A SYSTEM THAT WORKS??

Well there you have it. I've been a keen bettor for more years than I care to admit. Value or 'edge' is everything.
Its drummed into you that you cant beat the house. You cant win long term by betting fixed odds games.

Of course your right. But what if you bet 'short term'?

And now either
a) I'm right and the Wizard is wrong or b) there is a blindingly obvious error in my idea

Try it for real? of course. It worked
Write a computer program to simulate it over millions of games? yup done that.

If it works why tell the world? good question. I think 99.99% of you will think its rubbish anyways.
But what bettor (sic) place to bear my soul?

Rules

Your bankroll should be an amount of money that you can afford to lose. It is highly improbable that you will lose, but its not impossible. You do this knowing there is a risk. It is gambling after all. So far I have not had it fail me, but one day it might.
If it does I am well in front anyway and I'm now using the house money, eg my profit.

You split your bankroll into 20 amounts or 'banks'

Whatever the house edge, your stake percentage is half the house edge. Lets call it 'A.'

Each bank plays at least 100 games but never more than 1000 games

If you lose you raise your stake by 'A.'

If you win but the current bank is still less than the previous highest bank the stake stays the same.

When your current bank is bigger than your previous highest bank then
the previous highest bank = the current bank and the stake resumes to 'A ' of the current bank.

Repeat until your have reached the total of games for that bank or your stake is bigger than your remaining bank and you cannot bet thereby leaving part of your bank untouched.

Start the next series of games with the next bank

You will be in profit before you have used 20 banks.
You will never lose all of your bankroll.
Starting with say 2000 units for your total bankroll you will never exceed the table limits

Why this works?
You are using the game's variance to your advantage and betting level stakes whilst winning and compounding the bank.
You are only raising stakes to retrieve losses at half the expected rate for the game.
You have limited your liability to 1 bank which is 1/20th of your total bankroll for each series of games and stakes will never exceed each bank. Therefore you cannot lose all of your total bankroll (betting funds).
It is virtually improbable (but not impossible) to lose after a series of 20 x 1000 games, you will be in profit well before 20 series.

If you keep betting infinitely (which in the real world is impossible to do anyhow) then at some point the house edge will beat you.
But by limiting your liability you are in fact just waiting for the variance of the series to turn in your favour.
When the variance is against you, you only lose 1/20th of your bankroll.
But when the variance is on your side then your bank grows and compounds to a limit of 1000 games at a rate of half the house edge.

If you play more than 1000 games per bank then your positive winning variance will change to negative and you will always eventually lose.

You could of course cut and run while ahead, but statistically you will overall win more by simply keep playing up to 1000 games per bank. Take away the initial bankroll from your winnings and you're then playing for free with the house money.

Am I wrong?

Quote: CeeJay

Am I wrong?

Simple answer, yes. I'm sure others will provide details on why...see any other system post.

But welcome to the forum and I like the time you put into this.

One problem I see: if you make a profit with your first bank, you still run the risk of losing money over the run of the remaining 19. Maybe I am reading this wrong, but you don't seem to consider that a loss. (Are you supposed to stop at the end of a bank when you are ahead for the session?)

One thing I don't understand: why do you say, "Each bank plays at least 100 games"? What happens if your bank can't cover the next bet before then?

Let me make sure I understand how this works:
Your initial bank for each of the 20 "rounds" is 1/20 of your total initial bankroll.
Your first bet is your initial bank multiplied by (house edge / 2)
When you lose, increase your bet by (initial bank multiplied by house edge); if you don't have enough money left in your bank, that round ends, and the next round starts with the bets reset.
When you win, if your current bank total is less than or equal to your initial bank, repeat your last bet; otherwise, bet your current bank multiplied by the house edge.
The only time (house edge / 2) is used is on the initial bet of a round.
A round ends after 1000 bets, or when you cannot afford to place the next bet.

For example:
Your bankroll is 2000, so the first round begins with a bank of 100.
If the house edge is 5%, your first bet is 100 x (5% / 2) = 2.50.
Assume you lose; you are down to 97.50, and your next bet is 2.50 + (100 x 5%) = 7.50.
Assume you lose again; you are down to 90, and your next bet is 7.50 + (100 x 5%) = 12.50.
Assume you lose a third time; you are down to 77.50, and your next bet is 12.50 + (100 x 5%) = 17.50.
This time, you win; you now have 95, but that is less than your "high water mark" of 100, so you bet 17.50 again
You win again, you now have 112.50, which is your highest value in this round so far, so your next bet is 112.5 x 5% = 5.625; which I will round down to 5.5.
You win a third time; you now have 118, which is your highest value in this round, so your next bet is 118 x 5% = 5.9, which I will round up to 6.
Keep going until this round has had 1000 games or you can't cover a bet.
Is this right?

