Euromillions
- Thread starter PapaLazaru
- Start date
As you say, if those casinos who pay ITV1 a hefty amount (those programs last for hours), were being killed by sophisticated computer programs, then eventually, they would run out of money to pay ITV1 and would go bust.
Why would you need camera angles?
Surely the number which the ball lands on, is all that is required?
Or are we suggesting that these programs analyse the way that the ball bounces before landing on a number?
Soldato
- Joined
- 8 Mar 2007
- Posts
- 10,938
Studying past results is useless, roulette table don't have a brain or a memory. It's not like it can think "Oh 32 hasn't come up for a while, I better bring it in now", the trouble is people confuse the general rule that over a long enough period all numbers will come up roughly the same amount of times and think that because X hasn't come up for a while or is under-represented it must be more likely to come up soon given the rule I've just stated (this is called the Gambler's Fallacy). The reality is every number has exactly the same chance of coming up on any given spin regardless of past results.
There is no programme that can analyse a spin and accurately predict where the ball will land using physics. There are a lot of people that claim they can do it or invent some sophisticated computer that could do it but no one has yet, if they did Roulette would have been dropped.
Watch these two spins....
See those bumps around the edge? How the hell can you calculate the exact angle and speed it will hit one at? Any any fraction difference in speed or direction at point of impact can change where the ball will land dramatically, from one side of the wheel to the other. Notice also on the second spin how the ball is about to land then violently jumps out and falls in another section of the table.
Anyone that claims they can calculate all those variable with any accuracy, and all in the space of a few seconds (the croupier stops taking bets shortly after spinning the ball) is lying and a charlatan.
But online casinos make thousands out of people that do believe they can do it.
Don't get me started on 'dice control' in Craps!
wizardofodds.com is a great site that looks at odds and myths like these.
And this is the crux of what allows Casinos to make money.
Player says, "I have a system".
Casino says, "Would you care to try out your system?"
Player plays the game and more often than not, loses money.
I'd be interested in knowing how the program which gave a 147% return managed to achieve this. Was it luck? Was it calculating the trajectory of the ball? Was it using past results to predict future results? What metrics was this program using? And how many spins were used during the test?
Getting back to the subject of syndicates. I looked up online regarding the use of these. However, it would appear that syndicate websites take a large percentage/cut. It would seem that it would be better to run a syndicate website than buy syndicate lottery tickets.
There are only a handful of different models of roulette tables used in casinos, the company who were running this competition had there own table that was used in casinos around the country.
To use the program, the croupier would set the ball spinning first, and then the program would take a number of images, using these images it would calculate the trajectory of the ball, including where it would hit different parts of the table designed to make the results more random. The bumps and deflectors placed around the table are the hardest part of the modelling process. In the example I was referring to, if the ball was projected to hit a 'horizontal' deflector the program returned no prediction, and the user wouldn't bet.
Once the program had calculated where it thought the ball would land, it would return list of numbers around the calculated result where the user would bet their money.
The program of course did need to be calibrated, the model works flawlessly in ideal conditions, but each roulette table would be placed at ever so slightly different angles, and the ball and table surfaces would be of different states of ware. Calibration is a fairly simple process where the program records the first 20 or so spins, notes the position, and changes a couple of the predefined variables in the model.
The key difference in the online casino games which offer live video roulette, is that you cannot place your bet after the ball has started on its trajectory. In a real casino there is a time window between when the croupier starts the ball spinning, and when no more bets is called.
The program only needs to be accurate to 1/4 of the table for a user of it to make money.
Of course if you were to start recording roulette games on your smartphone in a casino, you would quickly be 'asked to leave'
To use the program, the croupier would set the ball spinning first, and then the program would take a number of images, using these images it would calculate the trajectory of the ball, including where it would hit different parts of the table designed to make the results more random. The bumps and deflectors placed around the table are the hardest part of the modelling process. In the example I was referring to, if the ball was projected to hit a 'horizontal' deflector the program returned no prediction, and the user wouldn't bet.
Once the program had calculated where it thought the ball would land, it would return list of numbers around the calculated result where the user would bet their money.
The program of course did need to be calibrated, the model works flawlessly in ideal conditions, but each roulette table would be placed at ever so slightly different angles, and the ball and table surfaces would be of different states of ware. Calibration is a fairly simple process where the program records the first 20 or so spins, notes the position, and changes a couple of the predefined variables in the model.
The key difference in the online casino games which offer live video roulette, is that you cannot place your bet after the ball has started on its trajectory. In a real casino there is a time window between when the croupier starts the ball spinning, and when no more bets is called.
