[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math

View post   
File: 78 KB, 1020x474, grail.jpg [View same] [iqdb] [saucenao] [google]
2659170 No.2659170 [Reply] [Original]

They told me beating roulette long-term is impossible...they were wrong.

pic related

>> No.2659175

That's not long-term at all.

>> No.2659183

>>2659175

compared to other systems that have failed it is, furthermore my graph sets a sawtooth trend, which proves my system has potential but may still need tweaking.

they are real-live wheel spins btw, RNG is a fool's game.

>> No.2659196

>>2659183
You had a lucky streak after game 250. Whoop-de-doo. Show my 10 000+ games with profit and come back.

>> No.2659201

>>2659170
When does your system choose to stop betting?

>> No.2659203

>>2659183
what's the difference between real live spins and RNG? apart from superstition?

>> No.2659205

>>2659196

In time I will, also your missing the point here it's not about my lucky streak at the end of the graph it is about the system's ability to recover from a loss.

I am the first one to say my method is not mathematically perfect to beat the odds, but it definitely works.

>> No.2659207

>>2659203
Most RNGs will end up in a predictable pattern if you ask for numbers continuously as part of a simulation. Timer-based methods end up making patterns in time, procedural methods end up repeating by iteration. The repeats can end up favoring tactics that might not be viable in a real world situation.

>> No.2659210

>>2659201

when the progression has ended.

>>2659203
RNG algorithms are programmed for you to lose in the long-run, meaning they are not purely random, they follow your betting pattern and then hit you hard. Many other people with experience can tell you this, including myself.

>> No.2659215

>>2659210
When does the progression end?

>> No.2659216

>>2659207
couldn't you just use random.org? either that, or do 10.000 games, I don't think it starts repeating after 10.000 games already, especially since there are far more than 10.000 seeds possible

>> No.2659225

>>2659215

sometimes I reset the progression, but I can usually quit my sessions, depending whenever I feel like doing something else.

>>2659216
I can but it will still take me time to process the numbers through my system, furthermore real-time test results are more reliable as it's real.

>> No.2659230

How many green spaces (zeroes) does your roulette board have? Real roulette has one or two of these. They are what makes roulette a losing game long-term.

>> No.2659246

>>2659225
If your system "plays until it loses" then you lose.
If your system "plays until it gets $X profit" then you have a chance >50% of losing.

>> No.2659247

>>2659230

I use a one-zero table for obvious reasons; and yes it does heighten the odds.

>> No.2659257

>>2659246

my betting sessions are done at my free time and leisure, I would imagine if I do it throughout the day I would have more of a balance,

when I feel like it I stop betting. as it takes roughly an hour of my day.

>> No.2659259

>they were wrong.

No, they weren't. The odds against winning are higher than the payout for every possible bet. You're either playing by incorrect rules, or the game is not really random, or you're full of shit.

>> No.2659260

>>2659170
the technique of betting red and double the bet until you win would probably be positif with only 600 try too... (with a starting stack of 1000, and a betting start of 1)

>> No.2659274

>>2659259
>incorrect rules

playing at an online casino

>not random

live wheel

>full of shit

that actually gives me more confidence, thanks bro.

>>2659260
Martindale progression is bad I agree with you, I don't use it.

>> No.2659292

There are betting systems you can use that give a decent chance of winning in medium term, but that chance is never greater than 50%, and so the best case scenario is that you gradually lose all your money in the long term.

You got lucky. lol, and quit while you're ahead.

>> No.2659290

>>2659274
My point was that even a known losing technique could appear to be winning with so few tries.

>> No.2659299

We're all waiting for you to post the algorithm brah

>> No.2659309
File: 48 KB, 586x446, 1266733632101.jpg [View same] [iqdb] [saucenao] [google]
2659309

>>2659299

>> No.2659320

>>2659259

He's probably not full of shit. He just doesn't understand that he got lucky. A losing strategy (which all strategies are in a fair game) can still win over a long number of games - it's just more likely that it will lose.

>> No.2659321

>>2659292
>>2659290

I started off with 1000 and once I reached the 2000 mark I withdrawn 1000 units,

so in other words I got my money back and the only thing I might be wasting is time.

>> No.2659325

The player only has one advantage: he can choose when to quit.

You're ahead. Quit.

>> No.2659329

>>2659321

You could do the same thing in one bet. It's just more likely to lose.

>> No.2659334

>>2659309
>>2659299

even though I am playing real, I am still testing. once I finished testing and made a bit out of it, i'll then consider giving it out.

if I post it and eventual testing shows that it is flawed, you will lose aswell...I can't have that on my conscious.

>> No.2659339

>>2659334
What's there to discuss then? We run long term testing and just have to trust you on your word. I'd love to try your system in a simulation in MATLAB.
>if I post it and eventual testing shows that it is flawed, you will lose aswell...I can't have that on my conscious.
Personal responsibility bro. If people try out your algorithm chances are they're gamblers anyway.