Assuming it is, here's what I get:
With a 1% HA, you only come out ahead after 20 "banks" 6.45% of the time, and, in fact, each "bank" wins only 6% of the time.
The average length of a bank is 188 bets.

Hi
I'm glad I'm wrong. That way no one will copy me and I won't give my winnings back.
Suggest you write a sim.... lol

Don your multiplying the bank not the stake.

Also it resets with the new bank.

Quote: Boz

Simple answer, yes. I'm sure others will provide details on why...see any other system post.

But welcome to the forum and I like the time you put into this.

Absolutely wrong :o) But I cannot be bothered to prove it.
Take your money to a casino and DO IT.
That will be explanation enough.

This system will not win. Let's also be clear... This isn't a site with a bunch of non-educated people. This is a site with many very educated people and smart mathematicians whom are reviewing what you say carefully and determining your system is a loser. So when you get 100 responses saying the same thing, really think about how that many educated/intelligent individuals allllll came to the same conclusion that you're apparently missing.

Any system that starts with "let's just look at the short term" is a poor system. The only way to beat a game is to gain a LONG TERM advantage, and even then you're fighting the short term variance to realize the long term Expected Value (EV).

Even if you found a system that would guaranteed let you win TONIGHT, but lose in the long run, what good is it to win a few hundred, or a few thousand, in 1 night, if you can't ever play it again? If you can play it again, then short term is completely erroneous and you need to look at the long term effects.

Lastly, if you'd like mathematical proof on why your system will 100% for certain not work, post a realistic example...

Such as:

BR = $20,000
Each Bank = $1,000
Game = X
House Edge = X

Now play this many times until you reach this or that, then do this, blah blah blah... Spell out exactly how to play the system, and then many of us can show you how you have a negative expectation all along the way.

Whenever it comes down to you or the Wizard being wrong, the math says it's safe to assume you're wrong.

It's easy to prove it doesn't work. He's here broadcasting it!

Quote: CeeJay

You will be in profit before you have used 20 banks.

I AGREE. So quit ! Just like a Marty will see you in profit before 1/20 of your money wagered.

Quote: CeeJay

You will never lose all of your bankroll.

I sort of agree. You will lose MOST of your bankroll, roughly half of all lifetimes.

Quote: CeeJay

Am I wrong?

Yes. Still dead wrong

OnceDear's immutable law:-

For any game, session, set of sessions, lifetime of sessions
Probability of reaching Target profit <=(Starting Bankroll)/(Starting Bankroll+Target Profit)

With house edge its '<' and never '='

The system works perfectly, all systems do. In the long run, your losses will closely approximate your total money bet multiplied by the House Edge expressed as a decimal.

Quote: Mission146

The system works perfectly, all systems do. In the long run, your losses will closely approximate your total money bet multiplied by the House Edge expressed as a decimal.

i.e. Expected Value = TotalAction*HouseEdge = (NumBets)*(AvgBet)*(HouseEdge)

If you have a negative house edge, the expectation will always be negative.

Quote: CeeJay

there is a blindingly obvious error in my idea

There is.
By playing in the short term, your (likely) achievable profit is a small proportion of your initial lifetime bankroll.
I.e. Small profit is likely but pointlessly insignificant.

Quote: CeeJay

Hi
I'm glad I'm wrong. That way no one will copy me and I won't give my winnings back.
Suggest you write a sim.... lol

Don your multiplying the bank not the stake.

Do you mean that I'm supposed to be multiplying the stake (which makes no sense), or you think that this is what I am doing?

I did find one mistake in my description; I assumed "A" was the house edge, and not half of the house edge. Here's the new description:
Assume the bank starts at 1000 (i.e. your total bankroll is 20,000), and the HA is 2%.
The first bet is 1000 x 2% / 2 = 10.
Suppose you lose; you now have 990, and your next bet is 10 + 10 = 20
You lose again; you now have 970, and your next bet is 20 + 10 = 30
You win; you now have 1000, but that's not your highest so far, so your next bet is also 30
You lose; you now have 970, and your next bet is 30 + 10 = 40
You win; you now have 1010, which is your highest so far, so you reset your base bet (your "A" value) to 1010 x 2% / 2 = 10.1, which I round down to 10.
You lose; you now have 1000, and your next bet is 10 + 10 = 20
You lose again; you now have 980, and your next bet is 20 + 10 = 30
You lose again; you now have 950, and your next bet is 30 + 10 = 40
You win; you now have 990, but this is not greater than your highest amount so far (1010), so your next bet is still 40.
Note that your "base bet" is going to remain 10 until you get to 1100 or higher, when it becomes 11.

