Repeaters roulette systems

[Steve]

You just can't beat roulette.

You can. There are only just a handful of ways it can be done, and they all revolve around winning number prediction.

[Proofreaders2000]

Firefox (a user here) was right about not using past spins imo.

(For example) You know every dozen has an equal chance of winning, right?

Why not bet equal parts all three dozens (the same numbers every time)

-or-

a sequence of your own: bet first dozen , third dozen, second dozen, repeat steps?

Same with even chances: Bet Odd, then Even, then Odd

-or-

Bet Odd twice, then even twice, then Odd twice....

[Steve]

Why not bet equal parts all three dozens (the same numbers every time)

-or-

a sequence of your own: bet first dozen , third dozen, second dozen, repeat steps?

Same with even chances: Bet Odd, then Even, then Odd

-or-

Bet Odd twice, then even twice, then Odd twice....

If your accuracy is same as random, then what has the system changed?

[Proofreaders2000]

If your accuracy is same as random, then what has the system changed?-Steve

You can test various combinations for the best hit-rate.

[Steve]

You can test various combinations for the best hit-rate.

Yes that's the start. You'll find some combinations perform better.

Next, you need to be sure your results aren't just luck. After all, you could bet red for 100 spins and profit. Then you may think betting red is always best. You'd be missing the bigger picture.

So you must test as many spins as is practical AND what's required for statistical relevance (ie almost eliminate the chance of luck playing a part).

In my experience, and the experience of countless others, you'll find all the combinations you thought might work actually have ZERO effect on prediction accuracy. This puts you back to square one, because if you dont change accuracy, you changed nothing.

Is there no hope? No, I'm not saying that. I'm saying try NEW things.
I think I've said it before.

And recognize when something you think is new is actually the same old thing repackaged.

[Steve]

And the reason I think 369 may have a clue is everything in the universe has an order. Even RNG, radio chatter from space etc are not "without cause". There is always cause and effect. But we cant so easily model cause and effect when there are countless variables.

Roulette computers essentially look at the most obvious variables, but that's after ball release. But what about with bets before ball release? There are still some usable variables. But what about the HG with bets before ball release or even RNG? That's a different story, because there are countless variables - everything from the dealer's itchy nose, to algorithms, when you click a button, to a bee fart. It's too much to calculate everything. So maybe look for fractals in the overall behavior, so you can predict waves.

Yes predicting waves like streaks has been done countless times before - in the proven wrong way. I don't think it's going to work, but if you're going to try, it should be in ways that haven't been done before.

If I had the time I'd design a program to:

1. Define "events". They must have clear definition.

2. Look specifically at proportions between the events, with parameters like 3,6,9.

An earlier program I made (programmer coded it) randomly looked for patterns. I mean everything like to see if a sequence of something like 34,5,1,7,19 had any relation to 2,6,15 etc. I'd leave it running for days at a time. I have no idea how many spins I tested. Billions or trillions maybe. I set it to check for repeaters too. And except in specific conditions and with real wheel data, the past spins had no correlation to future spins. I'm not the only one to have done such a thing.

It could be adapted to do this but I dont have the source code anymore. But there are programmers here. Why not have a structured discussion about what should be tried, then raise some funds as a community to have the software coded? Its better than a free coder because then it's unfair to them, and besides paid coders are faster.

[Proofreaders2000]

Ok, how about some parameters for a possible HG:

Flat-bet or positive progression.

Easy-to-remember bet-selection

Must be in plus (double bankroll or better) after 500 bets.

Time management: In profit
within 15 minutes for a typical session.

[Joe]

Joe, with statistics you can never 100% prove something. Simply the more data you have, the more assured you are.

Even if you test over 100,000 spins with the same system, sometimes the player will profit - especially when the system uses triggers, and ends up betting only on 10,000 spins. And especially with progression, where perhaps 1000 of those spins are high bets, and 9,000 are low bets. In such a case, there would only be 1,000 significant bets - which is still short term.