>> No.2659338

>>2659329
>>2659325

ok bro, I hear your advice, but note I am still on the testing phrase despite it being real money. I will update you guys on this board monthly

>> No.2659343

>>2659339
>We run long term testing
We CAN'T run long term testing*

>> No.2659344

>>2659334
>testing

See, there's the problem. If it were valid, you should be able to prove it mathematically, not rely on trial and error.

>> No.2659348

>>2659325

casinos can force you to quit while you're behind though specifically to stop that strategy

>> No.2659356

I run simulations in stuff like this, at one time I was up after 10000 tries.
Try a much larger sample, you are deluding yourself.

When I tested systems I usually ran 10 million tries at least. You should do the same.

Also the rand() function in C at least has some exploitable tendencies, so at least try a better one, or a hardware one would be best if you are serious, and if you try this system with real money then definitely.

>> No.2659358

>>2659339

not really you can argue it's not long-term as I only have 500+ spins, but then you can also argue:

long term = infinite

like I said the system is not perfect, but I really think it does have potential,

btw it's likely gamblers won't be able to use my system since it requires patience and concentration. In other words once they lose a progression or go under their bankroll...they would instantly dismiss it.

>> No.2659373

>>2659356

I can, and I might consider it, but I need the following:

RX coding knowlege
a PURE RNG

the first one can be gained, but I have the problem with the latter, is the RNG you recommend good?

>> No.2659376

programing a roulette is pretty simple, and having a million tries pretty fast, you will know for sure if you are a future billionaire or just a lucky man and withdraw your gain as fast as you can.

>> No.2659378

>>2659358
>not really you can argue it's not long-term as I only have 500+ spins, but then you can also argue:
I'd like to run it for at least 100000 rolls before taking it seriously. Proving it mathematically will be more convincing of course, but I simply don't know how to do that.

Wouldn't it help you as well if someone scrutinizes it?

>> No.2659381

play for an infinite amount of time using random betting. At some point you will have double what you started with, and you can quit with twice what you started with.

Keep leaving and entering the casino to reset your luck. Enjoy your infinite moneys

>> No.2659384

>>2659344

This. You don't have to "test." Every strategy has a CALCULABLE probability that it will be winning after N games, and that probability is ALWAYS less than 1/2.

>> No.2659395

Here's my strategy:

Roulette wheels aren't perfect, and tend to land on certain numbers more than others (if your casino is particularly dishonest, it will land on 0 more often)

Go to the casino and watch the wheel all day, and make a mental list of which numbers win the most that day (note: you can't use a pen and paper and you have to act like you're gambling or the casino will kick you out. I suggest hopping on a penny slot machine while you watch to avoid getting kicked out)

Now after a case study performed over a few weeks you will find the most lucky number. For mine it's 4 black, but it could be anything! Simply bet on that number every time and statistically you might come out even. This works every time.

>> No.2659403

>>2659378

I agree, the more spins the better, I am struggling to think of an explanation mathematiclly to this, but eventually I will reach 10000 spins while making money when I do they'll be a printscreen with your name on it.

>> No.2659404

>>2659384

for N games, but you don't have to play n games. You could play n-1 games or n+6 games

>> No.2659406

>>2659404

N is arbitrary....

>> No.2659411

>>2659403
>10000 spins
I said 100,000 though, not 10,000.

>> No.2659420

I'm sure we can deduce the algorithm from the pattern in the graph. Looks like you have a set amount to bet, and increase it after you lose.

>> No.2659422

>>2659411

that'll take years, however if I am still winning it will all be recorded bro.

>> No.2659423

>>2659406

but you don't understand. As a gambler, you don't play an arbitrary number of games. You set a goal of a certain amount of money you want to make or lose and leave once you hit that goal. Assuming you have a large amount of money to spend and can leave as soon as you turn a profit, that means there is some betting strategy that exists to maximize profit.

>> No.2659424

>>2659420

yes I said that already bro

>> No.2659427

>>2659422
>that'll take years
Not in a simulation though. After coding I could have it done in under an hour. I have my university's server available.

>> No.2659426

>>2659384
Ok then here's a simple strategy, I play perfect pontoon, which is a blackjack type game, where the casino has a 0.4% edge.
The only thing I do differenty is deciding when to quit. I stand up every time I'm up 5 units, and I have a stop loss at 10.
Whats my long term expectation?

>> No.2659440
File: 92 KB, 1020x474, grail.jpg [View same] [iqdb] [saucenao] [google]
2659440

for anyone who didn't get the graph

>> No.2659447

>>2659440
>over 9000 hours in ms paint

>> No.2659456

>>2659426

Not who you're responding to, but obviously you'll lose. To come out ahead, you'll need to "win" more than twice as often as you "lose," and that won't happen.

>> No.2659464

>>2659423
>but you don't understand. As a gambler, you don't play an arbitrary number of games. You set a goal of a certain amount of money you want to make or lose and leave once you hit that goal. Assuming you have a large amount of money to spend and can leave as soon as you turn a profit, that means there is some betting strategy that exists to maximize profit.

