Roulette Computer Comparison Chart

Forester's FFZ and FFA unlimited

Choice of the following:

1. Virtually any late model cellular phone: modified to ensure suitability of hardware. Practical, functional and includes the required features to beat any modern roulette wheel, including those without physical imperfections.

2. Any late model PDA: Not as discrete

3. PIC Microcontrollers: not recommended though as this hardware has too many limitations that ultimately reduce accuracy. It is only suitable for physically flawed wheels. We have two different PIC models. Contact us for details.

4. Image recognition hardware and software: aim a camera and predictions are made automatically. See Hybrid Roulette Computer.

Psion 3a PDA: Hardware is suitable for required functionality, although PDAs are far too large.

Mobile Phone: We were sent his phone version for free by an unhappy purchaser who believes the computer is a scam (photo shown left). I can only describe this device as the clumsiest roulette computer I've ever seen - it is actually worse than his previous computer and I can confidently say it indeed is a scam. The original purchaser who sent it to us for free believed the same, but Howe refused to give a refund. There is no chance it earned Howe almost US$8M as he claimed - Howe is well known for his outrageous false claims, both about his competitors and his own products.

NOTE: Johansen's NOK device is actually Mark Howe attempting to sell his device under fabricated names. Mr. Howe is extremely dishonest and manipulative. He is also the publisher of a review site roulettecomputerreviews.com where he claims to be "Johansen" - he gives top reviews only to his computers, and other computers that don't exist to add credibility to his site. See further information about Mark Howe:

Mark Howe's main identity | Johansen (fabricated)

This device has been repeatedly publicly tested on a level and unbiased wheel. Each time, the results were no better than random. Shown below is a photo one of the demonstrations of this device.

NOTE: The only significant difference is the FFA version gives audio predictions, whereas the FFZ version gives electric zaps as explained below.

PIC Microcontroller: very basic functionality. In some respects, it is similar to our PIC models. This hardware cannot support a design that is sufficiently sophisticated to beat modern roulette wheels.

This device has been repeatedly publicly tested. Participants unanimously agreed the device was simply not practical or possible to use in the real casino environment. Shown left is a photo one of the demonstrations of this device. The only way to use it is with slow motion video and a top view of the wheel, which is simply not possible in real casino conditions.

The device has also been independently tested by a casino consultant (Mike Barnett) under ideal conditions (with slow motion video). He found the device produced virtually random predictions on modern wheels that are not flawed (95% of wheels). Of course the device developer attempted to refute the test results, although the predictions were virtually random because of inadequate design. Some of it's issues are explained on this page.

Despite the hype created by the seller, this device is extremely simplistic, and simply not capable of beating modern roulette wheels that are not heavily flawed. With this device, you can produce exceptional results at home in front of your DVD player. But when you take the device to a real casino, you quickly find it is rather useless. This is why many people end up selling the device. If you ever wish to purchase this device, we suggest buy it secondhand as they are often sold at half price. But expect to end up selling it to someone else.

Learn more about this device and comparisons with ours

If you have this device and want to compare it with ours, see http://www.rouletteforum.net/cgi-bin/forum/Blah.pl?m-1262064653/

NOTE: You can clearly see our device has the highest concentration of points in the smallest area. This somewhat shatters absolute nonsense spread by envious developers of other devices who claim mobile phones are incapable of processing accurate timings. We will soon publish a video of such testing.

This testing only confirms which computer is most accurate for the FIRST part of the prediction process. The testing does NOT confirm overall accuracy, although testing of overall accuracy even more clearly demonstrates that our device is far superior.

Learn more about comparative testing with ours and other devices

FFV: N/A (very basic design. no menu or audio predictions)

FFA: English

Image recognition hardware & software (Hybrid Roulette Computer): simply point a video camera at the wheel for predictions. This method is not subject to human error. The technology is patent pending and not available elsewhere. NOTE: hit rates stated below are relevant for the standard computer only. Hit rates are higher for the Hybrid.

