Automated bots, queing theory, grid of bots and pyramid scheme.

[mr.ore]

Imagine you are in a supermarket and there are several queues of customers waiting to pay for their goods. It takes random time before any customer is done, but it usually takes short time. If there are many queues, it is good to go to the shortest one. What if some customer takes too long to be done? People go to another queues. There exists algorithms to distribute customers among the queues.

I have tested, but not enough, an idea of distributing losses among several systems which are played at the same time at one table, with limits included.

The idea I had was that there are several barrels which can hold water, and there is a hole at the bottom which allows the water to go out and empty the barrel. Now there were one pipe which were filling the first barrel with different amounts of water, and it could happen that the barrel would overflow. To avoid overflowing, the barrels are connected with pipes with different diameter. If the water were poured too fast to the first barrel, it would be brought by those pipes to another barrels, to be taken care of. The problem is that when water appears in some barrel, another tap is being activated, pouring randomly water into it.

Well, the idea is old. If you play several systems at once, the bankroll is more stable. But with progressions, if one of them is losing, it is exponential, but other system(s) is gaining only at constant speed, so one system's failure means overall failure. If you have several more systems which are at the time doing well, why not just decrease bet from the one which is losing, and distribute it to another ones? I made an interface to one class from which my other classes with roulette systems are derived, which served this purpose, and created flow matrix to distribute the amount unevenly. It did not worked, it even magnified drawdowns. The idea is, what if there existed mathematically correct way how to distribute play among several players, thus increasing probability of winning. There would be two groups of players - those playing at the same table, and those playing independent games. It seems that the second group does not matter, because independent games can be played by one player one after another, but bankroll is more balanced if you average those of players. When there are more players at the same table, the impact of table limits is lessened, and the probability of winning could be increased. So I might be thinking of such an algorithm.

It would be interesting to create software, where several computers with bots would be connected to a grid, and balancing their systems, or backing up one another. They would be connected into a network, and the only problem would be how to distribute won or lost money among the players. It would have be based on volunteers, and the application would have be to be open sourced so that anyone could check the algorithms to distribute money. The volunteers would have to send few dollars to each other from time to time(on PayPal, Monebrookers, PayPay and others), according to some algorithm, which would also try to minimize number of transactions. On casinos with low minimal bet like 0.01$ it would not be such a problem, because spending some 50$ = 5000 units would not be so risky, and definitely worthy for the fun of watching the "show". I would not expect this to win long term, but it would be definitely fun. How long would be able to survive a grid of players against a game with negative expectation?

Another idea would be to not split money evenly, but the algorithm would prefer those players, who would bring another players into the grid, or who risked more money. This would give those first players positive expectation on this game, since they would be shifting some of their risk to others. As long as the number of the players would be increasing, thus creating a n-ary tree, the thing would be profitable for those in higher levels. The buy-in should be low, to have enough players to volunteer. The system would automatically ban those who would not obey the steps the algorithm would tell them to do(transfer money to another player). Well, this last idea would be problematic, and it could be even considered pyramid masked as a collective gambling. But this idea appeared in my head, so I had to write it there. I know it is a BAD idea, but mathematically speaking it is beautiful idea. While profit from pyramid would be almost for sure(for those who leave while in profit, in the end the whole system is predetermined to be a GREAT failure), the fact that that there would be the roulette part would allow for a great experiment, which I would be interested to see, even if it cost that fifty bucks

[mr.ore]

Food for thought: link:://:.damas.ift.ulaval.ca/_seminar/filesH06/dolgov04resourceMDP.pdf

[VLS]

Ore, congrats on putting the neurones to work.

Sounds an interesting one here for doing "loss management & distribution" 

[ThomasGrant]

Good post.
Made me think.

What if...

What if all the members in here joined their pc's to play at one casino.
With a variety of systems. And number crunching.
Would we be better of?

[mr.ore]

It would have to be decided by simulation, if it is worthy. Even if it were possible, the problem are the casinos. There were groups of players in some countries who teamed without some algorithms, just played martingale on the same table, and because of that, players from those countries are banned on several casinos. The question is, if it would be possible to balance risk among several players playing independent games with progression. The aim of this would be to decrease variance and personal risk of any individual participating. Or one could also play alone on several casinos, with several bots.

The idea with "Ponzi scheme" is also interesting, because when gambling the money, there might be say some 70% chance to win 25% of overall shared bankroll, and that profit distributed to list players who would in normal Ponzi scheme would have negative profit, but there would be possibility of breaking even or with small profit even for them. With 30% probability everyone would lost. The problem is that at a certain point it would be almost certain, that it is going to lose, and rational player would take the profits for themselves and leave. There would have to be a lot of balancing and transactions to avoid that possibility, and because each transfer have some cost, it would add another game with negative expectation as a subsystem. If Jack sends money to John via PayPal, and John sends what he received back to Jack, and Jack sends it again to John and so on, then if PayPal takes 1% from each transaction, in the end they would have nothing. It might be also considered suspicious transaction by the system...

