How do you calculate cumulative probability?
#244
Flippin' radar engineers, they get everywhere
Just going to get out my copy of Handbook of Roulette Wheel Scattering Statistics by Ulaby... Hold on...
I knocked up a VERY noddy simulation of this in MATLAB, only took 5-10 mins, very basic, but it does show that this strategy... lets see what is the diplomatic way of putting this... ah, I know, sucks.
Basically the pyramid scheme is a red herring, it doesn't help. For the pyramid to be effective, you need to double your money more often than you lose it, i.e. an expected return of greater than what you put in. Unfortunately your game strategy does not achieve this.
I haven't explored with variations or optimisations of the strategy (I could do, but it would take more than 5 mins and I would have to charge ) but on the basis above I don't think there is likely to be a solution.
Just going to get out my copy of Handbook of Roulette Wheel Scattering Statistics by Ulaby... Hold on...
I knocked up a VERY noddy simulation of this in MATLAB, only took 5-10 mins, very basic, but it does show that this strategy... lets see what is the diplomatic way of putting this... ah, I know, sucks.
Basically the pyramid scheme is a red herring, it doesn't help. For the pyramid to be effective, you need to double your money more often than you lose it, i.e. an expected return of greater than what you put in. Unfortunately your game strategy does not achieve this.
I haven't explored with variations or optimisations of the strategy (I could do, but it would take more than 5 mins and I would have to charge ) but on the basis above I don't think there is likely to be a solution.
#245
You don't need to use Matlab.....or even our best VHDL engineers to route this problem into a Xilinx......
All you need is this : 35/37 (or 36/37 if that is the European table).
I've been banging on about this number since the start. It is the only solution to this problem. It really is as simple as that.
ie, over the long term, use this formula for your overall return, where x is the initial amount :
total profit = (x * 35/37) - x
All you need is this : 35/37 (or 36/37 if that is the European table).
I've been banging on about this number since the start. It is the only solution to this problem. It really is as simple as that.
ie, over the long term, use this formula for your overall return, where x is the initial amount :
total profit = (x * 35/37) - x
#246
Scooby Regular
Thread Starter
Join Date: Mar 2002
Location: Bradford
Posts: 13,720
Likes: 0
Received 0 Likes
on
0 Posts
But if you played my system where you are trying to lose that number from a starting value of £100,000 would be 0 and not -£5405 like it should be. Through your method of playing you have altered your formula albeit for the bad in this case but altered nevertheless hence by logic it cannot be as simply represented as that formula
#248
True, you don't need MATLAB, but it is one hell of a lot easier to code in MATLAB than VHDL. Trust me on this one
Of course 35/37 is behind it, but glibly quoting that figure doesn't help to understand why the problem fails. It is easier to understand by considering the individual game and the gearing of expected returns. Essentially the expected return is this 35/37, but you can gear it to make a highly probable small win against a highly improbable large win, i.e.
Expected Return = fixed
Expected Return = P(small win)*small win value + P(big loss)*big loss value
A well optimised system should offer a good balance here, i.e. the ratio of P(small win)/P(big loss) is directly related to (big loss value)/(small win value).
Thus you can make it highly likely (but not guaranteed) that you increase your initial stake by a small amount, which is the appeal of this system. We know that ultimately we may lose, but with sufficient "credit" we can make this infinitely improbable. (Hitchhikers guide anyone?)
However the failing is the requirement to achieve double or nothing, which is the principle of Saxo Boy's "sub-games". To achieve double or nothing, our (small win value)/(big loss value) is near one, which means the P(small win)/P(big loss) is near one, and with the tables "advantage" (that good ole 35/37 again) the odds will be stacked slightly against you.