With the exception of microcontroller versions, all our computers predict which diamond will be hit, the part of the diamond that will be hit (accurate to within 1/3rd of a diamond), and the manner in which the ball will fall.

In other words, if the ball clips a horizontal diamond then lands on the other side of the wheel, the computer predicts this. Or if the ball misses all diamonds completely, the computer will adjust for the different type of ball fall.

No other computer even remotely has such capabilities.

Assumes the ball will fall exactly the same way irrespective of whether the ball hits a vertical or horizontal diamond, or if the ball doesn't hit any diamond at all.

This computer does not correctly predict where the ball will fall from the ball track in the first place, nor is is capable of dealing with the errors from manual timings (clicks of a button). It is therefore not even remotely capable of correctly predicting which diamond the ball will hit, and certainly not which part of the diamond will be hit.

Assumes the ball will fall exactly the same way irrespective of whether the ball hits a vertical or horizontal diamond, or if the ball doesn't hit any diamond at all.

This computer does not correctly predict where the ball will fall from the ball track in the first place. It is therefore not capable of correctly predicting which diamond the ball will hit, and certainly not which part of the diamond will be hit.

The hardware used for this device is neither capable of supporting the algorithms, or storing the data required to make such a feature possible.

FFZ: 1 in 28. This is under "ideal" conditions where video footage is slowed to obtain the prediction. In real casino conditions, application is almost impossible due to the method of prediction, so not even this accuracy rate is realistically achievable.

FFA: 1 in 28. The FFA version uses virtually identical algorithms as the FFZ - the only significant difference is the FFA "speaks" the prediction.

1 in 8

FFZ: 1 in 16. This is under "ideal" conditions where video footage is slowed to obtain the prediction. In real casino conditions, application is almost impossible due to the method of prediction, so not even this accuracy rate is realistically achievable.

FFA: 1 in 16. The FFA version uses virtually identical algorithms as the FFZ - the only significant difference is the FFA "speaks" the prediction.

The green area shows the area for 95% of wheels. Accuracy increases as wheel tilt increases. The computer deals with level wheels, heavily tilted wheels, and semi-tilted wheels. Approximately 95% of wheels are semi-tilted

The green area indicates 95% of modern wheels, where accuracy is minimal. This is due to a fatal design flaw, where the computer incorrectly assumes the wheel is either:

1. Physically perfect (which it never is), or

2. Heavily flawed (which it very rarely is).

In reality, 95% of modern wheels have slight imperfections, but these subtle flaws make an enormous difference to where the ball will actually fall.

The green area indicates 95% of modern wheels, where accuracy is minimal. This is due to a fatal design flaw, where the computer incorrectly assumes the wheel is either:

1. Physically perfect (which it never is), or

2. Heavily flawed (which it very rarely is).

In reality, 95% of modern wheels have slight imperfections, but these subtle flaws make an enormous difference to where the ball will actually fall.

YES, and both the wheel clocker and bettor hear all predictions and computer audio prompts. This ensures both players are synchronized.

FFZ: NO. Only one player can use this device, which means you must look at the wheel to take timings yourself, then make your bets quickly, and this is not even considering the severe impracticality of the device. The design makes it blatantly obvious to casino staff that you are a device user.

FFA: The FFA version of this computer IS wireless capable, but has critical problems with real-casino application.

FFZ: N/A (no wireless)

FFA: The wireless transmissions are very easily detected - the device emits a constant radio emission as if you had a two-way radio left on. Furthermore, network scanners used by the casino instantly reveal the presence of the device, not to mention the range is extremely short, and subject to audio distortion to the point where predictions are often completely unheard by the second player. The developer appears to either given no thought to real-casino application, or has taken shortcuts to produce a cheaper device. You basically pay to be detected, which is beyond a poor product - it is virtually criminal.

NO, you have no way to validate accuracy of samples until you have tested the device and potentially depleted your bankroll.