Assuming I have it right this time - and if I don't, then could you give me an example of how it works? - then, for a 1% house edge, I now get wins only 29.44% of the time, and the average bank lasts 369 bets.

The problem I see with this system, assuming I am implementing it correctly, is, when you have a run of losses, your bets don't keep up with the amount lost, so a subsequent run of wins will keep the bets the same as you don't reach your "high water mark", but further losses raise the bet even more until a string of losses wipes out your bankroll.

Something else I don't understand; in this case, 20 banks would last around 7400 bets. How many hours does it take you to get through 20 banks?

There are a hundred variations of this out
there, a version of it is in that 100 year
old gambling book. Move along, nothing
to see here.

Dude, cash in ... $30,000 is up for grabs

http://vegasclick.com/gambling/betting-system-challenge.html

Quote: Ibeatyouraces

It's easy to prove it doesn't work. He's here broadcasting it!

If he were sure it worked, he reeeeeeealy wouldn't broadcast.
Or maybe he's just a really nice guy....

Quote: TwoFeathersATL

If he were sure it worked, he reeeeeeealy wouldn't broadcast.
Or maybe he's just a really nice guy....

It Does work!!!!! If you define 'work' as . . .

Quote: CeeJay

You will be in profit before you have used 20 banks.

No. Really. It Does that.

But... Oh yeah...

Quote: OnceDear

You will probably be at a loss after you have used 20 banks.

Also it resets with the new bank.

What happens is when the games variance is in your favour the bank increases exponentially.
Whereas when the variance is against you the loss is less than 1/20 th of the bankroll.
You will make a profit before you have used 20 banks. So as I say you can cut and run or just keep going. You can't lose because you have limited your liability. But have virtually up to 1000games in which to grow the bank.
It's sort of similar to 13 against the bank except you won't have house limit problems

I'll post up a bit of code later. I quite understand the hostility and was expecting it. But apart from the usual negative expectation argument no one has shown why I'm wrong and Don still hasn't grasped it. No matter I'll try and make it clearer

Quote: CeeJay

when the games variance is in your favour

Hear that boys? Variance is a GOOD thing. I don't just want to take my EV and go home, I want to feel the emotional roller coaster along the way.

Quote: CeeJay

Also it resets with the new bank.

It seems like 1000 is an arbitrary place for the variance to change from positive to negative, assuming you ever get that far. I also don't see how each game is guaranteed to be at least 100 games long; you can certainly bust out of a bank before 100 bets with long losing streaks pushing your bet higher, then staying high until you get back above par.

It seems to be a modified Martingale, which still has a negative expectation overall. If TDG's analysis (the corrected second one) is right, then a system that only wins (not go bust) 29% of the time doesn't seem like it would be useful, unless you have a real expectation for wins on 6 banks collectively large enough to offset losing the other 14 (where they would not only need to win, but each to more than triple-up before you reach 1000 hands). Or is your system depending on stopping short of all 20 banks at risk, so at any point you profit overall you stop? If so, all your system is doing is keeping you from going on tilt THAT session, not paying you consistently.

Perhaps you could provide a practical example of a betting progression using numbers on a single bank so we can see/verify the pattern you're describing.

Quote: CeeJay

Also it resets with the new bank.

That's twice you have said this. What "resets"? Your bets? Your entire system?

One problem is, you are not clear as to when to stop. Do you play out each bank until you get to 1000 bets or run out of money for that bank? And did you take into account that, after 1000 bets, you can end up with less money in the bank than you started?

It would be easier if you gave an example of what to bet.
For example, suppose you are playing a game with a 2% house advantage, and your total bankroll is 2000 units - this is 20 "banks" of 100 units each, right?

(a) What is your first bet?

(b) Assume you lose your first bet - what is your second bet?

(c) Now assume you win your second bet - what is your third bet?

(d) At what point do you say, "This bank is done - let's start over with bank 2"?

(e) What is your first bet with bank #2? Shouldn't it be the same as for bank #1? Is this what you mean by "it resets with the new bank"?

Quote: CeeJay

What happens is when the games variance is in your favour the bank increases exponentially.
Whereas when the variance is against you the loss is less than 1/20 th of the bankroll.
You will make a profit before you have used 20 banks. So as I say you can cut and run or just keep going. You can't lose because you have limited your liability. But have virtually up to 1000games in which to grow the bank.
It's sort of similar to 13 against the bank except you won't have house limit problems

I'll post up a bit of code later. I quite understand the hostility and was expecting it. But apart from the usual negative expectation argument no one has shown why I'm wrong and Don still hasn't grasped it. No matter I'll try and make it clearer

So, because the usual argument is usual, you don't believe it?

