Author Topic: Roulette Computer | How they work  (Read 3134 times)

0 Members and 1 Guest are viewing this topic.


  • Administrator
  • *****
  • Posts: 2272
  • Owner
    • View Profile
    • Professional Roulette System
  • Rated: +90
Roulette Computer | How they work
« on: March 03, 2013, 09:46:28 PM »
My roulette computer site has a detailed explanation of these devices including what they do and how they work, at . You need to see each page for the different components so you understand. But many people read through it and still don't understand the difference between a roulette computer that is useless, and one that is capable of beating modern roulette wheels. So here I'll simplify it.

NOTE: Roulette computer sellers say whatever they must to "sell", which includes unsubstantiated claims about my roulette computer technology. They don't expect you to actually check the accuracy of their claims. Read their claims and the simple facts at To avoid a particularly relentless scammer, see and

What Does a Roulette Computer Do?

This depends on the actual computer, but to keep things simple, we'll first consider what the typical roulette computer does:

Almost every roulette computer does the following:

1. First the user sets a "threshold" time. This is basically a "target ball speed" measured in "milliseconds per revolution". Let's say we set 1000ms (1 second). This means the roulette computer's target ball speed is 1 second per revolution.

2. Now the player takes the ball sample. To do this, they just choose a fixed timing reference point such as a diamond (metal deflector), and click each time the ball passes it, until the ball finally falls. Now the player is ready to actually get predictions. But at this stage, the predictions are more like "rough estimates" because we dont yet know how far the ball will actually bounce.

3. Now play mode starts. The player first waits for the green zero to be at the "reference point" (diamond), then a hidden button is clicked. When the green zero is at the same diamond for the second time, the button is clicked again. This establishes the "wheel timing".

4. Now the player starts clicking each time the ball is at the "reference point" (diamond). They keep clicking UNTIL they get a prediction.

This prediction is based simply on the number that is expected to be under the "reference point" (diamond) at the time the ball is expected to fall. Very simple. Again it is only a rough estimate at this point. These initial predictions are not tuned because they don't account for how far the ball will actually bounce, so I call them "raw predictions".

How To Tune to Roulette Computer "Raw Predictions"

Initially the roulette computer gives us a raw prediction on each spin. To find out how far the ball actually lands from the "raw prediction", we simply compare the distance in pockets from the "raw prediction" to where the ball actually landed.

So if our raw prediction was 0, and the ball landed on 32, this would be +1 pocket because 32 is 1 pocket clockwise of 0. I call this difference between the raw prediction and the actual winning number the "jump" distance.

Now let's say we tested the computer on 30 or so spins, and put the "jump" differences on a chart, to see if certain "jump" values occur more frequently than others. On a suitable wheel, there would be. Let's say the jump value of around +10 seems to occur frequently. All we need to do is ask the computer to adjust the raw predictions by +10 pockets. And from now on, we'll have "tuned" predictions.

The Basic Roulette Computer Algorithm

A basic computer does just what is described above. Simply put, it determines when the ball has passed the target speed, then determines how far the rotor will travel from that time to when the ball is predicted to fall. It is literally grade 3 mathematics. It does not account for anything like:

* which diamond will actually be hit
* the different distance the ball will bounce depending on rotor speed
* the ball deceleration rate changes resulting from ball friction changes that always happen

... and there is a LOT more such basic algorithms dont do. The computers that take this very simplistic approach are.. well, every computer available, including all of Forester's computers, and all of Mark Howe's computers. My computers are capable of this approach simplistic, when the settings are set to use such a basic approach.

The actual algorithm is basically:

Step 1: The user takes 1 revolution timing of the green zero (1 rotor revolution)

Step 2: The user starts clicking each time the ball completes a revolution... UNTIL the time interval between clicks is greater than the time threshold (in our example, we used 1000ms).

Step 3: The computer calculates the rotor speed, the defined deceleration, and the approximate ball fall time from when the ball passes the speed threshold. Then it simply calculates how far the rotor will move, or more precisely what number will be under the reference point at the approximate ball fall time.

