In this article, I’ll explain how to test my roulette computer. The same test could be applied to any roulette prediction device, computer or system. For this test, you would need at least one roulette video spin. If you can play video spin on a DVD player, it would be better then using computers DVD, because the computer may create time delays in the video picture, which could make the test harder or inaccurate. I am using a DVD player with built-in the hard drive where I have stored roulette spins. The DVD player is handy with an option to repeat a single spin continuously, and that is what we need for this particular test.
Four diamonds roulette computer test is a test to define how accurate the tested computer can predict according to measured known constant roulette wheel parameters.

If the tested computer is a smart device as the FF roulette computer; that learns wheel parameters from spins we clock, it is good practice firstly to predict few different spins from same roulette wheel so that the system can learn from it and tet tuned for a particular wheel. If we have only a single spin then is good practice; the first few spins to clock the ball each time at a different place. It’s good to give the system a chance to learn how the ball decelerates at various points during the spin before then we start with the test.

  • With the FF roulette computer first four valid spins is a fast adjustment, and probably after a first spin or two, you will be set close enough to have prediction within few numbers of accuracy. With each spin, the system settings are closer to match parameters of the particular wheel. After four spins the system is definitely within working range and starts looking and defining differences in between one single rotation to get the best accuracy when predicting. Also, the system starts various controls so already learned data can not be easily disturbed by accidental bad clocking. For each additional spin, the system is fine-tuning and monitoring what is going on with the wheel then adjusting according to it.

Roulette computer four diamonds test

Run one spin and try to predict it 40 times.

Every ten times use different diamond for clocking.

Clocking at diamond 1

Spin… Predicted number

1. 18 this is first spin and system learns, the prediction is number 18


3…26 here system is close to adjusted


5….32 stable prediction

6….32 it stays mostly within 3 numbers of accuracy





Clocking at diamond 2

Spin…Predicted number

11….12 It is same here, the system is seated with first 4 spins

12…..35; but here is fine-tuning for ball speeds at that range.

13….. 0








We can continue to test it at any other diamond or position on the wheel.

The good computer should detect ball speed correctly, eliminate human errors in clocking and predict according to measured ball speed. The narrower difference in between results means that the computer is more accurate. In this test, I actually used real data from testing FFZ’s IQE6 program.

Video spins move pictures in time frames of 40ms therefore at some positions error may slightly increase. Clocking at the real wheel should be more accurate, but we need video spins so we can have constant input parameters and we can compare output results. Ideally, all predicted spins would generate the same outcome.

At we have few members who did this test on some other roulette computers available on the market, none of the computers could produce reasonable results that would be worth spending any additional time for further testing.

Myrulet Four Diamonds Roulette Computer Test

Created debate all over internet and bulletin boards. Firstly some people tried to deny it and to call it invalid. Of course, such claims do not have even basic grounds since the test is very simple and logical. If we repeat the same spin, trying to predict it with roulette computer, it is obvious that if the computer is accurate that it should produce same or closes to the same results. The test is designed to help the user to get an idea of the quality of the device they have and to not be manipulated. However, there is some misunderstanding or perhaps deliberate manipulations of the test. The FFZ is very specific roulette computer. For prediction, it doesn’t need to know rotor position but it predicts from the point where the last ball rotation was clocked therefore this test is enough for such a device.

  • It has been two years since this test has been published and no other computer seller could design proper algorithms for levelled roulette wheel prediction, therefore, no further tests were required. Recently some are trying to sell roulette computers with tilted wheel algorithms claiming that it can predict and levelled roulette wheel. Following is an update on the test.

Roulette computers that have voice output (pronouncing numbers) from the moment we clock the rotor have to calculate rotor movement all the way from that point. To be able to predict particular number the computer has to know the rotor position at the time of prediction moment. The software can be manipulated so the computer actually can pass this test. So simple algorithms of tilted wheel prediction can be used with a preset wide time limit for the ball measurements and device may look as it can pass the test. Especially if rotor speed on video spin we use is slow.

If such tested computer passes four diamonds test further testing to define if the computer is working with tilted wheel algorithms or if it is capable of predicting and levelled wheel. For this part of the test, we continue with the same roulette video spin.

Tilted wheel algorithms explained

We clock rotor at 12  o’clock.  Rotor moves (computer has to calculate movement). At the moment when we get particular ball speed rotor moved let’s say one rotation. So the system predicts let’s say number zero.

We move the same test to 3 o’clock diamond. We do same, clock rotor, rotor moves, this time we are going to get prediction 1/4 of rotation later. For that time rotor moves only a few pockets extra and system still predicts the same number because with tilted wheel approach the system only calculates rotor movement which is not hard to calculate. Therefore the computer still may produce the same result.

Test if the computer works with tilted or levelled wheel algorithms

To differentiate if the computer, predicts as tilted or levelled wheel we need to do a bit more of testing.

After DIAMOND 1 test we still clock rotor at 12 o ‘clock but now we clock the ball at 6 o’clock position, the system has to predict the opposite result. If the computer predicts the same or close to same, the computer works with tilted wheel algorithms. (Tilted wheel prediction groups a wide range of ball speeds and produces the same result. For example, clocked ball time of 1000 ms per rotation may produce the same result as clocked 1200 ms/r.)
The computer needs to predict the opposite because the rotor is clocked at the same position, but we know that entered ball speed to the computer is by ½ rotations different. So if the computer is designed for levelled wheel and if it is accurate it must detect a change in ball speed, therefore it needs to predict differently.  (Until now only the FF roulette computer has such accuracy in ball timing to be able to do it)

Some roulette computer sellers claim that computer they sell calculates a function for all moments during roulette spin. In such case previously explained tests should produce the same results in any targeted ball rotation we decide to predict (for example (4,5,6,7,8,9 … rotations before then ball drops). If the tested computer can’t do it then, of course, you are entitled to a refund. Testing in different ball rotations should be done even if the computer works on tilted wheel principles.

