edge on roulette
You scan day by day and find a way to play that gives 6% edge in average
This play has won each of the 20 500-trial days from 1 to 10% of the total money wagged
The trick is that we found it out after taking the whole data.
This way of play is flatbetting on 15 numbers, but, each of them have got different chip value
Math question:
1)what is the chance of a way to play that is suposed to lose 2.7% of the tottal wagged have got +6% edge in 10k?
2)What is the chance to get +1 to 10% edge in the next 500 set?
3)How much positive fluctuation is here? 1 to 2 standard deviations?
4)What if we succeeded in the next 10 sets of 500 trials with similar performance?
warm regards
ybot
Quote: ybot
This way of play is flatbetting on 15 numbers, but, each of them have got different chip value
The chip value is irrelevant. You won't
win playing this, you will eventually
hit a losing streak that will wipe out
all your wins. 10K trials is nothing, run
a sim for 100K spins and you'll find the
real results.
Quote: ybot1)what is the chance of a way to play that is suposed to lose 2.7% of the tottal wagged have got +6% edge in 10k?
2)What is the chance to get +1 to 10% edge in the next 500 set?
3)How much positive fluctuation is here? 1 to 2 standard deviations?
4)What if we succeeded in the next 10 sets of 500 trials with similar performance?
1) it's definitely under 1%, starting to get close to zero
2) if everything is exactly the same as the first 10 000 spins, should be at least 50% (though that would even include the chance of doing better than +10%)
3) more than two standard deviations
4) YOU BE RICH
The way to calculate this event is beyond my resources.
What I meant when I stated that numbers had different chip value is that even being a flatbet bet, I asigned a chip value to each of the 15 numbers and kept the same bet for the whole 10k.
In this math questions I want to know the probability of facts I witness, we, then, might discuss wether I win or lose
10k playing the same way isn`t too much, but it has its weight, same numbers winning the entire number string.
Fluctuations are brave enemies, to know how brave they could be, we should find out a true/final edge in case it exists.
Quote: ybotEvenbob, I am an AP, we know basic and a little more complex probability conceps
The way to calculate this event is beyond my resources.
What I meant when I stated that numbers had different chip value is that even being a flatbet bet, I asigned a chip value to each of the 15 numbers and kept the same bet for the whole 10k.
In this math questions I want to know the probability of facts I witness, we, then, might discuss wether I win or lose
10k playing the same way isn`t too much, but it has its weight, same numbers winning the entire number string.
Fluctuations are brave enemies, to know how brave they could be, we should find out a true/final edge in case it exists.
so you are an AP huh.
interesting.
what area do you AP in?
now back to your problem. i think what your problem is you got 10k of data AND THEN you find a pattern. What you should do is invent a system first and then do a 10k trial.
Maybe that will help.
1)very dificult for random numbers to perform +6% edge , never losing a 500 run, Negative expectation starts to fade. Might we especalute in a positive expectation below 6%?
2)Did you mean that there is 50% chance to this way of play to get from 1 to 15% edge in the next 500 spins?
How did the fact that this play performed well affect a 50% success prediction?
What is the chance for any other 15 numbers to get from 1 to 15% edge in the next test?
3)in case we face over 2 sd at 10k, we might predict a result closer than predicting from random priors
4)I could be rich, yes
Does it change anything wether there were 20 sets of 500 than 30 sets of 500?
Quote: ybot
What I meant when I stated that numbers had different chip value is that even being a flatbet bet, I asigned a chip value to each of the 15 numbers and kept the same bet for the whole 10k.
.
What makes you think the 15 numbers you
chose will win more often that the rest of
the numbers in every session.
Quote: andysifso you are an AP huh.
interesting.
what area do you AP in?
now back to your problem. i think what your problem is you got 10k of data AND THEN you find a pattern. What you should do is invent a system first and then do a 10k trial.
Maybe that will help.
Hi andysif, we know this way to analize data can curvefit info.
As data is perecible, we must work with the updated and fresh
To get 10k takes a time, let alone another 10k, in the elapsed time many conditions may vary
In case we invented a system first, we needed up to 2sd to confirm a positive expectation, then, we must study a lot to determine the accurate advantage.
Quote: EvenBobWhat makes you think the 15 numbers you
chose will win more often that the rest of
the numbers in every session.
I scanned data after being taken , found a way to succeed in the whole 20 day of 500 spins
I think nothing, I obey to probability, the only way to know what to do or not
I try to be the most objective and study it with probability tools, I lack of strong knowledge about as an engeneer or an expert in probability has got.