The program only needs to be accurate to 1/4 of the table for a user of it to make money.
Of course if you were to start recording roulette games on your smartphone in a casino, you would quickly be 'asked to leave'
Soldato
- Joined
- 8 Mar 2007
- Posts
- 10,938
Yes you can, I've got account with various bookmakers who offer this service and you can place your bet after the ball has been spun.
Look I just nipped into one and took a screenshot...
As you can see she has spun the ball and in the top right there is still 3 seconds of time left to bet (there was actually 5 seconds when she released it).
Yes you can, I've got account with various bookmakers who offer this service and you can place your bet after the ball has been spun.
Look I just nipped into one and took a screenshot...
As you can see she has spun the ball and in the top right there is still 3 seconds of time left to bet (there was actually 5 seconds when she released it).
Hmmn, we were told they didn't exist. The program I've got on my hard drive works from a top-down perspective, it might be worth trying to alter it to work at that angle... What resolution can you get the image of the table at?
Excuse me while I read the NDA I signed
Soldato
- Joined
- 8 Mar 2007
- Posts
- 10,938
You can hit the 'expand' button and get it like this....
After watching a few spins, she launches it with 3 seconds remaining on average (and LOL at the comment in the window).
After watching a few spins, she launches it with 3 seconds remaining on average (and LOL at the comment in the window).
Soldato
- Joined
- 8 Mar 2007
- Posts
- 10,938
You're talking about a biased wheel, this can happen but is very unlikely nowadays with the calibration and and regular checks casinos do. If it was obvious a good croupier would spot it straight away and close the table, if it's not obvious then you'd have to watch 1000's of spins before you saw any correlation that was above and beyond standard deviation.
From the Wizard of odds himself...
Soldato
- Joined
- 8 Mar 2007
- Posts
- 10,938
I play Craps most weekends up at the G in Manchester and i can honestly say that Dice Control is a MYTH!
I cannot believe the amount of Tube videos dedicated to it too!
Quite, I will concede that if you don't hit the end wall (with the jagged surface) it would be theoretically possible to practice so you could throw in a way where certain numbers were ever so slightly more likely to come up than others but no casino in the world would let you throw dice in an obviously controlled manner without hitting that end wall (I don't play craps but I believe they can even make you roll again if it doesn't).
Man of Honour
- Joined
- 21 Nov 2004
- Posts
- 47,660
Think I am going to get a lucky dip 
If you continue to make throws which do not hit the back wall then, yes they can make you throw again.
In Vegas they also will ask you to throw the dice again if the throw is over eye level. In the G Manchester, this is allowed. Oh and just in case Dice Control is possible, Vegas casinos do not allow you to "set the dice", again over here, they allow it!
The odds are better over here as well
Back on Topic
I'm not sure on the correct way to word this so bare with me... I know I'm probably wrong but removing an element of chance should improve your odds?
For example...
If you have 2 hats with 1-49 in, and you randomly pull one number out of each, would you not remove the element of randomness on the second hat if you actually chose a number from 1-49?
For example...
If you have 2 hats with 1-49 in, and you randomly pull one number out of each, would you not remove the element of randomness on the second hat if you actually chose a number from 1-49?
Soldato
- Joined
- 8 Mar 2007
- Posts
- 10,938
Sorry, I really don't understand the question. can you re-word it.
Soldato
- Joined
- 23 May 2005
- Posts
- 2,964
- Location
- Auckland, New Zealand
Taking wheel bias out of the equation, Judge is not saying that it's possible to predict where the ball is going to land using variables guessed by a tracking device. I.e, you can't go "the ball is travelling at this speed so it's likely to drop off here and fit this bump and it's projected bounce will be..." etc. But you CAN make guesses about where the ball might land, based on some statistical model.
The resolution of those guesses might be pretty huge. i.e you might only be able to say that the ball will land in a certain quarter with a 50% accuracy.. but that then tells you it's probably not a great idea to place a bet on the opposite quarter.
I'd assume one would need a pretty sophisticated set up to make decent guesses, though. I'd imagine a high speed camera would be needed from a top-down view. So those programs would probably not produce anything usable from a 25fps picture off a tv.
The resolution of those guesses might be pretty huge. i.e you might only be able to say that the ball will land in a certain quarter with a 50% accuracy.. but that then tells you it's probably not a great idea to place a bet on the opposite quarter.
I'd assume one would need a pretty sophisticated set up to make decent guesses, though. I'd imagine a high speed camera would be needed from a top-down view. So those programs would probably not produce anything usable from a 25fps picture off a tv.