This vendor claims to use polynomials, although tests we've conducted indicate the claims are likely false.

YES, it takes 1 spin to set up the ball deceleration model. Play mode then begins, and the computer automatically "learns the wheel" during play. Accuracy improves the more you play.

PARTIALLY. This computer does have features that can be used for "quick resuming". However, it is simpy not accurate enough to beat modern wheels. Even on ideal wheels in perfect conditions, this computer can be 2-3 full revolutions off-target for predicting when the ball will fall. This degree of error is enormous, which is partially why it only beats heavily flawed (tilted) wheels. When it comes to modern and relatively level (unbiased) wheels, given that the winning number can be on the complete opposite side to the prediction if the computer is incorrect by 1cm, this device can not even remotely beat modern unbiased wheels.

YES, tuning is done automatically while you play. Accuracy is confirmed by audible beeps at the precise moment when the ball is predicted to fall.

The variation of the ball's deceleration rate is automatically learned.

FFA & FFZ: NO. It does part of this process correctly (about 15% of it), but fails to do other essential components. Additionally, it at least has a feature that informs you when deceleration rates have changed so you know when to take new samples. But this is only "adequate" for tilted wheels, not modern level and semi-tilted wheels where air pressure variations make it impossible to determine which diamond will be hit. This alone can make accuracy no better than random for this device.

YES. The feature can be toggled ON/OFF. When it is ON, accuracy can be maximized. When it is OFF, predictions can be obtained at any point in the spin to ensure you receive a prediction at the latest point in the spin possible.

Speakerphone (audio): excellent for practicing, OR

Vibration mode (covert): Vibrations can be used many different ways defined by the user. With the vibration feature, you can set the computer to vibrate when the prediction is above the winning number (as Forester's FFV is used), although this is no different to his FFZ, and very impractical). So if you use the vibration feature, it can be used in a variety of much more practical ways. The user can select the desired option and switch between them instantly

Speakerphone (audio): excellent for practicing

FFZ: After taking clicks to establish timings, the device will give you a painful electric zap. When you get the zap, you see what number is under the fast moving ball. This is almost impossible to with any kind of accuracy, but it's not even your predicted number yet. To get your actual prediction, you must then manually and visually rotate the initial number by the required angle, and see what number is at that angle on the wheel. You do this for every spin which is almost impossible to in the comfort of home let alone real casino conditions. This is because the ball is way too fast to see the numbers accurately, and in real casino conditions you cannot possible see the wheel from a top view. Click HERE to download video demo of this device to better understand why it is impractical. In this video, the computer was set to give predictions about 3 seconds before the ball falls (very, very late - to make it very easy). Even with these remarkably easy conditions, the following results were achieved:

Out of 31 spins, this device failed to give any prediction 10 times, and predictions were not 14 times. The seller of this device claims his device's method of prediction (zaps) is best, but you can see for yourself how ridiculous it is.

FFA: Single number only. However, the unlimited version, after the initial prediction (for tilted wheels only), you can obtain a new prediction each time you click the button (clocking switch) as the ball passes the reference diamond. The seller claims this to be a strong selling point, but in reality is is nonsense. Why? Because when you use the computer, you do not stand by constantly clicking a button to obtain the next number to bet on - in reality, you get your prediction as quickly as possible and place your bets before no more bets is called.

FFA AND FFZ: N/A (no vibration), although this seller has a vibration version which replaces the FFZ's electric zaps with vibrations.

Single player version: Two wheel rotor clicks (either full, half revolution, or two revolutions), and either 2,3,4,5 or 6 ball clicks. The order of which is clocked (timed) first can be set by the user (either ball or wheel rotor first).

Two player version: Two wheel rotor clicks (either full, half revolution, or two revolutions), and either 2,3,4,5 or 6 ball clicks. The order of which is clocked (timed) first can be set by the user (either ball or wheel rotor first) OR both players can take timings of the wheel at the same time. This feature enables players to achieve accurate predictions in as little as 1 second after ball release. No other computer has this capability.

