Roulette has no memory?

[Steve]

You are continually repeating the same thing : increase accuracy!

Should simple truth change?

Sorry but that’s too vague for most of people here

For vague people, yes. For others, it's pretty clear. I've also gone into detail many, many times. And still vague people don't get it.

I am sure you would immediately say physics is the way to go!

It's one way. The obvious way. What else determines the winning number? Voodoo? Streaks? Numerology?

I've also explained other possibilities in detail.

I don’t want to disappoint you but physics approaches are too difficult to implement nowadays, why ?
Simply because casino are imposing tight restrictions on the roulette table

Nonsense. Almost every wheel can still be beaten with one or another physics approach.

And try new approaches

I say it often. People don't listen.

one of the most solid approach in this regard is the so called (reinforcement learning)... I could recently gather some evidences that it could beat the game, the only thing is that it needs special programming skill, unfortunately the majority of people lack it

Reinforcement learning? What's that technique as it pertains to roulette?

Either way, most people don't even understand basics. That's where they need to focus, as boring as it is.

[Roulettebeater]

Should simple truth change?

For vague people, yes. For others, it's pretty clear. I've also gone into detail many, many times. And still vague people don't get it.

It's one way. The obvious way. What else determines the winning number? Voodoo? Streaks? Numerology?

I've also explained other possibilities in detail.

Nonsense. Almost every wheel can still be beaten with one or another physics approach.

I say it often. People don't listen.

Reinforcement learning? What's that technique as it pertains to roulette?

Either way, most people don't even understand basics. That's where they need to focus, as boring as it is.

RL is like teaching a software agent how to behave in an environment by telling it how good it’s doing, it’s an area of machine learning inspired by behaviorist psychology

[Steve]

A form of machine learning is how i developed one of my systems. I've written lots about it and what I found. Many times ive left software running for long periods to see what it found. I even published a very rudimentary version of it free.

Basically it correlates variables with spin outcomes to find statistically significant anomalies to exploit.

[gianfrancopierino]

A form of machine learning is how i developed one of my systems. I've written lots about it and what I found. Many times ive left software running for long periods to see what it found. I even published a very rudimentary version of it free.

Basically it correlates variables with spin outcomes to find statistically significant anomalies to exploit.

Steve can you put here link with your tests and results please?

[Proofreaders2000]

What abt the premise previous outcomes don't matter.
That said one bet is no better than another.

[FreeRoulette]

A form of machine learning is how i developed one of my systems. I've written lots about it and what I found. Many times ive left software running for long periods to see what it found. I even published a very rudimentary version of it free.

Basically it correlates variables with spin outcomes to find statistically significant anomalies to exploit.

If the AI that plays Go, were to play roulette, I wonder if the best outcome would be not to play. If you forced it to play, would it only do the minimum bet on one number only once and stop? Lets say you give it $1,000 that it has to play to maximize the profits. Would it blow its wad on one spin?

I would be interested in what your machine learning algorithm did.

[Roulettebeater]

With all respect to your contribution but RL is not same as simple machine learning, RL is an adapted technique which allows the agent to adjust its decisions regularly based on the data flow

[FreeRoulette]

loothog.com/Systems/repeat2/repeats.html

[FreeRoulette]

loothog.com/Systems/repeat2/repeats.html

Has anyone tried to have each step have its own progression? When I get time, I will make a program to keep track of 25 progressions at once. It might only reach 25 once, but you only be out $25 if it missed, then 50 if it missed the second time, if it ever reached 25 uniques again.