Exchange Games
Welcome to Live View – Take the tour to learn more
Start Tour
There is currently 1 person viewing this thread.
08 Nov 09
08:50
Joined:
Date Joined:
30 Jul 09
Probability what a word, it explains everything and tells us nothing.
Probability states that somewhere in the universe there is a roulette wheel that only ever spins reds. There is also one that only ever spins blacks.
Probability also tells us that there is a chance we will find one of them, however it also states there is a 50/50 chance that we will bet red forever on the black wheel or black forever on the red wheel.
Probability tells us that however well backtested a stategy is one day it will lose, ignore this fact at your peril. The one thing that kills all strategies & systems is the dreaded ANOMALY which is unpredictable.
So the final word is if you find a good strategy by backtesting sling it in the bin because the anomaly is waiting to pounce.
Probability states that somewhere in the universe there is a roulette wheel that only ever spins reds. There is also one that only ever spins blacks.
Probability also tells us that there is a chance we will find one of them, however it also states there is a 50/50 chance that we will bet red forever on the black wheel or black forever on the red wheel.
Probability tells us that however well backtested a stategy is one day it will lose, ignore this fact at your peril. The one thing that kills all strategies & systems is the dreaded ANOMALY which is unpredictable.
So the final word is if you find a good strategy by backtesting sling it in the bin because the anomaly is waiting to pounce.
Show More
Why would that be necessarily true if you have successfully backtested a flat bet strategy over a large enough sample ?
A big IF I know.
But I do question whether folks on here really tell the whole truth if and when they claim that their strategy is a flat bet one, with no deviations from this betting rule at all.
If you backtest such a strategy cpmprehensively enough and obtain the required varaiance from the norm ( say 3 to 4 standard deviations) to beat the bad value odds compounded by the commission off take, then the laws of probability would say the opposite in fact, that is that it should continue to win in the future.
A big IF I know.
But I do question whether folks on here really tell the whole truth if and when they claim that their strategy is a flat bet one, with no deviations from this betting rule at all.
If you backtest such a strategy cpmprehensively enough and obtain the required varaiance from the norm ( say 3 to 4 standard deviations) to beat the bad value odds compounded by the commission off take, then the laws of probability would say the opposite in fact, that is that it should continue to win in the future.
Not really the point I'm making.
I'm talking about playing a flat bet system over a huge number of games, such as 500,000 and I'm saying that statistically the results will be consistently the same, whether they are past games or future games.
I'm talking about playing a flat bet system over a huge number of games, such as 500,000 and I'm saying that statistically the results will be consistently the same, whether they are past games or future games.
Are you just being deliberatively argumentative or do you honestly not get the point I'm making ?
Do you agree for example that if you backtest a sample of ( say ) 100,000 past games of even chance bets that the result will be 50/50 wiihin a small percentage of a standard deviation, and that the results of a future 100,000 games will be statistically the same.
Surely that is indisputable even by you.
Individual immediate past results bear no reflection on immediate future results. Nobody disagrees with you on that.
But if you have a system that generates its results slowly over a large body of games and does not chase past results in any form or manner then, if that system has backtested a satisfactory and meaningful variance from the norm, then the probability is positie that it will continue to do so in the future on games to come.
This is not supposition, it is fact.
The key question is whether anybody has been able to find and successfully backtest any such system or bet selection strategy ( probably a better phrase).
Jazza I know that you are somewhat depressed by the games and your apparent failre to crack them. If it helps you're not alone in that.
Sorry if any of my above post appears to be personally rude or aggressive to you.
Not meant to be at all.
Do you agree for example that if you backtest a sample of ( say ) 100,000 past games of even chance bets that the result will be 50/50 wiihin a small percentage of a standard deviation, and that the results of a future 100,000 games will be statistically the same.
Surely that is indisputable even by you.
Individual immediate past results bear no reflection on immediate future results. Nobody disagrees with you on that.
But if you have a system that generates its results slowly over a large body of games and does not chase past results in any form or manner then, if that system has backtested a satisfactory and meaningful variance from the norm, then the probability is positie that it will continue to do so in the future on games to come.
This is not supposition, it is fact.
The key question is whether anybody has been able to find and successfully backtest any such system or bet selection strategy ( probably a better phrase).
Jazza I know that you are somewhat depressed by the games and your apparent failre to crack them. If it helps you're not alone in that.
Sorry if any of my above post appears to be personally rude or aggressive to you.
Not meant to be at all.
Or if you prefer, how do you backtest? I usually just test realtime paper trading. Think you've misunderstood me.
The dichotomy of past results not affecting future results and past results must revert to the mean is a good debating point.
Reason says that any result can happen in the future.
Statistics says that odds must revert to the mean, eventually.
