MMG Martingale script
- Thread starter zarqon
- Start date
A little. I can tell you my experience.
I was over 18 mill in the hole, now I am less than 3 mill in the hole. There are a couple of caveats.
1) You have to know you could lose it all at any time. If you look at a series, the odds are low, but they still exist.
2) The largest bet I've had to make to date is 10 million meat. By the time you make a 10 million meat bet, you've already lost 9,990,000 Knowing you are only making 100 meat a shot, that 20 million meat wager was hard to make. You gotta be deep and you gotta be committed. If you tuck tail and go home, it would take a couple of weeks to make back what you lost. If you press ahead, you could lose it all.
To fund a chain of 10 (1k+2K+4K+8K+16K+32K+64K+128K+256K+512K) is 1.023 million meat. Your odds of losing all 10 (you would only need 1 win anywhere in the chain) is 1/2^10, or .1% (one out of 1000). When I had to make the 10 million meat bet, it was because I had lost 13 in a row (I use 2.02 as a multiplier to counteract the house grab). The odds are on your side, but the chance to lose still exists.
I was over 18 mill in the hole, now I am less than 3 mill in the hole. There are a couple of caveats.
1) You have to know you could lose it all at any time. If you look at a series, the odds are low, but they still exist.
2) The largest bet I've had to make to date is 10 million meat. By the time you make a 10 million meat bet, you've already lost 9,990,000 Knowing you are only making 100 meat a shot, that 20 million meat wager was hard to make. You gotta be deep and you gotta be committed. If you tuck tail and go home, it would take a couple of weeks to make back what you lost. If you press ahead, you could lose it all.
To fund a chain of 10 (1k+2K+4K+8K+16K+32K+64K+128K+256K+512K) is 1.023 million meat. Your odds of losing all 10 (you would only need 1 win anywhere in the chain) is 1/2^10, or .1% (one out of 1000). When I had to make the 10 million meat bet, it was because I had lost 13 in a row (I use 2.02 as a multiplier to counteract the house grab). The odds are on your side, but the chance to lose still exists.
@Z: I understand probability pretty well, so I'm not asking because I'm wondering what a random walk distribution looks like. What I'm curious about is folks who play all the time. Are they deluding themselves about the outcomes? bad at math? Rich enough that they don't mind the meatsink? How do they cope with their losses - farm until they can re-invest? Etc. It's a psych question, not a math one. 
It's true. If we flipped a coin 10 times, the odds of you winning at least one of the 10 are pretty good. That's how that works.
Although if you were doubling your bet on every coinflip and you DID lose 10 in a row, it would be even the more painful.
@Z: I understand probability pretty well, so I'm not asking because I'm wondering what a random walk distribution looks like. What I'm curious about is folks who play all the time. Are they deluding themselves about the outcomes? bad at math? Rich enough that they don't mind the meatsink? How do they cope with their losses - farm until they can re-invest? Etc. It's a psych question, not a math one.
Well, you probably wouldn't want their win loss record. You would want their psyche evals.
zarqon
Well-known member
This script is currently sitting at 66 registered users. I too am curious about the draw given that, in theory, the risk is greater than the chance of winning.
I say in theory. Because what I'm more curious about is what role the imperfection of computer-generated random numbers plays in the probability of win/loss, and spreading the 50% chance thinner (as Martingale does) allows a greater sampling from which to possibly draw conclusions (to exploit of course). If I were to pick either 0 or 1 randomly, and kept picking 0, at some point we would say the random number algorithm was broken.
I would guess the MMG uses PHP's rand(), possibly seeded with srand(). I would also note the user discussions there about "random" and seeding, which often mention the idea of making things "more random" i.e. fewer repeated numbers and less predictable. In this case, the idea of making something "more random" works in Martingale's favor. What I'm saying is that while the game is, in theory, a 50% win situation no matter how you slice it, RNG's are not exactly perfect probability machines due to the human element in the perception of randomness, so there may actually be some small merit to using Martingale if you intend to play the MMG at all. The only question is if this imperfection is enough to outweigh the strategy's net loss. I doubt it, but would still be curious to see a larger data sample.
EDIT: hahahahahaha
I say in theory. Because what I'm more curious about is what role the imperfection of computer-generated random numbers plays in the probability of win/loss, and spreading the 50% chance thinner (as Martingale does) allows a greater sampling from which to possibly draw conclusions (to exploit of course). If I were to pick either 0 or 1 randomly, and kept picking 0, at some point we would say the random number algorithm was broken.
