Craps Not Set in Stone?

Introduction

“Dice control,” or, “Dice influence,” whatever you want to call it seems to come up here and there periodically. As far as the forum goes, you’ll see a new thread about it every few months; here’s the most recent example.

With that, we will briefly look at what exactly, “Dice Influence,” means and address both sides of the argument as to whether or not it is a viable means of playing Craps at an advantage.

Dice Influence

Dice Influencing is simply, ‘Setting,’ the dice in a certain way, as well as throwing the dice in a certain way, such as to purportedly make more likely desired results, make less likely undesired results (typically, sevens) or to do both. I guess by doing either of the two, even if the probabilities remain all else the same in relation to each other, you are doing the other. For example, if you could roll in a way that makes sevens less likely, then all other results become more likely.

‘Setting,’ the dice is simply orienting them in such a way as to, combined with the toss, shift the probabilities away from whatever they would normally be with a so-called, “Random Roll.”

Of course, “Random Roll,” is something of a misnomer because even a purported dice influencer’s results are going to be random. It’s not as if they can guarantee one result, or the other, which they wouldn’t ever claim to be able to do because their claims could be invalidated in as little as one toss of the dice.

With that, let’s read through the thread and grab a few quotes that I like:

OdiousGambit, who started the thread, kicks it off with:

“I once observed other players were quite convinced a dice setter could influence the dice and they were following him around and passing the dice so as to minimize their own influence. At least two players were making this clear, and I felt more were in on this. I went along with it and benefited, but didn't make all that much as I stuck to betting the minimum... though with max odds and plenty of come bets.

The reason for that was that I had no idea why they were so convinced. Very easily they may have been convinced by something that wouldn't convince me, such as there being one day where the guy had lots of winning sessions. Now, for sure he did set the dice, and it did seem as though the dice were hitting the back wall very gently tossing from next to stick. So admittedly those requirements were met for me. I talked to him at the cashier line and he was definitely bragging, not just saying 'oh I was just lucky'

I would still require more than the two sessions I saw, though they came out well for him. I would be influenced no doubt if I kept betting more and winning more. If he had an occasional losing session, I would consider that normal. I probably could never get to 100% sure, but if a clear majority of sessions were winners after 20 or so, I'd consider it a good theory he could do it, or at least he's damned lucky and who cares how it is happening. I think it would be an unrealistic maybe 200 sessions to be able to say, you know, I think I did run into someone who seemed to be able to do it.”

Immediately, OdiousGambit points out that the sample size in question is not, by itself, conclusive. Two hours of shooting, even if this guy were the ONLY shooter, would still represent only a few hundred rolls, at most, which isn’t enough to determine anything.

BillHasRetired comments:

“At the tables? That's really iffy. The guy might just want to run one good long roll (~25 rolls) and escape notice by the casino. So, maybe follow him around The Strip and watch him reel off a 25-roll session, head to the next casino, do it again, over and over. That would make it 'more probable than not', especially if spread over a full week. Coming up a consistent winner is a requirement

In the lab? I'd want to see at least a 2% edge in the Seven-to-Rolls-Ratio over 1500 rolls before I'd consider DI 'more probable than not'.

Hell, if Wizard could get someone to go to Vegas for that Two Dice thing he had last week, I'm sure someone could wrangle up the Golden Craps boys (Scoblete, Dominator, Sharpshooter, etc) to convince us all. NDAs would be required, I imagine.”

I tend to agree with BillHasRetired that more than a thousand rolls, at a minimum, would be required in settings (a Craps Table that would be casino appropriate) that are isolated, and even then, I don’t know how much that would prove.

Let’s start with the simple question of seven or no seven. As we know, someone would normally be 16.6666667% to roll a seven on any given roll. If we suppose that the purported dice influencer’s angle is just that he will roll fewer than expected sevens, and as a result, the advantage (whatever it might be) is going to come out in the wash, then let’s look at 1,500 rolls:

In the sample of 1,500 rolls, we would normally expect to see 250 sevens, but what we want is any number that is not that, and again, the idea is that the advantage will come out in the wash as long as the bettor sticks to low house edge bets. We expect to see 1,250 rolls that are not sevens anyway, so we can use this binomial probability calculator to see how likely someone would be to just be killing it, at random.

In this sample, someone would be 0.00021 likely to roll 1,300, or more, non-sevens. This would occur randomly about 1 in 4,762 sets of 1,500 attempts. As we can see, that wouldn’t really be enough to prove anything in that sample, but it would be sufficient to give a good indication that there is something worth proving over a larger sample.

With that, this person would have rolled only 80% (or less) of the expected number of sevens that we would normally see. I would be shocked if any dice influencer has ever even claimed to be able to reduce the number of sevens by a fifth, so that being the case, we would generally need to see a much larger sample size to demonstrate the potential veracity of whatever it is that they are claiming to be able to do.

