This article has been edited, since the web page which was referring to is removed. In addition argument about validity of my test is not applicable since my test procedure for testing roulette computers now is widely accepted as valid and very useful. People who were arguing about validity and effectives of such test now instead of arguing trying to prove that their product would satisfy test requirements.  
Instead on focusing who said what and commenting it I will edit this article so it refers to general debate.

One of comments related to discussion was that every roulette wheel is slightly tilted. That is ok as well. But when we are testing how computer will react on different ball speeds we do not really care much is wheel slightly tilted or not. (Maybe if it is very strong tilt it would matter.)

Tilted wheel effects only last revolution where the ball with different speed stops at same dominant diamond because of tilt, other effects are really minimal.

We want to test a system as if a roulette wheel is 100% leveled. We do not care for result.
We do not care how after prediction the ball will be affected.
I will use very simple explanation and round numbers for better understanding.
Let's say this is data from perfectly leveled wheel,

The ball at 12 o'clock has a speed of 1.000sec per rotation.
After some traveling same ball at 3 o'clock has a speed of 1.050 sec. per rotation.

The ball at 6 o'clock has a speed of 1.100 sec. per rotation.

The ball at 9 o'clock has a speed of 1.150 sec.  per rotation.

And again at 12 it is 1.200 sec. per rotation.

So the ball is slowing down, not exactly by these figures but close to it.

I explained in my test why we can do clocking at different diamonds and what we achieve with that. That should be clear and undeniable. However some with their product denied it. This time it isn't ridicules as on first version of his comments, where he was explaining FF's accuracy as a result of linear calculation. It didn't have any logic, simply because test is based on real spin and on real wheel the ball deceleration is curved. Whoever had opportunity to test FF would know that FF takes few spins to adjust.

This time the test was challenged with explanation that the ball doesn't decelerate equally during the spin. It is 100% correct. Problem with his device is that he takes samples of full ball rotations and adds it in to his computer as variables without having idea what is happening in between. Her we need to take in consideration that clocked ball time is in 360 degrees intervals, his computer can not have idea of what is happening to the ball in between. If different spin comes he can only calculate by curve that his computer may develop from clocked ball times, as he demonstrated on his graph. However he can apply only one formula, which is constant, and if wheel isn't perfect there is no way that he can know how much is that imperfection, so his explanation is nonsense.

The FF works completely differently. It does not do setting as his computer, but it does it with every spin gradually. First few spins it observes spin very widely, next few spins it is narrower, after that it narrows even more but then starts learning what is going on in between full rotations. This process is extremely important if we want accurate prediction on leveled roulette wheel, because ball deceleration is not only different in between rotations but it changes during single rotation and it isn't constant all the time.

Therefore FF all the time is monitoring what is going on with the wheel. We simply can not base system setting on single spin or as he is doing few samples from spin and estimate remaining parameters. All other roulette computers are doing it on same way so no wonder why there is so many complaints. They will never succeed to make anything useful until they change it. When we tested such devices we claimed that they produce random result. It might not be 100% accurate claim, but considering that FF during such test predicts with accuracy of 2-3 pockets to be able in real play to get some advantage. If other device during the test was producing something as 1:20 it was obvious to us that such inaccuracy wouldn’t produce any noticeable advantage where additional parameters are added.

Due to human errors during first spin clocking when setting the system, one seller recommends to buyers to buy camera from him. The cheapest one is $800 and slightly better one is $2000.

It is really nothing more then a basic camera valued at $100. Where you record a spin, analyze it at home in slow motion, take timings and enter it manually in to the computer. (Until you do it who knows what can happen to the wheel). Funny is that seller claims how computer follows everything, even temperature or humidity changes, but I did not see any sensors on his device to be capable of detecting such changes. For humidity, barometer is needed and for temperature of course, temperature meter. If it is true, then why system does need timings from camera. It is only marketing strategy and deception of public beliefs. In usual roulette computer sellers are competing who has a bigger head, promising everything and delivering nothing or very little.
It’s now two years from my test and this article in original format was published, and latest is that they do not use camera any more, to get more accurate parameters. Anyway, it wasn’t never accurate enough, since camera taking picture frames in 40ms steps while average person will make errors in clocking about 30ms per rotation.  

The FF roulette computer doesn't need camera or anything. The FF does not make decision on single spin and single sample of users clocking. It monitors it all the time, analyzes it and stores a lot of data.
That may be different subject to discuss so let's go back to the point. If a wheel has some degree of tilt, our timing may be different. So it may look like this:

The ball at 12 o'clock has a speed of 1 rotation per 1.000 sec.