Variance can't be in your favor or against you. Variance is always positive (unlike EV) because it's a squared number. The magnitude determines how spread apart your outcomes are around the mean, but that doesn't say anything about what that mean outcome is. In your case, you're making a bunch of independent bets, each one of which has a house edge. But you think somehow that the order in which you make those independent bets can guarantee a profit. Think about how that could possibly be correct.

FYI, when you suggest "you will make a profit" or "you can't lose" -- at the outset those are the claims you need to prove. If you're relying on the truth of those claims in order to prove that your system works, that's circular logic and your conclusion is invalid.

Hmmm you're not reading what I have written.

here's the code.

Example House edge = 5.26% for American roulette

bank= bankroll/20

stake = bank * 2.63 / 100

for j = 1 to 20

For i = 1 To Val(Games) 'minimum 100 maximum 1000

n = random number 1 to 100

If n > 0 And n <= 5.26 Then a = a + 1 'green or house

If n > 5.7 And n <= (100 - 5.26) / 2 Then b = b + 1 'black

If n > (100 - 5.26) / 2 Then 'red winner

c = c + 1

bank = bank + stake

If bank > OldBank Then
OldBank = bank
stake = bank * 2.63 / 100
End If

Else

bank = bank - stake
stake = stake + (stake * 2.63 / 100)

End If

make list = i, bank, stake
display a,b,c

If bank < stake Then Exit i 'stake too big for bank leaving remainder aside

Next i

list totals

Next j ' the series of 20 can restart at anytime

Quote: CeeJay

n = random number 1 to 100
If n > (100 - 5.26) / 2 Then 'red winner

Of course your system works on a roulette wheel where red shows up 52% of the time. So does flat betting. Try your system on a realistic simulation of a roulette wheel and it'll lose at the expected rate.

Here is a short series of only 50, you need more than 100 and maximum 1000

starting with a bank of 100 you need a bank roll of 2000

1 97.37 2.7
2 100.07 2.63
3 102.7 2.7
4 105.4 2.77
5 108.17 2.84
6 111.01 2.92
7 113.93 3
8 110.93 3.08
9 107.85 3.16
10 111.01 3.16
11 114.17 3
12 117.17 3.08
13 120.25 3.16
14 123.41 3.25
15 120.16 3.34
16 116.82 3.43
17 113.39 3.52
18 116.91 3.52
19 113.39 3.61
20 109.78 3.7
21 113.48 3.7
22 109.78 3.8
23 113.58 3.8
24 117.38 3.8
25 121.18 3.8
26 124.98 3.29
27 121.69 3.38
28 125.07 3.29
29 128.36 3.38
30 131.74 3.46
31 135.2 3.56
32 131.64 3.65
33 135.29 3.56
34 131.73 3.65
35 128.08 3.75
36 131.83 3.75
37 135.58 3.57
38 132.01 3.66
39 135.67 3.57
40 139.24 3.66
41 135.58 3.76
42 139.34 3.66
43 143 3.76
44 139.24 3.86
45 135.38 3.96
46 131.42 4.06
47 127.36 4.17
48 131.53 4.17
49 127.36 4.28
50 131.64 4.28


Here are the results after 20 series of 50 games

1 124.45 24.45
2 91.48 15.93
3 115.47 31.4
4 135.97 67.37
5 109.36 76.73
6 91.11 67.84
7 95.41 63.25
8 132.95 96.2
9 109.32 105.52
10 104.75 110.27
11 127.13 137.4
12 64.9 102.3
13 68.58 70.88
14 134.54 105.42
15 115.37 120.79
16 108.43 129.22
17 113.58 142.8
18 138.18 180.98
19 75.19 156.17
20 131.64 187.81 - this is the last series where the stakes are shown above

your profit is 187.81 with a 20x50 bankroll. This is a very typical result, but with 1000 games in a series your profit is considerably larger.

Quote: CeeJay

Here is a short series of only 50, you need more than 100 and maximum 1000

your profit is 187.81 with a 20x50 bankroll. This is a very typical result, but with 1000 games in a series your profit is considerably larger.

That's because your sim is broken. Fix it before carrying on like this.

youre a maths expert?

(100 - 5.26) / 2 = 52?

where is it broken? That's was the reason for my posting the system here.

Quote: CeeJay

youre a maths expert?

(100 - 5.26) / 2 = 52?

If you can't read your own code, I can't help you.

