Beating roulette with math..

[Fripper]

Nice discussion, keep it going.

@superman


What exactly are you betting? Which numbers do you choose?

[superman]

What exactly are you betting?

Inside numbers only.

Which numbers do you choose?

that's where it gets difficult, the bot monitors the last 40 numbers out, counts singles, doubles, triples, quads etc then calculates the remainder to build a list of numbers to bet on, I have had this method working for about 3 weeks now and decided to revisit it again with a simple progression, +1 if behind, until ahead then reset. Remember Dyksexlic (forum member) he didn't give us any real clues BUT after a few PMs' between Victor and myself I botted what we thought was his method, it got too risky and sometimes never caught up.

I have just about given up on EC Dozens Columns streets and all the other areas of the table, I play RNG only as its quick and I want a bot that can be left runniing all on its own, I know others will say it's impossible but I disagree, if you can cover enough to gain an edge then the only problem you have is progression, as you can see from the reveal file results, the most losses is around 4 to 6 in a row, you would think a straight forward progression to cover the losses and be ahead on the first win would work, it does, BUT the bets get too high and depending on your unit size you can reach the table limits. The reveal file is played with 10cent chips, yes that's over $1000 profit overnight, continual play.

It's still a work in progress, but I noticed this discussion and decided to add my bit.

Ignore colours hi/low etc go for the throat of random, it's the numbers that are the starting point for any result, the felt is just there to confuse/distract us, what does and RNG give? numbers, I've been concentrating on numbers now for over a month.

There isn't a hard rule as to how far back you gather your numbers, I have just plugged in 3 or 4 different methods and the reveal shows the results I am getting. It does look promising though.

[Chrisbis]

Fantastic work Neil. 

Great to read some progress on that Inside idea U had months ago.

Nice to see a new twist on a central theme.

[xxlakis]

Hey Chris how are you?I like a lot this kind of methods and the one i use now uses the last 24 different spawn numbers with a pretty mild progression.I am really interested to your idea superman and the way you choose these numbers.If you would explain some more about this it would be really great.And you are right about inside bets,concerning me it's the best way to go.

[Skakus]

@superman,

So you’re eventually betting on a big pile of numbers, and you’re using a progression?

If you were to put a pseudo RNG into a bot and then tell it to bet on ‘however many’ selected numbers and use your current progression, do you think you might get very similar results to your existing package?

[superman]

So you’re eventually betting on a big pile of numbers, and you’re using a progression?

Yes, depending on the amount of losses the number of numbers covered rises, which does 2 things, raises the expense but gives more hits, which also slows down recovery, I have tried as many different progressions as possible, like I said earlier, you would think with 4 losses in a row it would be safe to use a progression that covers your losses +1 well, that backfired, as did labouche as did d'lambert so to keep bets low I decided +1 win or lose IF behind, yes it can be a grind clawing back but its probably the safest. There was a member here called Dyksexlic that basically stated he covers a lot of the table and the general feel was he was covering whatever had come out and if it came out twice he would put 2 chips on it, and just keep increasing by a chip whichever number hit last, I tried that and yes it wins a lot but it goes past a point of no return, example, when 5 or more numbers have 7 or 8 chips and a numbe rlands that only has 1 chip, you still lose, a lot.

the one I use now uses the last 24 different spawn numbers

Like I said earlier, the number we chaeck/track back to is something that needs a lot of testing, if you are running over the last 24 you are probably most of the time only getting 16 -18 numbers, which is good for a progression to recover but you must be getting a fair amoun tof high progressions when all 24 are unique! I have tried everything from the last 10 right up to 36 I found that when you are just getting unique numbers coming out for 10 extra spins the progression can do 2 things, either table limit OR they are just too high for the risk. Currently I look back over the last 40 but I don't think that will be the final number, I've morphed a few methods together that all catch diiferent numbers depending on appearance so in short its sleepers hot ones and a bit of a wild card method I made a long time ago that was quite good, not good enough to put money on, but accuracy was fair, basically I test different seeds values against the last and previous outcomes as this is what an RNG does, and the sum that creates the seed value uses the last outcome for the next, I know, we have no way of knowing THEIR seed value but hey it had a fair strike rate, and it gives another number to bet on, probably just luck.

And you are right about inside bets,concerning me it's the best way to go

I'm leaning that way, I have always said we are playing the RNG not roulette, the roulette table is just a convenient place to place bets.

If you were to put a pseudo RNG into a bot

Trust me mate, I first test against the random() function of the software I build the bots in, if it passes that then I run it against Playtech and finally onto BV NZ, the results I put in the first post were from BV NZ overnight. I have also in the past run methods against PHP random() too. I always test this way, since RXTreme using it's internal RNG.

[Chrisbis]

Yes, depending on the amount of losses the number of numbers covered rises, which does 2 things, raises the expense but gives more hits, which also slows down recovery, I have tried as many different progressions as possible, like I said earlier, you would think with 4 losses in a row it would be safe to use a progression that covers your losses +1 well, that backfired, as did Labouchere as did d'lambert so to keep bets low I decided +1 win or lose IF behind, yes it can be a grind clawing back but its probably the safest. There was a member here called Dyksexlic that basically stated he covers a lot of the table and the general feel was he was covering whatever had come out and if it came out twice he would put 2 chips on it, and just keep increasing by a chip whichever number hit last, I tried that and yes it wins a lot but it goes past a point of no return, example, when 5 or more numbers have 7 or 8 chips and a number lands that only has 1 chip, you still lose, a lot.
 