Quote: ybotI scanned data after being taken ,
What does this even mean?
Quote: EvenBobWhat does this even mean?
he recorded the data, THEN he looked for pattern
Quote: andysifhe recorded the data, THEN he looked for pattern
Why? What good is it after the fact.
Quote: EvenBobWhy? What good is it after the fact.
What if the wheel is not balanced correctly?
ZCore13
Get another sample of 10K spins. Using those spins, see how your number picking / wagering scheme works out.
Here's a quick way to get started: The first 10K spins of data you have, figure out the frequency of each number hit. If the frequencies are very close to their expected hit rate, you're probably just going down some rabbit hole and you're not going to get any good data. But IF the frequencies of number hits differs a good amount from expectation, then you're looking at a biased wheel.
Good luck, but I wouldn't waste much time or spend much (or any) money on this endeavor unless you truly believe there is an advantage to be had. But I suspect there is none.
Quote: Zcore13What if the wheel is not balanced correctly?
ZCore13
You know how long it takes to record
10K spins on one wheel? Months and
months.
Quote: EvenBobWhy? What good is it after the fact.
It takes months to store 10k trials, we must work with it first
Quote: RSFigure out an algorithm that will suit your "number picking" scheme and wagers on each of those numbers.
Get another sample of 10K spins. Using those spins, see how your number picking / wagering scheme works out.
Here's a quick way to get started: The first 10K spins of data you have, figure out the frequency of each number hit. If the frequencies are very close to their expected hit rate, you're probably just going down some rabbit hole and you're not going to get any good data. But IF the frequencies of number hits differs a good amount from expectation, then you're looking at a biased wheel.
Good luck, but I wouldn't waste much time or spend much (or any) money on this endeavor unless you truly believe there is an advantage to be had. But I suspect there is none.
Rs, number frequencies differ , none of the 15 numbers have got the same hit ratio
Quote: ybot1)very dificult for random numbers to perform +6% edge , never losing a 500 run, Negative expectation starts to fade. Might we especalute in a positive expectation below 6%?
Divide profit by total amount risked to get your expected edge. After 10,000 spins there shouldn't be much of a reason to adjust that number
Quote: ybot2)Did you mean that there is 50% chance to this way of play to get from 1 to 15% edge in the next 500 spins?
Yes. Could even be well over 50% depending on exactly how you are getting your edge
Quote: ybotHow did the fact that this play performed well affect a 50% success prediction?
It was by far the most important factor
Quote: ybotWhat is the chance for any other 15 numbers to get from 1 to 15% edge in the next test?
With a 6% player edge, very high. With a 2.7% house edge very low
How do you calculate 50% chance to get from 1 to 10% edge in the next 500 trials?
Quote: ybotNo way that the same +6% is the true edge
Then just bet half what Kelly Criterion would otherwise say.
Or do a better estimating your edge
Quote: EvenBobWhat does this even mean?
He "shot the arrows and then drew the target" - in other words, he got the numbers first, then looked for patterns in that particular set of numbers.
Any system can "work" if you create it after the fact.
Quote: ybotRs, number frequencies differ , none of the 15 numbers have got the same hit ratio
How much do they differ from expectation? Find the one with the highest hit frequency -- if you were to flat bet that number, over that 10k trial, would you have come out ahead?
Quote: ThatDonGuyHe "shot the arrows and then drew the target" - in other words, he got the numbers first, then looked for patterns in that particular set of numbers.
Any system can "work" if you create it after the fact.
Yes, this is what I actually did
But, there must a thereshold
We know +6% edge is unreal, any edge probably, o none
In this math section I try to debank the way to calculate it
Quote: ybotRs, number frequencies differ , none of the 15 numbers have got the same hit ratio
How much do they differ from expectation? Find the one with the highest hit frequency -- if you were to flat bet that number, over that 10k trial, would you have come out ahead?
Quote: RSHow much do they differ from expectation? Find the one with the highest hit frequency -- if you were to flat bet that number, over that 10k trial, would you have come out ahead?
At 10k, analizing single numbers, you need at least 3sd to consider that its updated edge would lower to a more realistic frequency. Random fluctuations use to fool you, even to pro players. Single number study takes over 20k
Yes, there are few numbers with 2 to 3 sd
For players who have never played with an edge on roulette you must know that +6% on 1 number is not the same as +6% on 2 numbers or 15, each scenario yield different profits
