TwoCat tests the bot on casino account #1

[GogoCro]

On Betvoyager did not change limit on normal roulette.
You can play on these table like before but you will lose 2 cent on every 0, that is not much.

[TwoCatSam]

What is "normal roulette"?

[GogoCro]

They changed limit only on zero house edge roulette, in other roulette games limit is not changed.

[mr.ore]

In a no-zero roulette there is no need of any spread, it has been proved in this paper:


nolinks:nolinks. maa. org/joma/Volume8/Siegrist/RedBlack. pdf


Anyone serious about a roulette should read it, there is also mathematically proven optimal strategy for negative expectation EC games.      Well, that strategy is bet all you have and double until reach table limit or is satisfied with win, or lose, so it is unpractical for real play, but it is proven, that your chance to win some amount is higher than that of any other strategy, like flat-betting.      In 0% house edge game, it does not matter what is your size of a bet, and if you are playing positive expectancy game, you should play with minimal amount you can, to avoid fluctuactions as much as possible (or be Kelly bettor?).     

In case of no-zero roulette, even if you could bet only one unit, it would not affect anything at all, you would just have to play a lot longer do gain profit you need.      On the other hand, it would be inpractical for bot to play ten thousands of spins to gain a target profit, and since players bankroll is limited, while casion's not, better chances of winning would  be on the side of casino.      So yes, it is good to use some kind of progressive play in no-zero roulette, probably something milder, like a variation of six point divisor (some of the graphs in this thread looks very like that), which I think is quite good, and because of zero expectancy of game, you can allow to make divisor big and little decrement it for much longer time, without fear that you will go too down.     

Well, I think that the casino wont decrease spread much more in this game, because there is a lot of stupid players, who will lose more in this game because they can bet more, and their bankroll is limited.      I have looked to the terms of that voyager casino, and they take 10% of your winnings in no house edge when you get your winnings.      Well, still better than lose 2.   7% from each bet you make.     

I wonder how would you calculate REAL house edge game when you think of this 10% rule and of the fact your bankroll is limited.      Or rather: what's the probability you would loss all, depending on your bankroll, and how much you could expect to win.      Than, put it into graph.      I think, that it would look like a gaussian distribution, with center in 0, and each result on the positive side multiplied by 0. 9, while negative side would be normal.      So in the end, the integral(=sum of values over discrete interval in this case) from -initial_bankroll to -1 would be greater(losing),  than integral from zero to infinity.      Because the integral of gaussian distribution is always one, and the left losing side is bigger, the right winning side is lower, and so this game have also negative expectation value, I suppose very low, like 0. 1% house edge or so.     

[TwoCatSam]

"On the other hand, it would be inpractical for bot to play ten thousands of spins to gain a target profit,"

The first thing that grabs my attention is the misspelling of "impractical".  People who write papers on any subject should at least be smart enough to use a spell check, no?  This is sloppy and unacceptable.

But to the heart of the statement:  It simply isn't true.  First, to attain my 5 Euro goal, I usually play from 800 to 3,000 spins.  Somewhere in that range I will either lose 10 Euro or win 5 Euro.  In a 10,000 spin trot I would either win about 15 to 50 Euro or lose a ton!  I don't do that as I set the loss limit to 10 Euro.

I am doing a study to determine if there is a way to reduce volatility with the robot.  I am playing a certain system that will lose if left to its own method of play.  However, over many months and over a million spins I noticed a trend.  This trend has held up no matter which computer I use or which account I play on.  It has held on virtual betting, which means the casino has no idea what the bot is betting and therefore, cannot place a losing bet intentionally.

If--let me emphasize that-----IF this trend continues it will be the key to the vault at any casino which uses a fair RNG.  Zero or no zero.  And, with a large enough bankroll and a proper spread, it will work at any casino.

Remember the word IF...........It's the biggest word in the dictionary.

TwoCatSam

[mr.ore]

"The first thing that grabs my attention is the misspelling of "impractical".      People who write papers on any subject should at least be smart enough to use a spell check, no?  This is sloppy and unacceptable.    "

I'm sorry for that, my English is not perfect enough.     I have been using INpractical for years, because IN usually negates the word following it.     My stupid mistake.     I never use spellchecking, since my english is not perfect and never will be anyway, so there is no point in solving small mistakes, when big ones are present.   I have also resigned to using a/the and use them just randomly, they are useless anyway.     And I'm too lazy to temporarily change spellchecking in Konqueror(nolinks browser in UNIX KDE desktop) to English.     Well, on this message I will try it.     And it's ok, only INpractical was bad   What I don't understand, why after a modify of the message spaces after dot .   are prolonged. 

[TwoCatSam]

Sorry mr. ore

Didn't mean to criticize you personally.  I thought when you said "paper" you meant it was published in some journal or by some mathematician in some university study or some such.

Again, sorry.

Sam

[mr.ore]

Well, I have made several experiments: one million spins flat betting one unit on red.On the first image there you can see, that there is a very long positive trend for tens thousands of spins.Then follows a very long negative trend.In another experiment the graph was choppy and not wining or losing much.On the third it ended losing -1800 units. The fourth won 1000 units.  The fifth lost again, and finally the sixth won it again.  As generator I used ranlux389 from GNU GSL library, probably one of the best(and slowest) freely available RNGs, much better than the generator in standard C/C++ (rand() function), which may be implemented poorly in some C/C++ implementations, where is used congruent generator, which is incredibly fast, and quite good for normal use in most applications, but not as near to perfect randomness as should be this.After each experiment, my application saves the state of RNG to file, so these six results can be considered to be one experiment on six millon spins, if you beleive that RNGs stream should not be broken to be able to make some conclusions.

I personally think this:

the same rules that apply in small scale, applyies also in big scale.So when it is possible to have 37/18*20 = 41 long streak on EC, which should be 6 sigma, and actually 35 have been already seen on some casino, then it is possible for negative trend to continue for 41*1000000 spins, same for positive trend of course. If you have found one, it may continue for a thousand of spins, for ten thousands, even millions, but it can end anytime. I wish you your positive trend lasts as long as possible , but be aware, that you may be just a lucky person.  

[mr.ore]

Well, I tested little tweaked six point divisor and got interesting results on no-zero roulette, but nothing to relay on and be too excited from that.  Differences to normal divisor are:

1) After winning 10 units or losing 20 units we reset the system.

2) Each spin we increase the divisor by 0. 25.

It wery lightly pushes the bakroll up, but big swings are of course also possible.  It is also possible, I just found a good trend.  Well, so far no losing, but definitely some big swings down, so the possibility of failure is still there.  But there are many ways to improve, like waiting for local negative trend, and then bet more aggresively, hardly ever occured streak of sessions loses, and even so it ended up positive.  By the way, system totally failed without money management, so it is really important to reset after that 20 unit loss.  I did not tested, which is best target/stop loss for session, there may be beter one.

This is not a HG for no-zero roulette, of course.  If it was possible for a swing down, it is sure there will once come really long streak of loses for sessions, and failure will be the result.  The win +3000 units/1e6 spins is nothing nice also, but better than flat betting, and more stable so far.

[mr.ore]

-