As it happens, Saxo Boy's game is poorly optimised, and only successfully doubles his money around 10-20% of the time Basically you'd be better off sticking all your money on red and looking to double it that way. You'd still lose, but you'd waste a lot less time doing it
Of course 35/37 is behind it, but glibly quoting that figure doesn't help to understand why the problem fails. It is easier to understand by considering the individual game and the gearing of expected returns. Essentially the expected return is this 35/37, but you can gear it to make a highly probable small win against a highly improbable large win, i.e.
Expected Return = fixed
Expected Return = P(small win)*small win value + P(big loss)*big loss value
A well optimised system should offer a good balance here, i.e. the ratio of P(small win)/P(big loss) is directly related to (big loss value)/(small win value).
Thus you can make it highly likely (but not guaranteed) that you increase your initial stake by a small amount, which is the appeal of this system. We know that ultimately we may lose, but with sufficient "credit" we can make this infinitely improbable. (Hitchhikers guide anyone?)
However the failing is the requirement to achieve double or nothing, which is the principle of Saxo Boy's "sub-games". To achieve double or nothing, our (small win value)/(big loss value) is near one, which means the P(small win)/P(big loss) is near one, and with the tables "advantage" (that good ole 35/37 again) the odds will be stacked slightly against you.
As it happens, Saxo Boy's game is poorly optimised, and only successfully doubles his money around 10-20% of the time Basically you'd be better off sticking all your money on red and looking to double it that way. You'd still lose, but you'd waste a lot less time doing it
#250
Based on a handful of runs of my MATLAB script, yep. Actually the script is so noddy I could have trivially put it into an Excel macro. D'oh! Maybe I'll do that at some point (not going to get a chance now though... got to go out!) if you can wait until next week, I'll do a conversion.
By the way, the rule
total profit = (x * 35/37) - x
is based on a single application of a single strategy over an infinite period of time and doesn't teach us much about the "real world" sampling situation. Basically, it's true, but about as useful as a chocolate teapot
By the way, the rule
total profit = (x * 35/37) - x
is based on a single application of a single strategy over an infinite period of time and doesn't teach us much about the "real world" sampling situation. Basically, it's true, but about as useful as a chocolate teapot
#251
Scooby Regular
Thread Starter
Join Date: Mar 2002
Location: Bradford
Posts: 13,720
Likes: 0
Received 0 Likes
on
0 Posts
noddy - define please! I'm seeing a strange small guy with a red hat in my mind!
Sprint I'm trying to keep up but I'm struggling here The first problem with your model is you have based it on the 35/37 table when the European table is 36/37 Also, what system of 'banking' of winnings are you using because I have never accurately defined this simply because I'm not sure what would work best
Sprint I'm trying to keep up but I'm struggling here The first problem with your model is you have based it on the 35/37 table when the European table is 36/37 Also, what system of 'banking' of winnings are you using because I have never accurately defined this simply because I'm not sure what would work best
#254
No, 35/37 perhaps is a bit idealistic, but it definetely & absolutely defines the odds of the table.
I defer to your maths for now until I dig out my Uni maths books.....and work on this one as you have done.
I defer to your maths for now until I dig out my Uni maths books.....and work on this one as you have done.
#256
Originally Posted by Saxo Boy
Played for real money again and got £70 up - was willing to place the £30x2 bet (but no more) which I had to do twice - intense stuff
You said previously you would have *****, but you've almost lost them in your example. Your system DEFFO will not work if you crap out when the stakes are high.
#258
Scooby Regular
Join Date: Oct 2004
Location: London
Posts: 4,797
Likes: 0
Received 0 Likes
on
0 Posts
interesting website here which outlines quite a few roulette strategies, including the martingale, and also explains that they don't work.
interesting thing about the martingale that i hadn't realised before, is that in a sequence of losses, when you eventually get to the win you will still only win your initial bet value, despite the fact that if you have endured a significant sequence of consecutive losses, the bet will be very large. this makes it a very slow way to accumulate money, even ignoring the problem of the table limit. so with a £5 starting stake, by the seventh consecutive loss, you would be staking £320 just to get back your losses + £5. so the max you win in at any time is your smallest stake. i'm sure this is obvious to everyone else but it only just occurred to me.