More conservatively, we could suggest that the number of sevens is reduced by only 10%, so that would go from 250 per 1500 rolls to 225 per 1500 rolls. In this event, someone would be 0.04338, more than 4%, to have 225, or fewer, sevens completely at random. That would mean that this would just randomly occur 1 in 23.05 times and would, therefore, prove nothing.

I think that any direct claims that have been made have been even more conservative than reducing the number of sevens by 10%, so that being the case, you would need a much larger sample size than even the 1,500 to ever come close to proving that claim.

As far as monetary results are concerned, I would suggest that actual monetary results would be pretty meaningless. That’s true even if the person makes the same spread of bets, in terms of amounts bet, every single time. When you throw in the fact that they could sometimes bet more, then you add more variance when it comes to actual cash results.

AlanMendelson states:

BillHasRetired even if a DI surpassed all of the requirements stated by the members of this forum they would never acknowledge Dice Influencing.

They'd say "variance" was responsible for the great results.

Stop wasting your time here.

In my opinion, this is an unfair characterization of what my mathematical opinion would be.

What I mean is that I would not say variance WAS responsible for the great results because I could not be in a position to know that. At a minimum, the person said that they were going to exert enough influence over the dice to achieve a particular outcome and then they did it.

Depending on the sample size in question, yes, of course I am going to say that the results could have been due to variance, but it’s unfair to suggest that I would postulate that as some sort of unassailable fact.

With that, what would or wouldn’t be satisfactory to demonstrate the veracity of the dice influencer’s claims is going to depend largely upon who you are asking. If someone is on the side of wanting to believe in the viability of dice setting, then that person is going to generally have a lower standard of proof to demonstrate that a different person can do it. If someone does not, for whatever reason, want dice influencing to even be potentially viable, then yes, that person might demand a standard of proof such that the results of a particular sample otherwise be statistically near-impossible.

For example, to roll the dice 15000 times and only come up with 2000 sevens is all but impossible. If someone would demand a dice influencer perform such to demonstrate that they can actually do it, then that sort of requirement for proof would be totally unreasonable. Again, I don’t know that anyone has ever made the claim that they can reduce the number of sevens they throw by 20%, so it would be a ridiculous standard to hold them to.

How much of a sample size would someone need who is neutral on the subject, or at least willing to pursue the question on a neutral basis? Quite frankly, that’s going to largely depend on what the purported dice influencer is claiming to be able to do.

If the dice influencer says, “I can roll fewer sevens than expected,” then they are not making a claim that can be given reasonable test parameters. If they were to roll the dice 150,000 times and only roll 24,999 sevens, then they have technically succeeded vis-a-vis their claim, after all.

With that, you would want the person in question to make an actual specific claim. The claim having been made, you could then determine an appropriate sample size for the demonstration of that claim.

It’s also important to note that the probabilities work both ways. For example, if someone claims that they can reduce the frequency of sevens to, say, 15% of all rolls, then the claim should be graded (if they don’t succeed) based on the probabilities of 85% anything else and 15% sevens.

For example, with 5,000 rolls, this person would be claiming that 4,250 of them should not be sevens. If they end up rolling only 4,225 non-sevens, then the probability of rolling that number, or more, would be 84.373% based on their claim, but only 1.35% normally.

In other words, this would have less than a 1.5% chance of happening for someone just throwing the dice, but if the person is claiming to normally only throw 15% sevens, then these results would be pretty poor, by their standards. In both cases, the results would be insufficient to prove anything, but would be such as to warrant further testing.

In any event, there’s really not much of a point if a person is not making a quantifiable claim. Qualified claims are meaningless; how would a person who simply believes they can throw fewer than the expected number of sevens (without a specific quantifiable) even know what their edge over the casino was, if any? If you’re going to claim to be a dice influencer, then at an absolute minimum, you should have some idea of how successful you have been at doing whatever it is you think you are doing to change what would otherwise be the natural probabilities.

Gordonm888 makes this comment:

If people can do DI, they would not want you or anyone else to know that they are capable of doing that. And certainly they would have absolutely no desire to organize and endure the extraordinary effort of videotaping 1000 throws with some authoritative person nearby to certify their throws.

I get impatient when people require extraordinary proofs requiring other people to act against their own interests and to stage monumental demonstrations. The world doesn't naturally organize itself to provide convincing video proof of 'illegal' activity so you can sit on your couch and see the proofs without making any personal effort at all.

I would say that I both agree and disagree with this take. I certainly agree that some or most dice influencers, if any exist, wouldn’t want to make a big production of how successful they are at it. At least, not without a substantial wager being made pursuant to the results of whatever the terms of the experiment are.