NASA technology? No. Grade 3 mathematics? Yes. So now you know what roulette computer sellers mean when they say "the best roulette computers".

Mark Howe's Roulette Computer

If you know Mark Howe, you'd know he is quite literally a sick person and creates all sorts of lies to sell his scam roulette computer. But let's try to give him some credit: his best roulette computer is his Psion 3 model, which is available for free thanks to people Mark scammed (see His approach and algorithms are just like the typical roulette computer algorithm, with the below exceptions:

a. Rotor must be timed AFTER the ball: this is wrong. Every other computer does rotor first, then the ball. How Mark could get such a critical thing wrong is a mystery.

b. Mark tries to model the ball deceleration in a remarkably poor way. He can get predictions anytime in the spin after the 4 ball clicks are done, but then his algorithm loses between 1 to 3 ball revolutions accuracy. This is a huge accuracy loss, even if for only the easily beaten wheels. This is actually less accurate than an average visual ballistics player who can be accurate to within +/-1 ball revolution.

Anyway that's Mark's best computer explained in a nutshell. As for his current computers that he calls the best ever (the only ones that havent yet been given away for free by people he scammed), they are far worse than Mark's psion model. Why? Because they are just thumpers. What are thumpers? Just a device that you click to start, and the device just gives little pulses... Yes, that's it. The equivalent is if you set a phone to vibrate every 1000ms. Nothing more.

Anyway there is a detailed review about Mark Howe's roulette computers at

Of course Mark rants on about his computers do this and that, but he just lies and hopes his customers don't have any idea what they're doing. Nevertheless, his computers are cheap, so why not give it a try and see for yourself. But EXPECT you will never get a refund, and you will never make money with it. Many people have even sent him money but he actually sends nothing - not even proof of shipping. Anyway if you do give his computers a try and he actually sends something, do the basic testing recommended on so you can see for yourself you've been scammed. I've had so many people complain to me about him, and to that I say just report him. But Mark is a small-time scammer that isnt a police priority, so don't expect anything to come of it. Last I heard he fled to Norway to evade arrest.

Forester's (Miro Zirdum's) Roulette Computers

Past all of Forester's claims of superiority, his roulette computers are just essentially "the basics". There's a summary of one player's experience with forester at which they wrote after being banned from his forum for asking reasonable questions. A few other players were treated the same way for asking forester logical questions.

The following is a timeline showing how his computers have evolved:

1. First FFZ: The basic algorithm explained above. The player received an electric zap when the ball was above the RAW PREDICTION. This used just a few lines of programming and grade 3 mathematics. It was 1 step better than a thumper. See the near the bottom which explains what a thumper basically is. I did explain it above, but the FFZ is slightly different. A full review of the FFZ is at - His FFZ device is much the same as my Basic computer which I give free to my players except my basic computer (free) gives vibrations instead of electric zaps.

2. First FFA: This did much the same thing as the FFZ, with the difference being the computer SPOKE predictions, which made it actually possible to use it (disregarding the fact that accuracy was unchanged)

At this time, Forester's computers were at a super-basic stage. But of course Forester claimed his were NASA tech. He used big claims like his computers were hundreds of times more accurate because his hardware was 0.000000s accurate. On this note, when you manually click a button, there is about 50ms inaccuracy because human's cant click that accurately. Let's not even consider the inaccuracies introduced with FFZ by trying to see what ball is under the zap when it is moving so fast, with the rotor in opposite direction, then trying to count how many pockets you need to adjust. Anyway, FFA just speaks predictions, but given the manual clocking errors of 50ms, any advantages of his hardware are INSIGNIFICANT. He knows this. He just doesn't care. He's running a business and needs to appear competitive.

3. FFA unlimited: At first, FFA could only predict at the ball threshold speed, as basic computers do. Mine did, so he claimed what I achieved was impossible and therefore a scam. He created FFA-unlimited, which gets predictions at almost anytime in the spin. Basically it simply assumes the ball will decelerate at a nice even and LINEAR rate. A very vague assumption, and very inaccurate.