Incorrect. That way you'll have many small wins and a few large losses, that in the long term will always work against you.

>> No.2659468

ITT: We learn that gamblers don't know math. (If they did, they wouldn't be gamblers.)

>> No.2659482

hey op, einstein did research on roulette and after his research he came to this conclusion:

"The only way to beat roulette is to steal chips from the table"

now are you attempting to refute one of the most mathematical intellectuals that ever lived?

>> No.2659495

Okay OP, if you're not going to post your algorithm I think I'll make may own. There are roulette functions for matlab available online for free.

This is actually kind of fun, so thanks for starting the thread.

>> No.2659513

>>2659495
>may
*my...

>> No.2659516

<3 /dev/random

>> No.2659575

My friends developed a good winning strategy for roulette. I'll even share it with you all. They would find a low-limit table and then bet a large amount, distributed over the field. Pretty much they had a 35/37 chance of winning something; they just wouldn't ever bet on 0 or 00. Thing is, for most of their numbers they would win a small amount, or at least only lose a small amount. So usually they'd win something. Of course, this doesn't work in the long run. But, there were two things they would do. If they managed to get up by $100 or $200, they'd take their money and leave. Even if they lost, they'd lose only about $100 or $200. In the mean time, they'd have gotten tons of free drinks, an evening of entertainment, and between the two of them enough points to get a free night or two comped by the casino.

Their strategy was not to maximize their winnings, but exploit the comp system. The longer you play, the more comps you get. Sure, they paid a few hundred dollars over the weekend, but given all the free rooms and food and drinks they got, that's a fairly good deal.

This is a question of game theory, where you have different strategies and goals. It's much more about a wise choice of goals than the statistics of the game.

>> No.2659587
File: 78 KB, 1177x620, strat.jpg [View same] [iqdb] [saucenao] [google]
2659587

here is the results i just came up with, it work like this:
devide bord into 3 areas:
1,2,3 (this can be done any way)
if previous number was 1, bet on 2 and 3
if previous number was 2, bet on 1 and 3
if previous number was 3, bet on 2 and 1
repeat

this will obviously not work in real live, but it seems to work when moddeld

>> No.2659600

>>2659587
>implying past roles influence future roles.

>> No.2659602

>>2659468
The March Meeting (the big physics annual meeting) was held once in Vegas. The Vegas people figured that people who make six figure salaries would probably love to gamble, so they gave them each something like $100 or $200 to get them going. Of course, they physicists, being at least decent at math and statistics, did the smart thing. Bet only as much as they had already been given, and then walk away.

They weren't invited back.

>> No.2659629

>>2659602
Physicists don't make six figures...

>> No.2659625

>>2659207
You cannot into RNG theory.
There are plenty of good RNG algorithms that are just as good as "real spins." The fact that you don't know about them doesn't lessen their usefulness.

>> No.2659632

>>2659587

you made approx 500 units in 2000+ spins

op made approx 800 units in 600+ spins

op post your algorithm...now

>> No.2659639

>>2659602

I would have just kept the $200 and loaded up on free drinks and appetizers.

>> No.2659641

You will lose to the house by 0.4%.

>> No.2659761

>>2659600
thats why I sayd it shouldn't work, but in the models it works. my point is you cant try to model this shit

>> No.2659768

>>2659761
If only there were a way to secure confidence within come kind of interval.

>> No.2659779

great
Now all I need is a casino that will not kick me out for playing >100 times

>> No.2660162

>>2659629
Many do. Many of them don't work in academics. Many of those in academics also do consulting on the side to supplement their regular salary.

http://www.payscale.com/research/US/Degree=Doctorate_%28PhD%29,_Physics/Salary

True, many physicists make less than $100k/year, but many do. Often their official title is "researcher" or "scientist" but their backgrounds are in physics. Even those that don't make $100k, most make something close to that. They're not exactly poor.

>> No.2660204

>>2659768
Monte Carlo simulations. If I had nothing else to do right now (damn QC thesis!) I'd do it. Wouldn't take more than a few hours, but you need to know a bit about Monte Carlo simulations to make it work right.

For those interested, set up a strategy X and apply it to M spins of the roulette wheel. Record the amount of funds after each spin. Now do this N times (N >> 1, obviously, preferably something like N = 1000 or 100 000), each time recording the funds. Now, for each spin n = 1..M count the number of simulations with an amount K in them. This will give you a probability distribution for the amount of money after spin n. You can draw contours for the probability.

These contours will go to x = X0 - p n for n -> infinity, where X0 is your starting amount of cash, x is your remaining funds, and n is the number of spins. The standard deviation of x will go to zero as n gets large. However, for small n it will grow to some maximum value before inevitably decreasing. That value n0 (which is a function of the contour value) is the point at which it makes sense to stop.

Also, the lower your contour, the more likely you are to reach it, of course. It means smaller payout, but more likely to have SOME profit. You can do the math yourself to figure out which contour conforms to getting paid minimum wage of about $6/hour. Of course, my goal would be to play as long as I'm not down any more than I can expect to recover.