in sb's sytem it would be even slower as you only get back your stake plus 2/3rds...
interesting thing about the martingale that i hadn't realised before, is that in a sequence of losses, when you eventually get to the win you will still only win your initial bet value, despite the fact that if you have endured a significant sequence of consecutive losses, the bet will be very large. this makes it a very slow way to accumulate money, even ignoring the problem of the table limit. so with a £5 starting stake, by the seventh consecutive loss, you would be staking £320 just to get back your losses + £5. so the max you win in at any time is your smallest stake. i'm sure this is obvious to everyone else but it only just occurred to me.
in sb's sytem it would be even slower as you only get back your stake plus 2/3rds...
Last edited by ProperCharlie; 13 February 2004 at 09:01 PM.
#259
Scooby Regular
Thread Starter
Join Date: Mar 2002
Location: Bradford
Posts: 13,720
Likes: 0
Received 0 Likes
on
0 Posts
Here is a way to look at it: There is only one way to loose and many ways to win. The absolute only way we can loose it to have a run of 5 losses in a row. If this occurs you will loose £810 exactly. If this doesn't happen you WILL either remain at zero or gain money. Because each spin is 64% likely to be a winner you win a lot more often than you remain at zero hence you make money. You could just bank each of your initial stake wins (£5 clear profit at a time) but this would take £810/5 = 162 wins to hit your target. Or you could take advantage of the fact that you statistically win more than anything else and reinvest you winnings looking for runs - which will be longer as well Exactly how to do this yet I'm not sure but I'm working on that side of the strategy. It took me about an hour of play to double £810 to £1620 so if you think making £810 in an hour is slow then I'm curious to know what you do for a living
#260
Originally Posted by Saxo Boy
Here is a way to look at it: There is only one way to loose and many ways to win. The absolute only way we can loose it to have a run of 5 losses in a row. If this occurs you will loose £810 exactly. If this doesn't happen you WILL either remain at zero or gain money. Because each spin is 64% likely to be a winner you win a lot more often than you remain at zero hence you make money. You could just bank each of your initial stake wins (£5 clear profit at a time) but this would take £810/5 = 162 wins to hit your target. Or you could take advantage of the fact that you statistically win more than anything else and reinvest you winnings looking for runs - which will be longer as well Exactly how to do this yet I'm not sure but I'm working on that side of the strategy. It took me about an hour of play to double £810 to £1620 so if you think making £810 in an hour is slow then I'm curious to know what you do for a living
#261
Originally Posted by Saxo Boy
reinvest you winnings looking for runs - which will be longer as well
I feel this thread is going round in circles now
#263
Scooby Regular
Thread Starter
Join Date: Mar 2002
Location: Bradford
Posts: 13,720
Likes: 0
Received 0 Likes
on
0 Posts
Statistically 15 loses in a row is far far far less likely to occur than 15 wins and we can ONLY lose £810!! Never more, always £810. Ok so you don't 'look' for runs, I mean we have to devise a strategy to try and ride them when they do come along which they will as we have our strategy in place to stop losses eating our bankroll. This is like playing Goldeneye on the N64 in golden gun mode with the opposition set to easy and dumb. In other words its one shot kills but its a total piece of **** to run around killing them. The only problem is can you make in through ALL of facility without one of them getting a lucky shot
The following is the spin record of play for fun which I'm now about £100 up on again. Please note that 'q' represents a loss (I just sat with my fingers on 'q' and 'w' on logged in notepad).
wwqwqwqwqwqwwqqqqwqwwqqwqwqwqwwwwwqwqqqwqwwqwqqwww qwwqwqwwwwqwqqwwqwqqwwwwqqqwwwqwqw
You'll note ther was a run of four early on which called upon the £270x2 bet There are 84 spins and 37 losses which is actually 44% i.e. considerably more than the 36% it should be yet I'm still up. This was a particularly hard segment to play and took much longer than usual to get £100 up. Nevertheless, based on this pattern I'd need 84x8 = 640 spins of the wheel to double up.