First 2 clicks for the wheel, then continual clicking for the ball until a prediction is given.

FFZ: NO, you need a top view of the wheel for this device to be possibly viable. This is because you aren't actually told where to bet - you only get an electric zap. At the time of the zap, you observe the number under the ball. This alone is nearly impossible, but then you need to mentally add/subtract angles to determine the final predicted number. This is simply not a viable or realistic process.

FFA: YES, you only need to see the ball and zero pass.

It is no different to using linear (straight-line graphs).

This mathematics can only work for heavily tilted wheels - NOT for level wheels. This is because the maths determines only ball drop time, but it does not tell you accurately WHERE the ball will drop.

The developer has done well with limited resources, but limited resources means cutting corners. As a result, the device can only be applied successfully on older wheels, and even then it doesn't achieve the highest accuracy possible.

FFA: Very good to Excellent: If you attempted to predict the same spin repeatedly on DVD, predictions are almost all within a 6 pocket arc, which are NOT the stated results on the developer's website. In our tests we found this to be odd, considering the hardware he uses is actually the most suitable for timing applications (but not by a significant amount). However, it takes much more than accurate timings to create a truly effective roulette computer. We believe the results of related testing were poorer than our computer because it uses less sophisticated algorithms, and does not learn the wheel parameters as well as it could.

FFZ: Excellent. However, it does not make any calculators for the rotor, so there is fewer errors possible. In real casino conditions, if you were able to predict the same spin repeatedly, there is no chance you would achieve obtaining the same prediction within a 3-6 pocket arc. Why? Because to achieve this, when you received the electric zap, you would need to note the number under the ball when it's travelling fast, with the wheel rotor moving in the opposite direction. Then you would need to manually see which number is X pockets away from the initial number, and that's assuming you could even see the ball at the time the zap occurs.

FFZ: 1

FFA: 2, although as explained above, the wireless / 2-player setup is extremely poor.

Single player version: 1 burst. The only wireless signal ever sent is an extremely short burst of data that tells the 2nd player where to bet. There are no emissions at all during the timing process.

Two player version: 1-2 bursts

5 bursts. This is 5 times easier to detect than a single burst. There is one radio emission for every click of the button when you take timings.

FFZ: n/a (no wireless)

FFZ: Rather than one device transmitting the prediction to another via a short burst, the wireless setup is no different to a two-way radio being set to constantly transmit. So there is one long and continous radio burst, which is no different to a radio becon that alerts the casino surveilance staff to the presence of the device.

FFZ: N/A (no wireless)

FFA: Yes, but the actual radio emission is incredibly easy to detect. There is no frequency hopping or anything to reduce the potential of the device being detected.

YES. Important points for consideration:

• The 100 spins are divided into two segments of 50 spins to validate the results. The results are:

i. First 50 Spins: 1 in 16 exact number hit rate

ii. Second 50 Spins: 1 in 16 exact number hit rate

• Actual tuned predictions are shown, not misleading "raw predictions".

• The footage is continuous and uncut which can be verified by the continuous movement of the lava lamp in the background.

• The computer's screen is shown displaying the prediction so you know predictions are authentic, and not merely audio added later.

• The computer's scatter chart is displayed so you can confirm groupings of high bars are in the predicted area.

• The wheel does not have any significant common ball drop point. It is neither perfectly level or heavily tilted - such a condition is most difficult to predict. This is because the computer can neither assume the wheel is mechanically perfect, or heavily flawed.

• Virtually none of the computer's additional features that increase accuracy are used. Such features include but are not limited to:

i. Automatic adjustment of predictions based on wheel rotor speed for each spin.

ii. Risk announcement so the player knows when a particular prediction is likely to be less accurate than others.

iii. Only 5 clicks are used to establish the ball timings. Up to 6 ball clicks can be used to increase accuracy.

