:10 against the possibility of that happening. I certainly would be.
Quote: mkl654321Quote: WizardThe one guy who accepted my betting system challenge didn't argue when I told him he lost. Not for one second. Here is the whole story.
The first and the final paragraphs of the story imply that the behavior of that person was highly unusual, even unique. You state that every other person you've dealt with in a situation like this has tried to weasel out of it.
98steps has passed through the initial stages of the gauntlet--he has tendered the actual system and has tentatively agreed to the contract--but it's a long way from contemplation to consummation, as I learned on high school dates. I would be gratified to see this challenge actually come to fruition, though.
"but it's a long way from contemplation to consummation, as I learned on high school dates."
mlk...you are so good with words and descriptions in your posts.
This old man, me, got a kick out of that.
The best simile or metaphor ever! ( I dont' know which one it is, and too lazy to look it up)
Keep them coming
added: I see you are an English teacher.
I got it now. Never liked my English teachers in school, but back then they could hit you with a stick.
Quote: 7winner
"but it's a long way from contemplation to consummation, as I learned on high school dates."
I liked that too. It was a VERY long way in my case.
Quote: Zcore13Question...
Is someome present when mrbluejay runs his simulation.
Mr. Bluejay would have to start the simulation.
I think when the Wizard ran his, he offered the challenger to be present and press the button so to speak.
Quote: Zcore13Is it possible this system, by some miracle of miracles, produces a profit after the first try, but then it gets run again and loses?
Then the 2nd run is what is used as proof that it failed? I would think someone from both sides would have to see the running of the similation???
Yes it is possible, but only by a total failure of the RNG or Mr. Bluejay's computer catching fire..
The probability of the miracle that you described actually happening is 0.
From Mr. Bluejay's own website:
"I realize that we all want to believe that there's a way to beat the casinos, because knowing a guaranteed way to win would be a lot of fun.
But no such way exists, and wanting something else to be true doesn't make it so. "
could it be said any better?
maybe mlk could!
Quote: MichaelBluejayI'm willing to hear what MathExtremist says about my RNG results.
Quote:Finally, I simulated 100 million dice rolls (10x more than suggested) and here were the results:
1: 1667514
2: 1667513
3: 1667511
4: 1667512
5: 1667515
6: 1667513
First, this only adds up to 10M rolls, not 100.
Second, there are two more tests I'd do prior to kicking off the sim:
1) Game result distribution test. Do a distribution of random(6) + random(6) and test for the appropriate probabilities for each sum [2..12] (1/36, 2/36, etc).
2) Serial correlation test. Do a distribution of (random(6)+random(6), random(6)+random(6)) and test for appropriate probabilities for each sum following each other sum. In plain English, make sure that the chances of rolling a 7 aren't greater after certain numbers than certain others.
I may have a reply from runrev.com by Wednesday or so. I'll post that here.