Steve, the point I was trying to make is that you don't need to make a huge number of bets to show statistical significance. It's very inefficient and not necessary. The point of statistical testing, confidence intervals etc is to draw conclusions by taking a sample from a population. In the real world you can rarely get the whole population, and even if you could it would probably be too expensive or require too many resources, that's why statistical methods were invented.

And in the case of testing roulette systems if a large number of bets are needed it would mean that either you need to be able to code your system or would have to spend many hours manually testing.  By using a test like the Chi-square, you need only make a few hundred bets. If the results are significant, that's not the end of the matter; of course you can't conclude that you have a holy grail from that alone. It would be a good start though, and from there you could either get another sample, this time using a lower significance level to reduce the chance of a false-positive, or maybe get the system coded.

You may think you have something 'significant' because you're making a profit, but making a profit doesn't necessarily mean that. Normal variance can keep you in profit for thousands of bets, especially if you're using a progression. Yes coding a million spins will ultimately reveal the truth, but you could do a hypothesis test on a few hundred spins which might tell you that your results are due to luck. Few people can code complex systems or have the patience to manually test thousands of spins. If a system seems to be working on paper the temptation is go out and play it for real money, which may be a really bad idea if your results were just a lucky streak. Hypothesis testing is a practical alternative to either coding or intensive manual testing.

[Joe]

1. Flatbet,
2. 50:50 even chance payout,
3. Longterm hiitrate - 70%,
4. Max drawdown - 30%

A 40% edge, that would be impressive.

[Tinsoldiers]

A 40% edge, that would be impressive.

Exactly.  And Steve that was the challenge he threw.

[Elite]

Hi LuckyFella, attach are the numbers, can you test this and give some suggestion, if your system in profit

[Person S]

Hello to all!
Any thoughts on how to limit variance?
I think that if you create a space in which there is a property of the presence of some constant (for example, 37 unique ones will not seem). Then the dispersion will have a weaker effect.
This is my idea, but possibly erroneous from the point of view of mathematics.

[Person S]

Get a template with controlled drawdown properties, see what is there? Sleepers, and those who woke up, some of them are more active. How about the fact that they can turn into pairs or can be lonely. And all this rotates against a background of binomial distribution, which can be compared with the sky, where there are no clouds)

[Steve]

A 40% edge, that would be impressive.

If I bet black for 3 spins, and win twice, do I have an "edge", or is it just a "result"?

Exactly.  And Steve that was the challenge he threw.

Unfortunately I dont have time to prove mundane facts like bet 3 x 1 numbers for a 3/37 chance of winning. I guess I'm an unknowledgeable chicken.

Any thoughts on how to limit variance?

You cant. Variance is a word for "unpredictable".

You can try decreasing bets when you think variance is unpredictable, but your accuracy is unchanged. You could miss big wins, or hit them. Chances are you'll have a bit of both, but mostly lose due to unfair payouts (hours edge).

Sleepers, and those who woke up, some of them are more active. How about the fact that they can turn into pairs or can be lonely.

This has nothing to do with future spins.

Flat-bet or positive progression.

Unless you first increase accuracy of prediction, progression wont help. All you'll have is different size bets on different spins with the same expectation.

Yes coding a million spins will ultimately reveal the truth, but you could do a hypothesis test on a few hundred spins which might tell you that your results are due to luck.

Yes you can take shortcuts, like when looking at wheel bias, you can correlate with other data like wheel observation. But with pure statistical analysis, you can extrapolate but still because of variance the only way to know for sure is large-scale tests. Even then it's not 100% assured, but it doesn't need to be. 99.99% is close enough. Most players wouldn't have the knowledge to interpolate/extrapolate, but using automated testers over a large amount of spins is easier, and a better option for typical players.

[Tinsoldiers]

I guess I'm an unknowledgeable chicken.

Stop advising people, if that’s the case. And don’t just read what you want to and understand what you want to, He is not asking you to bet 3 numbers, go back and read his post.  May be you don’t have time for that as well.