• Predictions are made 4 - 7 seconds before the ball falls. The computer is capable of accurate predictions with 10+ seconds remaining until the ball falls, depending on ball track deformations that contribute to erratic ball behavior.

• The wheel is not stopped. There is no common release point of the ball. Even if there was, it wouldn't matter because different diamonds are used as timing reference points between spins.

• Teflon 3/4" ball is used. This ball produces a more random scatter than heavier ivorine balls.

• The wheel is disassembled with photos at genuinewinner.com to confirm no magnets or tricks are used, although such trickery would actually be more complicated than the development of a genuinely effective roulette computer.

If all features are enabled, accuracy is greatly increased. However, to prevent trade secrets from leaking to casinos, most features in this video demo were disabled.

NO. The only video provided has the following credibility issues:

• The demo is conducted over only 60 spins which is insufficient, especially when the footage ends abruptly with no comment. The vendor could have easily taken the best 60 spins and cut the footage short where the results were still favorable, instead of doing the demo over a set number of spins.

• The footage is badly corrupted and the wheel becomes invisible about half way through the footage. This is extremely suspicious, and would make it extremely easy to cut and join two favorable sets of 30 spins.

• A heavy ball is used with minimal bounce.

• There is absolutely no continuous movement in the footage's background. This would make cutting and editing footage much easier.

• Most notably, at the end of the video, a scatter chart is shown which is extremely misleading. This is because after any amount of spins, a "peak" will inevitably form because there will never be a perfect spread of values. It is much like how after 100 spins, there will almost never be exactly 50 red and 50 black numbers. the vendor claims that his computer's edge is where the peak has appeared. The problem with this is especially over a low number of spins, because there is never a perfect spread of values, a peak will inevitably form. What the vendor appears to have done is merely claim this inevitable peak is where his computer has the advantage. To an amateur, this video may be convincing proof. To someone that knows better, the video shows many signs of manipulation.

Various video demonstrations available, although they are only on heavily flawed wheels where the ball tends to hit the same diamond on almost every spin, with the ball barely bouncing. This could be referred to as a "heavily tilted wheel with virtually no scatter", and such wheels are extremely easy to predict.

To the inexperienced player, the accuracy in the video demonstrations may appear impressive. However, when people use the device in real casino conditions and on modern wheels, it is very common for them to end up selling the device for as little as half price.

FFA: NO. The sound volume and radio emission from the induction loop (the part that transmits audio to the wireless earpiece) is set far too high, and cannot be changed. This means anyone with a hearing aid as much as 2 feet away from you can hear what you hear. This presents a serious risk of detection.

FFZ: n/a (no audio)

FFA: YES

FFZ: n/a

FFA: YES

FFZ: n/a

FFZ: NO, plus the actual mechanism that tells you where to bet is a high voltage electric zap - the worst option possible. This is extremely easy for casino surveillance equipment to detect.

FFA: NO

YES, although it is not needed. It is only used if a roulette wheel spins very infrequently (perhaps once every 4+ minutes) so the user can setup fully after just one spin.

Our computer does NOT need this capability because the wheel is learnt automatically, but it is better to be with than without this feature.

FFA & FFZ: NO. Even IF the wheel was learnt correctly (which it isn't), in casinos where the wheel spins very infrequently, the player would need to waste approximately 20 minutes just to learn the ball deceleration, although it is actually constantly changing. And this is without consideration to the computer learning the ball behavior once it actually falls and hits the rotor/pockets.

YES, you can see EXACT output including predicted rotations. This helps you confirm you are getting what you paid for.

You can also use the 4 diamond test (as shown in video 2 on the videos page), although this particular feature is more for the player to verify the accuracy of the mathematics.

Visual: This is where you look at the wheel each spin and "visually" determine the number of pockets from the "raw" prediction to the "tuned" prediction. This is only recommended for experienced players.