That said, there are any number of people who have claimed to be dice influencers who have been very open about their identities. Often, there has been a profit to be made as a result, which often is not made inside of an actual casino, but they still don’t seem shy about being known.

In any event, I think Gordon’s claim has merit, but I disagree that I would take it as any sort of absolute. At a minimum, there HAVE already been people who legitimately believed that they could do it that have attempted to prove the same. I suppose some of them might not have legitimately believed they could do it, but I’m not going to toss accusations around.

With that, let’s get into some other arguments for both sides:

ARGUMENTS FOR AND AGAINST

I would say that I am pretty Agnostic on the subject of dice influencing. I definitely lean skeptic, so I am going to do this on kind of a believer-argument, skeptic-counterargument basis.

FOR: Dice influencing is largely a physical skill that can be learned. Essentially, all it requires is knowing how to set the dice and throwing them correctly. It is for that reason that dice influencing is possible; you simply have to become great at throwing the dice in a predictable-enough way to shift the long-term probabilities compared to a, ‘Normal,’ throw.

AGAINST: Free throw shooting is a physical skill that can be learned. However, with free throw shooting, we have examples of people (basketball players) who are demonstrably better at it than other people. Another key differentiator for free throw shooting is that anyone with the strength (which most people have) to get the ball there will quickly become better at it with practice. The results of practicing free throw shooting are observable as your percentage of success will quickly go up and then either level off or rise very slowly.

Jordan Poole was the NBA’s free throw shooting leader for 2021-2022 hitting a ridiculous 92.48% on 246/266 free throws.

Imagine that I went outside and started shooting free throws, the best I can remember, I haven’t even touched a basketball in several years, so my percentage will probably be abysmal.

Let’s say that I make forty of my free throws. That’s 40%. Let’s say that is my long term expectation.

Obviously, I would not beat Jordan Poole in a meaningful free throw shooting contest, but if we had one by which you are out as soon as you miss and the other person makes, then I would have a .0752 * .4 = .03008 or 3.008% chance of just beating him immediately.

The larger the sample size of the contest; the less likely that I am to beat Jordan Poole, which is why sample size is such a relevant factor.

Anyway, I’m not going to get into the pyramids and stuff like that. The differences between myself and Jordan Poole are stark when it comes to shooting free throws, but obviously, a dice influencer would not require such a huge swing in percentages to be profitable at Craps. However, because the usual assumed shift is a small shift in percentages on an event that is not terribly likely to happen on an individual trial (a seven) anyway, you’re going to need a large sample size to ever prove anything.

So, if DI is possible, then it is because it’s a physical skill. That said, those asking for proof of concept have, one and all, been left wanting.

FOR: I saw (insert person) and they dominated this one session. I saw (insert person) and they won every table they played for three days.

AGAINST: Yeah, and so do people who don’t claim to be dice influencers for whom, we would assume, roll sevens about 1/6th of the time.

Also, were you tracking the number of sevens these people rolled in these sessions? Even with Place Bets on six and eight, you could roll MORE sevens than expected and still be profitable, you just have to also roll more sixes and eights than expected.

With that, profits over one session or a few sessions are meaningless and would constitute proof only to those who want to believe anyway. If someone is neutral on the subject, then they will know that an isolated session, or few sessions, are irrelevant. Anyone can have a few winning sessions.

FOR: The only reason you don’t believe is because you don’t want to believe! Nothing would satisfy you for proof!

AGAINST: This is just a terrible argument; it’s not even really an argument. Someone could state that I don’t believe in Santa Clause simply because I don’t want to; I’d certainly have no way to prove wrong their saying that’s why I don’t believe in Santa Clause, but I more likely don’t believe in him because he doesn’t exist.

Personally, I don’t want to believe and I have no motivation to disbelieve.

I’d even expand upon that and say that I can see no reason for anyone to specifically want to disbelieve. The only parties who would really have an interest in it not being possible, I would think, are the casinos. Why would I be opposed to someone being able to play Craps profitably? It would make no sense for me to be opposed to that.

Requiring some sort of statistically meaningful proof is not being overly demanding considering that any claims as to dice influence are, fundamentally, mathematical in nature. The entire goal of dice influencing is to change the probabilities of, ‘Random’ dice in some way as to yield an advantage for the player, so there can be no dispute that it is a mathematical claim in nature.

It would take a lot to absolutely PROVE it; I grant that much. That being said, I would be satisfied with a preponderance of the evidence (more likely than not) sort of standard, so those demands would be lesser demands. Again, merely beating a 50/50 chance, whatever the metric being used is, would not satisfy the standard that it is more likely than not, in my eyes, simply because there remains some level of being able to just randomly do that over a particular trial even whilst exerting no influence over the dice at all.