So, can the future be predicted from past results?
If past results are found to be far away from the mean, do they have to come back?
Any comments?
Reason says that any result can happen in the future.
Statistics says that odds must revert to the mean, eventually.
So, can the future be predicted from past results?
If past results are found to be far away from the mean, do they have to come back?
Any comments?
Yes everything will revert to the mean with a small running variance.
The problem is that it moee often than not reverts slowly and unevenly, that is not in a straight line, and the combination of bad value odds and commission gets you in the end.
That is it will revert but not always, or even mostly, in a way to capitalize profitably on unfortunately.
This is not using past results to predict futre results.
It's just saying that over any reasonable sample of games, past and/or future, a statistical norm will be achieved always.
The problem is that it moee often than not reverts slowly and unevenly, that is not in a straight line, and the combination of bad value odds and commission gets you in the end.
That is it will revert but not always, or even mostly, in a way to capitalize profitably on unfortunately.
This is not using past results to predict futre results.
It's just saying that over any reasonable sample of games, past and/or future, a statistical norm will be achieved always.
Sure - but I should preface my thoughts by saying that I am not an idiot and am just throwing some ideas around. If you are also not an idiot, you should not go and follow this advice. hah.
Martingale is brilliant. If you took a bottomless bank to a roulette table with no limits, you would never lose. Yes, there are runs of huge streaks, eg 15, 18, 20 reds/blacks or odds/evens or high/lows in a row, but there is no roulette table in the world that has never spun a black (etc).
Yes, these things come in eventually, so mean reversion is at work. Taking this away from Ex-games a bit I know, but roulette is an easy to understand mathematical example. I've got some software that works using this in that instead of charting the last few numbers in sequence it charts the accumulative totals. For example, if the results went:
R R B R R R B R R R B R R B R R R R B B B
(R=red, B=black)
Some gamblers would say "bet red, it hasn't come up in three consecutive spins", whereas this software would be recommending putting money onto black, because for the average to be c50/50 we are 'owed' some more black results.
However, this doesn't answer the question as to whether betfair exgames **, because it wouldn't be at all difficult to get back to random averages through a winding and longwinded path that just happened to go against every bet you ever made...
Martingale is brilliant. If you took a bottomless bank to a roulette table with no limits, you would never lose. Yes, there are runs of huge streaks, eg 15, 18, 20 reds/blacks or odds/evens or high/lows in a row, but there is no roulette table in the world that has never spun a black (etc).
Yes, these things come in eventually, so mean reversion is at work. Taking this away from Ex-games a bit I know, but roulette is an easy to understand mathematical example. I've got some software that works using this in that instead of charting the last few numbers in sequence it charts the accumulative totals. For example, if the results went:
R R B R R R B R R R B R R B R R R R B B B
(R=red, B=black)
Some gamblers would say "bet red, it hasn't come up in three consecutive spins", whereas this software would be recommending putting money onto black, because for the average to be c50/50 we are 'owed' some more black results.
However, this doesn't answer the question as to whether betfair exgames **, because it wouldn't be at all difficult to get back to random averages through a winding and longwinded path that just happened to go against every bet you ever made...
FAFH,
With a reasonable sample of future games a statistical norm will always be acheived.
That sounds like a prediction to me.
So if a sample of 1000 coin flips landed 45% Heads would you back Tails until 50% was achieved?
Would this be a good bet?
What if it was 10000 flips and 40%?
Would this be a better bet?
With a reasonable sample of future games a statistical norm will always be acheived.
That sounds like a prediction to me.
So if a sample of 1000 coin flips landed 45% Heads would you back Tails until 50% was achieved?
Would this be a good bet?
What if it was 10000 flips and 40%?
Would this be a better bet?
No Ivy I don't think what you suggest would necessarily be a good bet.
As I said above it could revert to the mean in a number of ways, some very slow and gradual and some conversely fast and direct..
It can for example effectively revert to a normal variance from the mean by simply going 50/50 for a very extended future time.
One cannot predict anything in this micro sense.
All one can say is that any current statistically excessive variance from the norm will revert to a normal statistical variance over time.
Not anything you can use to develop a betting strategy with I'm afraid to say.
Not a sure fire one that is.
As I said above it could revert to the mean in a number of ways, some very slow and gradual and some conversely fast and direct..
It can for example effectively revert to a normal variance from the mean by simply going 50/50 for a very extended future time.
One cannot predict anything in this micro sense.
All one can say is that any current statistically excessive variance from the norm will revert to a normal statistical variance over time.
Not anything you can use to develop a betting strategy with I'm afraid to say.
Not a sure fire one that is.