These are hard facts and also show that the dealer is actually getting the 'luck' at the moment. Also suggests that the play for fun module is NOT trying to help us - quite the opposite Also of note, had I been playing straight up bets I'd be miles down by now!
The following is the spin record of play for fun which I'm now about £100 up on again. Please note that 'q' represents a loss (I just sat with my fingers on 'q' and 'w' on logged in notepad).
wwqwqwqwqwqwwqqqqwqwwqqwqwqwqwwwwwqwqqqwqwwqwqqwww qwwqwqwwwwqwqqwwqwqqwwwwqqqwwwqwqw
You'll note ther was a run of four early on which called upon the £270x2 bet There are 84 spins and 37 losses which is actually 44% i.e. considerably more than the 36% it should be yet I'm still up. This was a particularly hard segment to play and took much longer than usual to get £100 up. Nevertheless, based on this pattern I'd need 84x8 = 640 spins of the wheel to double up.
These are hard facts and also show that the dealer is actually getting the 'luck' at the moment. Also suggests that the play for fun module is NOT trying to help us - quite the opposite Also of note, had I been playing straight up bets I'd be miles down by now!
Last edited by LG John; 13 February 2004 at 10:04 PM.
#264
Originally Posted by milo
page 14 and you've only just realised that?
I'll give him his dues for persistence, but 0/10 for listening
#265
Originally Posted by Saxo Boy
Statistically 15 loses in a row is far far far less likely to occur than 15 wins and we can ONLY lose £810!!
#266
Scooby Regular
Thread Starter
Join Date: Mar 2002
Location: Bradford
Posts: 13,720
Likes: 0
Received 0 Likes
on
0 Posts
SaxoBoy persists in thinking he has discovered something new that works
SaxoBoy persists in thinking he has discovered something new that works
#267
Originally Posted by Saxo Boy
Its the journey to discover eactly 'why' that stimulates and drives me. In fact this thread alone has been interesting and enjoyable enough to justify paying £810 for
I'll give you a quick example.
How did life begin
There. That should be another classic thread. Hehehehehe......
#268
Originally Posted by Saxo Boy
If you take £810 and decide, 'I will use this to pay my roulette system' and through losing it your game ends how can you lose more unless you don't have enough stength of character to stop. Its like putting £1 into a video arcade: you can't possibly lose more than that one pound unless you decide to play another game. The machnie can't pick-pocket you while you play
So, you've banked 3 x £810. If you lose the 3rd set of £810, you're left with £0, but 2 x £810 in the bank. As you said above, the game has now come to an end as you are at £0.
So, does this mean you never touch roulette again, or do you use the 2nd set of £810 and start again? If so, the game has not come to an end has it!?
#269
Originally Posted by Saxo Boy
I have NEVER said that it works. Its the process that I'm finding stimulating. Finding something, testing it, refining it, testing the theory, maths, logic, testing again, refining and more live testing. Its a learning curve and like I said back on page 1 I know that its 99.999% certain it won't work. Its the journey to discover eactly 'why' that stimulates and drives me. In fact this thread alone has been interesting and enjoyable enough to justify paying £810 for
What went wrong with poker? Given your goal in all of this research is to accumulate money, I thought poker was easier to make money on as you are taking money from "the weak"?
#270
Scooby Regular
Thread Starter
Join Date: Mar 2002
Location: Bradford
Posts: 13,720
Likes: 0
Received 0 Likes
on
0 Posts
No, re-read the concept of each £810 generating 2 other £810 before its banked. If BOTH those £810 get destroyed then that tear of your pyramid is destroyed so no you don't touch the original £810 (father)