What I would probably want to see in a controlled environment, and over a meaningful sample size (based on whatever the person is claiming to be able to do) is them roll in a way that is, at an absolute minimum, less than 5% probable to just happen randomly. That’s going to be my minimum standard which would justify another test which results could be combined with the first test.

FOR: This guy’s lifetime results are $XXXX.xx over x years!!!!

AGAINST: That’s what the person who wants you to believe it is saying his results are.

OTHER THOUGHTS

One thing that I have always wondered about dice influencing is that most discussions are centered around reducing the frequency of sevens as well as making sixes and eights more likely, whether directly or indirectly. Most DI advocacy that I have seen, notably that of Frank Scoblete, seems to have placed some focus on the Place Bets of 6 and 8.

What that makes me wonder is: Why not just focus on rolling one thing and sevens being otherwise immaterial?

Or, alternatively: Why not just focus on rolling more sevens?

Hypothetically, what if someone could roll 20% more 12’s than would otherwise be expected? If that happened, 12’s would go from 2.777% to 3.3333%, so you would have this for a bet on Midnight:

(30 * .03333333) - (.96666666) = 0.03333324

In other words, being able to roll 12’s with that frequency would create a 3.333324% advantage on betting Midnight. Granted, you would still have to fade one of the line bets, but if you were on the Pass Line, you just wouldn’t throw your 12 throw on the Come Out. If you were on the Don’t Pass, then this (all else equal) would increase the number of pushes and slightly reduce the House Edge per bet made, even if you want to bet (and throw for) a 12 even on the Come Out roll.

We can even solve for x to get the percentage that you would want to hit to be dead even on betting Midnight:

(30 * x) - (1-x) = 0

X = 1/31

In other words, if you could increase the percentage of 12’s that you throw to 0.03225806451 or 3.225806451%, then you would be breaking even. This reflects an increase of roughly 16.129%, so you don’t even have to hit 20% more 12’s to get yourself to breakeven. Anything more than that and you would be long-term profitable on this bet.

Another thing that I have wondered is that you don’t see much talk about intentionally trying to throw sevens. Logically, seven (1/6) is the most likely number to throw anyway, and thus, should be considered the easiest number to throw. This would create a big advantage on the Don’t Pass if you could throw an increased number of sevens to a meaningful degree because your first throw would be just chucking them, and then when a point is established, you would throw for sevens.

The argument that I would understand about 12’s is that some might say it’s just too large a House Edge to buck and there is simply no way to be good enough to generate the significantly more 12’s that would be needed to conquer it.

As far as throwing more sevens goes, the player would be trying to conquer the DP+Odds, which have the combined lowest House Edge possible in the game of Craps, so I would think that throwing more sevens might be the best way to go. Having established a point number, the DP bet has survived the only point that it is at a disadvantage anyway, so now a player who could throw more sevens than otherwise expected would just be adding to the advantage (at that stage of the bet) that he already would have anyway.

That’s just my take on what I would probably try to do if I was absolutely determined to try to become a dice influencer, which is something that I am presently more on the side of skeptic (by a lot) than I am on the side of believer.

PROOF OF CONCEPT

Another thing that I would like to see as proof of concept would be either a mechanical dice thrower (built to throw an otherwise legal Craps throw) on an actual Craps Table. The reason that some sort of computer simulated one wouldn’t work, in my view, is because a simulation might be able to throw with precision that wouldn’t even be possible for a mechanical apparatus, so by extension, not possible for a human.

In my opinion, this would be the easiest way to prove the viability of dice influencing as a concept. Some might argue that the arm would never get tired and lose concentration, (as happens with any tired person on any physical endeavor) but with the amounts that you can bet on Craps in some casinos, it’s not like a person would have to throw the dice 1,000 times per day just to show a long term profit. For that reason, I don’t think the fact that it’s a mechanical arm would invalidate the proof, especially since it would just be proof of concept anyway.

In other words, the mechanical arm being able to do it does not automatically mean that a person could influence the dice to a profitable extent; it would simply mean that there is more worth exploring along those lines.

CONCLUSION

In conclusion, this comes up, off and on, in gambling circles, has for decades and will probably continue, to some extent, as long as Craps continues to exist.

For my part, I’m open to the possibility of it, provided someone can actually demonstrate that they can do it over a meaningful sample. I’m not so closed-minded that I would refuse to accept any form of proof or would set an unreasonable standard for it.

Finally, contrary to the apparent belief of some people, I’m not highly skeptical simply because I don’t want to believe it. I have no percentage in whether or not it can be done. If it could be done, I already know that I probably can’t do it because I don’t think my hands are steady enough for my throw to be as consistent as it would need to be. In the meantime, I also do not own a casino, much less a casino that offers live craps, so someone being able to influence the dice to a meaningful enough degree to create an advantage has no negative impact on me.