[mr.ore]

The idea of distributing money is no different from distributed p2p networks(like Freenet, Share,Winny,PerfectDark...), which is used to anonymous p2p filesharing, and they become more popular because of France's HADOPI anti-piracy law. The idea with these networks is to distribute data to it's nodes, and there are often things like who has big upload, can have faster download, to have up/down ration ratio ~ 1.

The same principle as anonymous p2p networking could be used to balance money, and even for anonymous payments as long as all participants would not cheat. A send money to B, be splits money to three parts and sends them to B,C,D, B and C send money to F who is the real recipient, and D sends money to E, and E finally to F. The route is random, and the recipient is decided by packet with encrypted data which is being distributed during transaction.

[F_LAT_INO]

Mr.Ore,
For several years am working on this idea,playing all 3 EC X 3=9 BETS EACH SPIN.

Testing it on thousands of past spins,it is constant winner....BUT...can't develop
the way to implement it in the real play...it is just impossible alone.
One EC/even 2/can handle,but 3,nop.....not enough time to calculate and place bets
before ---no more bets pls.----
btw--Am talking about en prison rules on which  I play.

[ThomasGrant]

Hmmm...

This could be a very interesting experiment.
What if a group of members did login into the same casino.

So.

Members.

1. Have the same casino.
For example.
All interested members must have a EuroGrand account.

2. All members must have funds to play with at the casino.

3. All members agree to meet in
3a. A live game.
3b. A rng group table.

Now I think the rng group table may or may not work.
I know that the Live table will work.

So if we all agree on a time and date.
It should be interesting.

Perhaps a group skype session.
Or even use the forum chat.

Anyone interested in this experiment.
Please let me know.

And I may see if I can organise it.

[mr.ore]

F_LAT_INFO: If you can play on with en prison rule, it's obvious you should stick to EC bets. How do you combine them? Are you somehow splitting a progression, or are you flat betting some bet selection based on statistics and low probability of a negative trend happening on all chances at the same time?

[mr.ore]

ThomasGrant:

Before a play for a real money, it would be better to make a simulation, and use a) RNG spins, b) real spins. This is not simple task. How can we trust each other that if someone's account is wiped out, yet overall profit of the group would be positive, that the other members would give part of their profits to those unlucky? The only way is that everyone bets the same, and that is suspicious and might be considered team play, and then casino might refuse to pay your win. You might even get to some blacklist of online casinos.

The answer whether it is really good idea to play like this have not been resolved yet. There would be a need to write an application which would tell everyone where to bet, and the target would have been to be established before the play. What is needed is a mathematical model of the whole situation.

And why only roulette, what if the algorithms allowed for splitting a risk to several games, like blackjack, or even forex, dll library can be easily created for MetaTrader, and with some networking code it could also participate

[mr.ore]

Excerpt from Eurogrand's terms and conditions:

COLLUSION is a problem

EuroGrand casino
...

7.4 We may at any time set off any positive balance on Your Account against any amount owed by You to any company within the same group of companies as the Operator, including (without limitation) where we re-settle any bets or wagers pursuant to paragraph 4.8 (Duplicate Accounts), paragraph 11 (Collusion, Cheating, Fraud and Criminal Activity) or paragraph 18 (Errors or Omissions).


11.4 If:
11.4.1 We have reasonable grounds to believe that:

11.4.1.1 You have participated in or have been connected with any form of collusion, cheating, unfair or fraudulent practice, or otherwise any other criminal activity; or

11.4.1.2 You have, in relation to any bet placed or game played, gained an unfair advantage over us or any other person participating in the relevant game; or

11.4.1.3 a game in which You participated included the use of collusive or fraudulent practice, or any cheating; or

11.4.2 we become aware that You have placed bets and/or played online games with any other online provider of gambling services and are suspected (as a result of such play) of collusion, cheating or fraud (including in relation to charge-backs), or otherwise any criminal or otherwise improper activity; or

...
then we shall have the right to suspend Your Account for an indefinite period of time; and/or withhold the whole or part of balance of Your Account; and/or close Your Account and terminate the Terms of Use; and/or recover from Your Account the amount of any pay-outs, bonuses or winnings which have been affected by the event(s) contemplated in paragraphs 11.4.1 to 11.4.4 (inclusive) above.


11.5 For the purposes of paragraph 11.4.1:

11.5.1 the basis of our belief shall include the use by us (and by our gaming partners and our other suppliers) of any fraud, cheating and collusion detection practices which are used in the gambling and gaming industry at the relevant time;

...

11.5.4 an ââ,¬Å"unfair advantageââ,¬Â shall include: the exploitation of a fault, loophole or error in our software (including a game); the use of automated players (sometimes known as ââ,¬Ëœbotsââ,¬â,,¢); or the exploitation by You of an ââ,¬ËœErrorââ,¬â,,¢ as defined in paragraph 18.1.

Other casino's have similar terms...

We would have to create system with algorithms where collusion would not be simply detected. I also don't like the the spread there, minimal bet is 0.5 Euro.

[ThomasGrant]

I only used EuroGrand as an example.
Never mind.

I like the way you think.