"n = random number 1 to 100
If n > (100 - 5.26) / 2 Then 'red winner"

Maybe I can help. Try implementing your game using the 38 numbers that are actually on a roulette wheel instead of trying to do whatever you're doing.

typo

If n > 5.7 And n <= (100 - 5.26) / 2 Then b = b + 1 'black


should have said

If n > 5.26 And n <= (100 - 5.26) / 2 Then b = b + 1 'black

but thats the point. where is the error?

Quote: CeeJay

but thats the point. where is the error?

You have three cases, green, black, and red (in that order). You're attempting to calculate probability ranges for those cases from 1 to 100, but your math is wrong because it doesn't match a real roulette wheel. In your case, red appears over 52% of the time, so of course your sim produces positive results: you're essentially playing an even-money game where you have a 52% chance of winning.

You could either fix it or, even better, draw numbers from 1 to 38 instead because that's how many spots are actually on a roulette wheel. You can use any appropriate mapping; here's one: 1 and 2 are zeros, 3 through 20 are black, 21 through 38 are red.

Maybe I can help. Try implementing your game using the 38 numbers that are actually on a roulette wheel instead of trying to do whatever you're doing


but this sim is just for red/black/double zero ?

The procedure is the same for whatever rules you are playing eg a random number to provide whether the bet is a win or lose.

Quote: CeeJay

The procedure is the same for whatever rules you are playing eg a random number to provide whether the bet is a win or lose.

That's the part that's broken. Your attempt to "use a random number to provide whether the bet is a win or lose" is buggy. In your code, red wins over 52% of the time. That is an error.

weird how it has worked for months for real.


Ok, thanks for the feed back which was the point of my thread.

Quote: MathExtremist

That's the part that's broken. Your attempt to "use a random number to provide whether the bet is a win or lose" is buggy. In your code, red wins over 52% of the time. That is an error.

Absolutely. Everybody knows black wins over 52% of the time. DUH !

Quote: muleyvoice

Absolutely. Everybody knows black wins over 52% of the time. DUH !

That's only when you're betting on red, though.

Quote: CeeJay

weird how it has worked for months for real.
.

Not weird at all. Many things work
for awhile and then stop.

Here are two that I found:

First, I don't see in your code where OldBank is set; it needs to be reset to Bank at the beginning of each pass through the "for j" loop.

Second:


When you lose, your next bet = your current bet * 1.0263. In your description, it says that your next bet = your current bet + 2.63 x the "old bank" - i.e.:



If your initial bankroll is 2000 and bank starts at 100, the initial stake = 2.63, but if you lose, stake becomes 2.63 + (2.63 * 2.63 / 100) = 2.699169; shouldn't it become 5.26?

Also, when you play at an actual table, what is your actual stop condition - when do you get up from the table? For that matter, when do you decide to start a new "bank"? 1000 spins of a roulette wheel per bank, and 20 banks per session, is 20,000 spins.

I think a system that works is waiting for a progressive jackpot to become +EV, then borrowing your boyfriends credit card and having some fun. If you win, you pay him back. If you lose, then it wasn't your money lost!

n = random number 1 to 38

If n <= 2 Then a = a + 1 'green or house

If n >= 3 And n <= 20 Then b = b + 1 'black

If n >= 21 Then c = c + 1 'red winner



Try your sim with this instead. Like MathExtremist said, you're currently rolling red 52% of the time.


100-5 = 95
95/2 = 47.5
N between 1 and 100
N will be > 47.5 52.5% of the time


Also, as far as "working for a few months so it must be a sound strategy": I know a couple guys that made 6 figures playing basic strategy blackjack with $25-$500 bets. They weren't counting, they had a distinct disadvantage and still won over the course of a year or two.

Eventually they lost it all back.

Quote: MathExtremist

If you can't read your own code, I can't help you.

"n = random number 1 to 100
If n > (100 - 5.26) / 2 Then 'red winner"

Maybe I can help. Try implementing your game using the 38 numbers that are actually on a roulette wheel instead of trying to do whatever you're doing.

I can't even code a security alarm and I understand the problem with what ME quoted!

Thanks for all the replies
So it looks like I was just plain lucky!

Should have bought a lottery ticket ☺

You do get credit for coming back and saying 'thanks I see where I was wrong' which most people would not do.

Most of the work showing where error lay was over my head. But one way I could tell you were going wrong was it seemed your system relied on breaking up your betting into small little units. This is always going to be the wrong direction for negative expectation, otherwise not only would your system be the first to work but it would negate the Kelly Criterion, which is based on the idea that positive expectation is best exploited by smaller rather than larger bets [and also determines the largest ideal bet to do this]

I'm just saying, if you are still obsessed with working on betting systems, don't go down this same road.