Manual with in-built charts: the computer builds a scatter chart for you that you can observe anytime. You observe the charts and rotate predictions to tune manually. The tuned predictions are rotated a set number of pockets. If the wheel is outside a speed range you set, the computer notifies you.

Advance Automatic Tuning: using chi-squared and standard deviation statistical analysis, the computer will automatically tune the computer for you as you play. If the wheel is outside a speed range you set, the computer notifies you.

Dynamic Tuning: The greater the wheel speed, the further the ball will come to rest from the point of "raw prediction". The "dynamic" feature will automatically tune for each spin based on the speed of the wheel when the ball is predicted to fall. For example, if the wheel timing is 4000MS, then the ball may come to rest +5 pockets from the "raw prediction". If the wheel speed is 3000MS (faster), the ball may come to rest at +15 pockets from the raw prediction. This is a 10 pocket difference. Without this feature, ordinary computers would give predictions as if the wheel speed were the same every time, which it's not. Without this feature, even if a computer perfectly predicted where the ball will actually strike the wheel rotor, the difference in wheel speed between spins may make other computers almost completely ineffective.

Manual: you have to check the scatter graph then manually tune based on your own analysis. If the wheel speed varies during play, you receive the prediction anyway whether it will be accurate or not. The mobile phone version does NOT have in-built charts, so you need to draw your own on paper by noting the prediction you receive, the winning number, and the distance in pockets between them. This is extremely poor as to any attentive casino staff, what you are doing is obvious.

If the wheel speed varies, you receive the prediction regardless. You cannot restrict acceptable wheel speeds, and it does not have anything close to "dynamic" tuning (adjusting predictions based on wheel speed). This contributes to enormous inaccuracies when wheel speeds vary between spins, which they inevitably do.

FFA & FFZ:

Visual: This is where you look at the wheel each spin and "visually" determine the number of pockets from the "raw" prediction to the "tuned" prediction. This is only recommended for experienced players.

Manual (no in-built charts): Neither versions have in-built charts, so you need to draw your own on paper by noting the prediction you receive, the winning number, and the distance in pockets between them. This is extremely poor as to any attentive casino staff, what you are doing is obvious.

If the wheel speed varies, you receive the prediction regardless. You cannot restrict acceptable wheel speeds, and it does not have anything close to "dynamic" tuning (adjusting predictions based on wheel speed). This contributes to enormous inaccuracies when wheel speeds vary between spins, which they inevitably do.

FFZ: PARTIALLY. When timings don't meet certain specifications, no prediction is given - the player is not notified though. This occurs often on over 70% of spins which is extremely frustrating, especially when the device does not reliably reset to play mode. After no prediction is given, often the device appears to crash and needs to be reset. This is what many others, including an independent casino consultant has found (Mike Barnett). The unreliability of the device alone makes it extremely impractical.

FFA: PARTIALLY. When timings don't meet certain specifications, no prediction is given and "error" is announced. However, unless the computer is setup perfectly, which is not so easy, errors are extremely common. For example, if your initial setup spin has too many clicks, the device appears to be unable to cope due to its limited memory, and "error" is announced. Our devices have no such unrealistic and sensitive limitations.

PDA version: NO, the computer has one central scatter chart for each direction and makes no adjustments based on wheel rotor speed.

Mobile phone version: NO, the computer does not even have an in-built scatter chart. Scatter analysis must be done entirely manually with pen and paper. When it is done manually, the computer makes predictions as if the ball bounces the same distance for any wheel rotor speed.

FFA & FFZ:

NO, the computer does not even have an in-built scatter chart. Scatter analysis must be done entirely manually with pen and paper. When it is done manually, the computer makes predictions as if the ball bounces the same distance for any wheel rotor speed.

Psion 3a version: Scatter chart only. You must manually check the scatter graph away from the roulette wheel, then manually tune. The tuning is not automatic.