The ball at 3 o'clock has a speed of 1 rotation per 1.045 sec. ****

The ball at 6 o'clock has a speed of 1 rotation per sec 1.100sec.

The ball at 9 o'clock has a speed of 1 rotation per sec. 1.155sec ****

And again at 12 it is 1 rotation per 1.200sec.

Please note change in time next to ****

First one is because the ball is traveling down the hill so it decelerates slower.

Next one is when it decelerates up the hill. If we measure time with laser we would have that difference, because laser detects speed at particular spot, but when we do clocking we always do full rotation, so what ball is gaining when is going down the hill of tilted wheel it loses when it goes up the hill, (important) clocked time really stays close to same because one rotation includes everything therefore time difference from diamond to diamond would be same. Even if it doesn't, how much it can be different?

5 ms difference would produce error of less then 1 number. Who cares for one number? I am very happy when FF predicts within 5 numbers of accuracy on particular roulette wheel.

Therefore previous argument is invalid. The only reason for that was for such test great system accuracy is needed. That would include accuracy in timing, error correction and algorithms which could properly calculate ball position at any moment in relationship to position of numbers on the rotor and rotor speed at any moment which no other product could do to within required limits..

Due to slight tilt changes of ball speed per rotation are close to none, therefore attempt to use few milliseconds to adjust for possible effect of tilt would be disaster even if we define where on the wheel is tilt. For this problem even most sopshicated instruments wouldn't help simply because with roulette on 2 different spins after ball is exactly let's say 1.0 sec per rotation, next rotation can be 1.2s but on the other spin it could be different as 1.25s. That is (50ms) difference and it is huge if compared to possible few milliseconds created because of tilt.
Adjustment for tilt effect would require changing prediction by almost full rotation based on only few milliseconds difference in clocked data.

Nobody can have such accuracy, and even if he does because that few milliseconds are floating on top of already explained much higher possible time difference of (50ms) so it would be suicide if we try to adjust it. Does FF uses many clocking for accuracy it was stated, definitely not, I need at least 2 rotations. But not every rotation has same importance; FF decides which clocked rotations are more accurate based on all clocked data and previously stored parameters. FF uses multiple rotations to define parameters on the wheel, not only to reduce clocking errors.

Interesting point here is that only few months ago at roulettesystems.com (which now we know that belongs to computer seller) the FFZ was described as not workable system, now he is talking about system with high accuracy but as unpractical because it requires a lot of wheel watching. Player must have clear view of the wheel. What I should say on this: "maybe no it is not true, player can go and watch nice ladies, all he needs to do is to show FF to casino staff and tell them how long he was planning to play and they will transfer money to his bank account". There is so much that can be commented from his page but I think this would be more then enough. Even some people like to call me a competitor. I surely do not feel that way. I do not compete with anybody. If so, I would never provide so much information. Same as this procedure is valuable tool for someone who wants to test computer, it is also most needed tool for anybody trying to design workable roulette computer.
If you manage to get good results after a test, you are probably close to be successful.
Excuses as imperfections and deformations on the wheel are not good enough. Sure they exist and final result may be affected but effects on a test is very minimal because we do not look at final result. Especially if multiple diamonds test is used where there is not much time difference in between clocked ball speeds. Understanding importance of such testing and accepting it as a tool during design, half of work of creating workable roulette computer is done. By my opinion even someone who classifies himself as a competitor should be thankful for such test procedure. Maybe by publicly denying it, I am positive that they may start using it to try to solve problems with their product and prediction. Yes, it did open many questions and pressure from customers but all of them will be much happier once the product is brought to a level where it can pass this test.

Recently I have had opportunity to meet with one of person who at start did was arguing about my test. He told me that first he couldn’t understand it well, but now his product can do it, with better results.
 
So I am asking myself why I was troubled when I nicely asked him if I could come with 6 of his roulette phone computers that people wish to return. All over the net he published that there is no unhappy customers and that I am making all story just to discredit him.
Taking into consideration that he was manipulating them and they were from all over the world, I would collect computers from customers who were unhappy with his product. At that time he gave me an address but it was wrong address. Particular seller never accepted return of his computers and he never provided his real address. All he does is advertising and creating more and more pages, if you do a search you will find close to 100, all with different domain names. He is trying to look convincing with intention to find uniformed overseas buyers. His challenge simply doesn't exist, same as his so advertised official test by scientists, or his roulette computer conference. However same person in private conversation just recently told me to forget about his previous models of roulette computers, but now the product is much improved.
I can’t comment it since I know only about people who have his product and claim that they are unsatisfied.

< Prev