The FFA roulette computer with IQE6 algorithms is the only computer designed to work on wheels without any dominant ball drop point. It all goes back to a time when some roulette computer sellers created a war in between each other arguing who has a bigger head. None of them did it properly, simply because they couldn’t. They do not use the right hardware capable of doing accurate time measurements. PDA’s and mobile phones programmed in Java are not timing devices.

Question someone may ask could be, but can such computer based on tilted wheel calculation predict and roulette levelled wheel or semi tilted wheel?

My answer is Yes; it can at least theoretically, if the computer is set on very tight limits for the ball timing and if the computer has reasonable algorithms for rotor calculation. In such a case, the computer may predict only abut 15% of spins. However such computer will not be able to predict at any time during the spin but it will predict only when the ball happens to be within the narrow ball speed limit. For example, a computer may be set to predict only if the ball is 1200 ms/r to 1230 ms/r (30ms time frame). It means that the operator will need to clock the ball until it comes to that particular speed range. In such case, many spins will not be predicted. In usual time difference in between rotations is about 200ms, therefore 30 ms preset time frame we will get only 30/200, 15% of spins predicted.
For spins that we get predicted we know that the ball is within very narrow speed range and most likely it will hit the same diamond which may produce an advantage.
With such an approach even if the ball is clocked with double errors, we still may get similar results, but of course, that is far from reality and needed accuracy for proper roulette levelled wheel prediction. Narrowing limits like that and getting 15 out of 100 spins predicted and claiming that the computer is capable of predicting each time when the ball will hit selected diamond is pointless. Especially when the test is done on semi tilted wheel where the particular diamond is getting more hits (meaning even 100ms wide ball speed range may still hit it). However, if the computer can do it accurately, and predict at any time during the spins that may be something. For example, we clock the balls 3-4 rotations and run 100 spins. We get accurately predicted 15 spins then we repeat the same test (without any adjustments) but this time we clock balls 3-4 rotations earlier in spin and still get same results and same spins predicted. Until now nobody could demonstrate it.

Also with such approach player also needs to experiment with time limits.
For example time window of 1200 ms/r to 1230 ms/r may on particular wheel catch ball speeds where the ball may be hitting high positions on the diamond, and some even pass it and go for another diamond. In such a case, the time limit must be blindly readjusted, and the player needs to repeat the same process.

The main problem with such approach is that from 100 played spins we get only 15 spins predicted. It may take us 5 hours to get a hundred spins if we need to readjust the time frame it may take additional 5 hours just to see if it is adjusted properly.
If parameters on the wheel change (as in usual) then it may take us additional 5 hours just to find out that we do not get accurate predictions.

(Here I would like to highlight that manually adjusting time frame even for tilted wheel prediction is not the very practical solution. The FF roulette computer with TILT2 algorithms uses automated time frame set up based on measured parameters. It saves a lot of time, and predictions are more accurate.)

From everything explained above it is clear that such design of roulette computer and claim that it can predict and levelled wheel is the more theoretical approach, an unpractical solution from someone who has no clue how to design proper algorithms for levelled wheel prediction. The FF roulette computer with IQE6 algorithms predicts every reasonable clocked spin because it has the hardware for proper time measurements and the program is designed for levelled wheel prediction.

The test how good computer can eliminate errors in clocking

For this test alternate clocking position, first rotation clock the ball at the diamond and next rotation do it 1-2 cm distanced. If you do it with a single click, it will produce an error of about nine numbers. But error correction should reduce it. Compare results with previous data and check if there is the difference and how much. IQE 6 is getting almost same data during this test. But also important it that if entered errors are too large the computer will protect the player from wrong prediction, and it will call clocking error.

Roulette rotor calculation tests

Still, we can use same video spin. Up to now you already should know what predicted the outcome is. Next is to check how much that prediction will change by giving to system different rotor time. So repeat spin clocking but when clocking rotor start when the ball is at clocking diamond but stop it 5 -10 cm earlier or later. You give to computer data that rotor is faster / slower. Of course, prediction should move according to different rotor speed clocked.

If your last click is earlier, then the system should predict further in ball direction. This is an only approximate test, but the change should be close enough.

FFZ uses detected ball and rotor speed when displaying result; therefore it will not show a 100% accurate result. A simulation that wheel is faster will show a bit less than real rotor change, and for the slower wheel, it would show more.

With tests explained we can have a general idea about system ability to define failing point of the ball. Of course, that is not a final prediction.

How accurate is this failing point in relationship to final result it depends on many additional factors such as:

How far in time are we from the point when the ball hits the rotor.

Imperfections on the wheel, deformations on ball track, the slight tilt may have various effects on ball travelling time, kind of ball, number of diamonds etc. With later prediction, we are more accurate, but we have less time to place chips.

IQE6 tested on different spins with high accuracy setting have all results within same or close to the same distance from prediction.

Roulette ball bouncing

This part is individual for every wheel/ball combination.

Because of that no matter how good we predict the game is still partly random.

Sometimes it works for us sometimes against, and we should look for an average.

The task of every advanced player is to have a good way of prediction but also to search where that advantage may be the highest.

For the end, if you have the opportunity to look computer in action as sellers demonstration. Never look final result it could be coincidence or video editing.

Simply compare distances from predicted number to number where the ball hits the rotor.


If the computer is good, and prediction is zero, then the ball hits rotor at number 18.

Next spin, whatever prediction is the ball should hit same distance from that number as it did from zero to number 18. Sure few spins could be out of that, but most of them should have the same distance.