[mr.ore]

Back to the theory. Imagine we have two players who would decide to cooperate in playing the following system:

[tex]\begin{tabular*}{1.4\textwidth}
\\\hline
bankroll&bet&after W&after L&Pwin &avg.return\\ \hline\\\hline
0&no bet&-&-&0.0000%&0.0000\\
1&1 unit(s) on 11:1&12&0&5.9832\%&0.9146\\
2&1 unit(s) on 11:1&13&1&11.9912\%&1.8372\\
3&1 unit(s) on 11:1&14&2&18.0359\%&2.7715\\
4&1 unit(s) on 11:1&15&3&24.1213\%&3.7188\\
5&1 unit(s) on 11:1&16&4&30.2736\%&4.6876\\
6&5 unit(s) on 2:1&16&1&36.4751\%&5.6720\\
7&1 unit(s) on 8:1&15&6&42.5956\%&6.6306\\
8&1 unit(s) on 8:1&16&7&48.8015\%&7.6165\\
9&1 unit(s) on 5:1&14&8&54.9216\%&8.5749\\
10&3 unit(s) on 2:1&16&7&61.2132\%&9.5882\\
11&1 unit(s) on 5:1&16&10&67.5030\%&10.6010\\
12&2 unit(s) on 2:1&16&10&73.7927\%&11.6137\\
13&3 unit(s) on 1:1&16&10&80.0825\%&12.6264\\
14&1 unit(s) on 2:1&16&13&86.5422\%&13.6935\\
15&1 unit(s) on 1:1&16&14&93.0892\%&14.7886\\
16&no bet&-&-&100.0000\%&16.0000\\
\end{tabular*}[/tex]

In the beginning each would have 4 units, so overall shared bankroll is 8 units. How should they play? One of them must bet on a square 8:1 and if he wins, he has to split a win with the second player, so he gets 4 units. What if he lose? Then he has 3 units and the other player has 4 units. Overall bankroll is 7, so we are on line 7, and the second player with 4 units bets 1 unit on square 8:1. If he wins, then bankroll is 15 units, and the second player has 12 units, and we need to make the last bet. Who should take the risk? If the player with higher bankroll continues and wins one unit, he has to give 5 units to the first player, so that they both have profit of 4 units. Now is a question who should play if the target was higher & more players in the game, so as their bankrolls are balanced so that in case of betrayal the negative effect on a group is minimized in it's impact. In case the table limit were 3 units, they would have to cooperate on line six, and if not enough players had units, transaction between their accounts would be needed. The aim is as maximal balance as possible to both avoid risk of betrayal and a need of transactions.

[Mikeo]

This is very interesting, but I don't see how this is any better than one person using a bot which concurrently plays a number of systems, using differential betting (which may be better than a number of people playing individual systems and placing conflicting bets), and possibly balancing risk between the systems. And you wouldn't have the concern about distributing winnings/losses among the players. And there would be no worry about player collusion being detected.

The only possible advantage would be to divide a bet to avoid the table limit. But then players are making the same bet, so the collusion might be detected.

Am I missing something?

[mr.ore]

The problem is that to survive some losing streaks, you need a big bankroll, but not everyone is willing to risk a really huge amount of money. If there were a way to coordinate many players and play at once, they could have better profit. It's very hard to more than double your bankroll, that says math about roulette. To win a million, you need a million. You need a huge bankroll. Imagine a huge network of a few hundreds of people, where a few chosen would have to do same bet. And each spin you randomly select another individuals. It would be casino's owner nightmare for some time, before they all lose, because of negative expectation. Now if you wanted such a huge group, you have to setup a set of rules, which would minimize impact of thieves, who would enter network hoping they will get some free money. I don't know if it would be possible, but imagine there would be p2p application that would be coordinating such a game.

The rules would be something like:

-low buy in so that many people are willing to connect

-winnings split evenly or in some relation to the time the player is in system and how much he helped

-the wealth would be distributed that no player at a time have more than 20% of average

-each player would be able to set how much he trusts someone, there would be an oriented 
  graph of trust, and algorithms would plan transfer among players using utilizing such information

-in case of betrayal BAN for life

-possibility to leave anytime, if the player does not owe to anyone, he does not have to play, can take his profit and leave, later return

This is just an idea, experimental thinking, brainstorming. From dividing loss among several systems to find a way, how to allow anonymous players to cooperate without a high risk. It might be risk, of course. How can you know that if you anonymously play with someone else who you don't know, that he will split his profits in case you should get something? What if is he just a scammer that is hoping that in case someone else wins, he would be given part of the profit, but if it is he who wins, he would not split it with others? Coordination is problematic, look at my example with two players - if the first player wins his first bet, group's target is reached, and the second player have not played at all. To whom should the money go? The first player might thing that since he was only one to bet, the win is only his own, and betray the second player. The fact is, that in case they would be serious companions, the second player was prepared to risk his whole bankroll if needed. It might seem weird, but most of time majority of participants in such a system would not bet at all, if they trusted each other and everyone was fair. Then there would be no need to have permanently balanced accounts, and swapping would occur only in case of very rare event.

The theory behind it is similar to p2p sharing networks, and it just caught my interest, that's all. Brainstorming, brainstorming, brainstorming...