Mobile phone version: no scatter chart at all. You must manually draw a scatter chart on paper. This is highly suspicious at the roulette wheel.

FFA & FFZ:

NO, there is no scatter graph of any kind. You must manually draw a scatter chart on paper. This is highly suspicious at the roulette wheel.

FFA: YES, although as the device does not make any adjustments for the inevitable variation of ball deceleration rates over time, obtaining early predictions dramatically reduces accuracy.

FFZ: NO. But provided the ball track is not too deformed, and assuming the ball deceleration rate doesn't change (which is does) then in theory this device can do it, but not in real casino conditions because you have to see the number under the ball at the time of the electric zap. And for numbers to be seen at this time, the ball is moving far too quickly. Bare in mind this is all assuming you can actually see the ball when the zap occurs because in real casino conditions, you cannot see the entire wheel.

FFA: Single number only

FFZ: Electric zap when the ball passes the "raw" predicted number (single number), then you mentally/visually tune to get your prediction on each spin (near impossible)

NO. At first sight, the device is quite obviously a roulette computer. If you are ever searched, there will be no doubt you have been using a roulette computer.

FFZ: No

FFA: YES, the computer announces "now" when the ball is predicted to fall. In reality it lets you see the accuracy is fantastic in the practise DVDs, but terrible in real casinos and modern wheels.

2,3,4,5 or 6 (user choice), OR

4 - 15 or so. There is no defined number of clicks as you keep clicking until the ball reaches the desired speed.

NOTE: You can set which option you prefer in the user settings.

Maximum of 1KB code and 68 bytes memory (not KB, just Bytes). This is extremely small with very simplistic algorithms that are incapable of beating modern roulette wheels. To give you an idea of how small 1KB is, the below image is just over 1.2KB:

YES, although some of our players have witnessed failed demonstrations. One individual was even heavily insulted when he refused to purchase, even at a discounted rate.

Yes, and the tester (Michael Barnett) came to the same conclusions we did.

TESTING OF FFZ:

Regarding Barnett's tests on a physically flawed (tilted) wheel, he stated:

"At the precise moment of this “zap”, the operator must observe the number that is directly under the ball. Given that the ball is traveling in one direction at the rate of around 2.5 meters per second and the rotor is traveling in the other direction at about half that speed, this is no mean feat, particularly in view of the fact that you don’t know when or where this zap will happen. If you are to have any hope in identifying the number under the ball at zap time you’d better hone your peripheral vision skills."

He also stated:

"FF does a good job of giving you a consistent reference point on wheels that exhibit stator bias (drop_zone) but there are several others that do this relatively trivial task (for a computer) equally well and don’t require you to have the visual acuity of a Nighthawk."

In other words, it's basic algorithms can beat raw heavily flawed wheels, but for the trivial task of beating such flawed wheels, the design is very poor.

Regarding his tests on a roulette wheel that is not physically flawed (level), he stated the following:

When the device did produce predictions the results were close to random. The testers were unable to obtain a statistically significant advantage on any of the test sessions.

We agree with his findings, although he neglected to state prediction with this device is only "possible" if you can see the ball when the zap occurs. This is likely because his testing was done with slow motion video to ensure the correct predicted number was obtained.

TESTING OF FFA:

Mr. Barnett stated the FFA device was far more practical, and that it was able to give predictions for most spins. He stated that accuracy was excellent, unless consideration was given to what happened when the ball fell - which is far from a complete test, because if a computer cannot accurately predict the actual winning number, it is of no use in real caisnos.

His tests were conducted in an environment where spins were very frequent, and unlike real casino conditions where air pressure variations gradually vary ball deceleration rates. The algorithms of FFA are almost identical to FFZ, and Mr. Barnett previously stated that the FF device was able to achieve a significant edge of tilted wheels. However, that it was unable to achieve an edge if consideration is given to what happens when the ball actually falls. In other words, it has no effective method to predict how the ball will bounce - our devices do not have this limitation.