I tried this when I first got introduced to Roulette, and you feel very exposed when the *common* numbers are increasing, and the "Already Ran" numbers are looking decidedly 'naked'.

It reminds me of our bot- Bis-Sector-Hot-WFS, which, if your quick and can get in on the Rebet button, just as the bot is awaiting for the last Spun result, You can bet on the last sector that obviously missed, then the Bot bets on the latest sector, and if these are ding-donging between each other, and a third, You can get the common "Overlap" numbers rising in value.
But of course, what happens- the lowest covered (value wise) hits, and You have to abandon that attack, and take the loss.
What I was waiting for was to land on one of the *Juicy* -well endowed numbers, and make the profit, but the correlation between an Overlapped number, and the reality of hitting it (since it may or may-not, be) one of the previously Marquee'd numbers now showing/or just left, is not set in stone!
Its more likely that the "Juiced-up" number(s) was/were a sleeper(s) anyway! 

[superman]

I thinks it's safe to say with RNG forget anything to do with roulette, wheel layout or table layout AND the marquee, they are just a bi product of the spun number, the nu,mber it first then it gets divided into different areas of the felt, sectors should be left to real wheels and bias.

It's very rare for an RNG to give 36 unique numbers in a row, it can and probably does happen, I have not seen it, but you can rest assured it will at some point, so, a method that bets on whats already happened and grows/shrinks with the flow of random is what's needed, and a safe recovery progression too. I have made many bots that can run continually but again table limits or comfort of bet size are the issues, I'm under no illusion that it may never work but we gotta try anyway.

[xxlakis]

Well i always had the feeling that RNG is "trying" to be fair about results,at least more fair than real roulette.By fair i mean even distribution on colors,dozens,columns,numbers,...etc.I run a test somewhen with about 200,000 spins from BV No zero.I divided numbers to 3 groups,12 different last spawn,12 spawn before them and 12 sleepers and guess what...even with this groups the distribution of hits was equal.If this would happen in every session we would play i guess we would be winners huh?...

[superman]

If this would happen in every session we would play I guess we would be winners

That we would, but it's not that simple, I too have done extensive tests on 36/7 numbers, splitting them into groups to see when the re-hits happen, they are all over the place so we can't work with that.

[Nickmsi]

 Bingo Superman . . . I too agree that RNG is the way to go but only if your bet selection based on inside numbers.  I plan on having a bot coded and have found no problem in getting the bet selections to win, but as you point out it is the money management that requires fine tuning.


The table limits, as you noted,  poses the biggest problem.  As of now, using only BV or low limit Casino's using $.25, $.10, $.01 would be practical and profitable using a basic loss recovery system, ie. If your bet selection is for 5 numbers, you can bet them for 7 spins totaling 35 unit loss before increasing bet by 1 unit, and betting until your losses reach 70 units before increasing another 1 unit, etc. This can be extended quite far using only$.01 betting units.


These are some of the variables I think helpful to test out this system further:


(  ) # of cycles
(  ) # of spins to track in cycle (like 20, 40, 37, 74, 111)
(  ) # of bets to make (2-5 bets seem to work out best so far)
(  ) Minimum # of repeats for a number in cycle (ie. For a cycle of 111, must repeat 6 or more times    to be considered)
(  ) Minimum # of sleepers for a number in cycle(must have 0,1 times to be considered)
(  ) Minimum # of numbers to qualify for Dozens/Column bet (needs further testing, if 4 of the numbers picked are in same Dozen, would it be more profitable to bet the Dozen??)
(  ) Maximum # of bets to make( to test or set loss limits)


Progression:
     (  ) Flat Betting  (  )  Loss Recovery (Gradually recovers your losses) (  ) GLC Special ??


(  )  Loss Limit, Profit Target, Betting Unit, Bankroll or other standard variables


If cycle does not meet minimum requirements, it is re-tracked.


Any other variables you think would be helpful for this bot?


Many Thanks.   Nick

[superman]

no problem in getting the bet selections to win

How often and whats your longest losing streak?

If your bet selection is for 5 numbers, you can bet them for 7 spins totaling 35 unit loss before increasing bet by 1 unit, and betting until your losses reach 70 units before increasing another 1 unit, etc. This can be extended quite far using only$.01 betting units

Been there done that, to me it's looking like the more coverage you have the less losing streaks you will get, over a sample of 40 spins, you will get around on average 23 unique numbers, on average every 4 spins after that will be a new number, I said on average, which means you will have losses to contend with. What I am trying to find is the best amount to backtrack over so the loss streaks are as short as 1 or 2 in a row, we can't use any MM that recovers on the first win while playing that many numbers because when you get 3 or 4 losses in a row the bet is too high, it might not reach the table max BUT it will be a lot of chips, so keeping the bets low is priority.

[Skakus]

Trust me mate, I first test against the random() function of the software I build the bots in, if it passes that then I run it against Playtech and finally onto BV NZ, the results I put in the first post were from BV NZ overnight. I have also in the past run methods against PHP random() too. I always test this way, since RXTreme using it's internal RNG.

Actually I meant if you let a random number generator select the numbers to bet on would the result be any different?

[superman]

Ah sorry Danny, in short no, that is something I have tried before too, if you did that with whats being discussed you would have the same amount of numbers each time unless you first randomly select an amount of numbers then produce a random set of numbers from that, good idea but you will end up in the same boat, money management! so, in my opinion, it's best to let the RNG you are playing against tell you what to play, that's the theory anyway.

[Bayes]

it's best to let the RNG you are playing against tell you what to play, that's the theory anyway.

I agree. 

But I'm not sure what the theory is.