Martingale does work if you have a bottomless bank with no table limits, however would anyone with a bottomless bank bother gambling in the first place.
People insist on thinking that the only reasons why martingale doesn't work is because they haven't got a big enough bank and because of table limits in the casino.
But the martingale system doesn't work whatever way you look at it!
Tell me what ROI you would have on your martingale system betting if you had a "bottomless" bank! Not much left when you divide something with infinity now is there?
How about if you are Bill Gates then? Well if you want to put 1 000 000 000 USD on a 1.000000001 shot then go ahead, but don't think it's a value bet. Actually the odds is much worse than that because of the green zero.
I think the martingale is then to blame for people fruitlessly trying to find some other staking system as they think it should be able to overcome the "problems" martingale suffers from in terms of casino table limits etc.
Then there's this misunderstood mean reversion thing. If red has been overrepresented in the last 1000 spins and have a lead of 100 spins against black, then that difference might as well increase as decrease for the next 1000, 10000, 100000 spins etc.
Percentage wise however red and black will over time get closer and closer to the same win ratio. But that's not the same thing! It's the sheer volume of spins that make red and black converge percentage wise. In absolute terms red and black can be quite far from each other but still have converged percentage wise.
For further reading I also suggest you google gambler's fallacy which many of you seem to suffer from.
But the martingale system doesn't work whatever way you look at it!
Tell me what ROI you would have on your martingale system betting if you had a "bottomless" bank! Not much left when you divide something with infinity now is there?
How about if you are Bill Gates then? Well if you want to put 1 000 000 000 USD on a 1.000000001 shot then go ahead, but don't think it's a value bet. Actually the odds is much worse than that because of the green zero.
I think the martingale is then to blame for people fruitlessly trying to find some other staking system as they think it should be able to overcome the "problems" martingale suffers from in terms of casino table limits etc.
Then there's this misunderstood mean reversion thing. If red has been overrepresented in the last 1000 spins and have a lead of 100 spins against black, then that difference might as well increase as decrease for the next 1000, 10000, 100000 spins etc.
Percentage wise however red and black will over time get closer and closer to the same win ratio. But that's not the same thing! It's the sheer volume of spins that make red and black converge percentage wise. In absolute terms red and black can be quite far from each other but still have converged percentage wise.
For further reading I also suggest you google gambler's fallacy which many of you seem to suffer from.
People insist on thinking that the only reasons why martingale doesn't work is because they haven't got a big enough bank and because of table limits in the casino.
But the martingale system doesn't work whatever way you look at it!
Tell me what ROI you would have on your martingale system betting if you had a "bottomless" bank! Not much left when you divide something with infinity now is there?
How about if you are Bill Gates then? Well if you want to put 1 000 000 000 USD on a 1.000000001 shot then go ahead, but don't think it's a value bet. Actually the odds is much worse than that because of the green zero.
I think the martingale is then to blame for people fruitlessly trying to find some other staking system as they think it should be able to overcome the "problems" martingale suffers from in terms of casino table limits etc.
Then there's this misunderstood mean reversion thing. If red has been overrepresented in the last 1000 spins and have a lead of 100 spins against black, then that difference might as well increase as decrease for the next 1000, 10000, 100000 spins etc.
Percentage wise however red and black will over time get closer and closer to the same win ratio. But that's not the same thing! It's the sheer volume of spins that make red and black converge percentage wise. In absolute terms red and black can be quite far from each other but still have converged percentage wise.
For further reading I also suggest you google gambler's fallacy which many of you seem to suffer from.
But the martingale system doesn't work whatever way you look at it!
Tell me what ROI you would have on your martingale system betting if you had a "bottomless" bank! Not much left when you divide something with infinity now is there?
How about if you are Bill Gates then? Well if you want to put 1 000 000 000 USD on a 1.000000001 shot then go ahead, but don't think it's a value bet. Actually the odds is much worse than that because of the green zero.
I think the martingale is then to blame for people fruitlessly trying to find some other staking system as they think it should be able to overcome the "problems" martingale suffers from in terms of casino table limits etc.
Then there's this misunderstood mean reversion thing. If red has been overrepresented in the last 1000 spins and have a lead of 100 spins against black, then that difference might as well increase as decrease for the next 1000, 10000, 100000 spins etc.
Percentage wise however red and black will over time get closer and closer to the same win ratio. But that's not the same thing! It's the sheer volume of spins that make red and black converge percentage wise. In absolute terms red and black can be quite far from each other but still have converged percentage wise.
For further reading I also suggest you google gambler's fallacy which many of you seem to suffer from.
I don't think people here are SUFFERING from anything. Just a discussion mate, no need to take it all so seriously...
Post Your Reply
Please login to post a reply.
Wonder
Instance ID: 13539