I would guess the MMG uses PHP's rand(), possibly seeded with srand(). I would also note the user discussions there about "random" and seeding, which often mention the idea of making things "more random" i.e. fewer repeated numbers and less predictable. In this case, the idea of making something "more random" works in Martingale's favor. What I'm saying is that while the game is, in theory, a 50% win situation no matter how you slice it, RNG's are not exactly perfect probability machines due to the human element in the perception of randomness, so there may actually be some small merit to using Martingale if you intend to play the MMG at all. The only question is if this imperfection is enough to outweigh the strategy's net loss. I doubt it, but would still be curious to see a larger data sample.
EDIT: hahahahahaha
Last edited:
halfvoid
Member
Hi. I have zero experience in editing ASH script but wanted to know if someone could help me alter the code to use a different betting scheme.
I bet in increments of 1k, 2k, 6k, 18k, 54k, 162k, 486k, 1458k
I start off with 5 bets of 1k, then for however many lose, i bet that many of the next increment, until all have been won, then i start over again at 1k.
i.e.
bet 1k 1k 1k 1k 1k
lose 3
bet 2k 2k 2k
lose 1
bet 6k
lose 1
bet 18k
win
(repeat)bet 1k 1k 1k 1k 1k (and so on...)
I bet in increments of 1k, 2k, 6k, 18k, 54k, 162k, 486k, 1458k
I start off with 5 bets of 1k, then for however many lose, i bet that many of the next increment, until all have been won, then i start over again at 1k.
i.e.
bet 1k 1k 1k 1k 1k
lose 3
bet 2k 2k 2k
lose 1
bet 6k
lose 1
bet 18k
win
(repeat)bet 1k 1k 1k 1k 1k (and so on...)
Maybe, maybe not. Picking 0 100 times in a row wouldn't in and of itself mean the generator was broken, as that is statistically possible for a functioning random number generator.
Here is where the falacy comes in.
If we take the house cut out of the occasion:
On a 1 chain, you have a 50% chance of winning 1000 and a 50% chance of losing 1000. Run it over enough trials to bear out the stats your net gain is 0.
On a 2 chain, you have a 75% chance of winning 1000 and a 25% chance of losing 3000. Run it over enough trials to bear out the stats your net gain is 0.
On a 3 chain, you have an 87.5% chance of winning 1000 and a 12.5% chance of losing 7000. Run it over enough trials to bear out the stats your net gain is 0.
And so on. Even though you are incresing the chances you are going to win a chain, you aren't increasing your chances that you are going to win any money, because when you lose you lose so big. The law of large numbers says that you are going to even out over time anyway you slice it.
When you put the house cut into it, you are going to lose more (or win less, as the case may be) for every step you go over step one, so in the long run, you won't break even. You will lose.
You can alter the situation to slide wining a chain in your favor, but in the long run, the odds of wining money are the same.
Last edited:
@Z: I'm not quite sure if you're joking or not - any respectable RNG will show an appropriate distribution of outcomes; the problem comes from if they're truly random versus psuedorandom (e.g. there is an underlying pattern to the even distribution that could be discerned). If you'd like proof of this, no need to bang around with mafia; just run php rand() a few billion times. I've done this, incidentally, not with php rand() but with the libc implementation (which may well be what php uses). I can tell you for sure*, it's 50/50. And I can tell you for double-sure** that if I'm wrong on the previous point, the margin of error is much less than you'd see from considering every instance of the MMG played to date.
So the thing I'm interested in is - when they crash, how hard, and what do they do about it?
EDIT: confirmed, php rand() does use libc rand().
* sure = within one part in a trillion or so...
** guestimating - to one part within a 10^18
So the thing I'm interested in is - when they crash, how hard, and what do they do about it?
EDIT: confirmed, php rand() does use libc rand().
* sure = within one part in a trillion or so...
** guestimating - to one part within a 10^18
Last edited:
zarqon
Well-known member
@halfvoid: A script very similar to what you are requesting has already been discussed and worked out on the preceding pages, with plenty of code samples. You might take a look at that and see if that gets you on the right track. Then, scrape something together, and if it doesn't work post it and ask why. People here are always happy to help fix existing code.
@dj_d: dang, I should have made this script do some data-gathering as it went, sending data to some central place. A repository of small wins and magnificent losses!
@dj_d: dang, I should have made this script do some data-gathering as it went, sending data to some central place. A repository of small wins and magnificent losses!
halfvoid
Member
The best i've come up with so far is a single bet version of my betting scheme. Just by removing the other 4 bets and setting the wins required to 1. works for me so far, but i don't know how to loop it to only bet again for the lost bets. Hehe. Still says betting five times when it only bets one too.