Now this assumption is ok IF you only consider a very range of ball speeds AND IF you are playing a very heavily tilted wheel. Thats because the wheel is so easily beaten that you can afford to have the sloppiest and least accurate timings possible, and still have reasonable accuracy. Thats what he demonstrates in his sales videos, but let's look at the real world: in the video I show exactly how inaccurate his "unlimited" version is. Upon review of the video, it is long and boring but still shows what you need. You will note that after each click, a new prediction is given. If his approach was accurate, the predictions would not change so much during the spin. But the fact is they change considerably.

Putting it another way, the closer each successive prediction is, the greater the accuracy. As you can see, not exactly good.

4. FFA with point set: this was an attempt at diamond targeting, but was later changed as he saw it didnt work for too many reasons to explain. He now uses a different method which is equally as unsuitable but he hasnt changed it yet. He still has to fix the problem with the inevitable ball deceleration rates that change during play for his computer to have an accurate version of "diamond targeting", and it cant be done with the typical simplistic algorithm.

5. FFA today: essentially everything explained above, plus:

a. ability to enter many different offsets, based on rotor speed. It is an attempt to account for different ball bounce behavior on different rotor speeds. Some of the problems with his approach are:

- How does the user collect the data? The user needs to manually save the rotor speed on each spin, write it down, manually chart it, then manually enter all the offset data. Even if it were practical, you would waste hours doing it. And then when ball deceleration rate has changed, you need to start all again. It cant be done that way. Again he needs to address other issues I explained to him many times (as explained below too).

- The computer does not account for the ball's gradual change in deceleration rates. This makes his diamond targeting and the above algorithm obsolete even IF they were correct in the first place.

A detailed review of the FFA device is at

Over the years, he has made my computers to be more like mine. But at this stage his FFA is about the equivalent of my lite computer $2000. My Uber version is $30,000 and Hybrid version is $80,000. While they are not cheap, they are quite a lot better than Forester's FFA or my Lite computer version - ie they achieve very high accuracy where other computers achieve none, plus they are far quicker and easier to use.

Unfortunately none of Forester's computers are any better than basic visual ballistics, which is why no-one has ever made significant money with them. Although Forester claims one person has made close to $1m and he apparently travels the world with the only problem being he gets stopped by airport staff because he carries so much money. But nobody but Forester knows him, and apparently he refuses to post on Forester's forum because he wants to hide his identity. You can figure this one out for yourself.


Most roulette computers are no better than the free visual ballistics technique I offer in the video below:

Free Roulette System | Video 1 of 3

But dont expect roulette computer sellers to admit it. Instead, expect them to tell you their computer is like NASA technology, and that all others are scams. Sure, I have my roulette computers too and claim they are best. So who do you believe? . . . Do your own research and find out.

VERIFY FOR YOURSELF what each roulette computer seller says. Forget the personal junk and empty claims like "he is a liar", or "look at what some guy said about him" . . . and focus on the critical data. A good source of comparison for the details is but in cases where the seller like Mark Howe blatantly lies about what his computer does, I cant do anything about that, and you'll just need to find out for yourself.

A few points to focus on:

1. Almost all roulette computers do exactly the same thing and will achieve the same accuracy as each other, unless a computer does something different (properly).

2. If you rely on anyone's word alone, you can expect to be misled. FIND OUT FOR YOURSELF.

3. Ask roulette computer sellers the right questions. I've had many people tell me doing this drives people like Mark Howe and Forester nuts, because they end up in a corner because of lies they've told.

In my public roulette computer demos like the ones below (recordings of events), I used a combination of the most basic settings properly, with a backup that does adjustments for ball deceleration variations. It is still well beyond even the most advanced settings of other computers, but for my computers I call it just basic settings.

Demo 1:  In-person public demo:

Demo 2 - Live webcam demo


  • Administrator
  • *****
  • Posts: 2272
  • Owner
    • View Profile
    • Professional Roulette System
  • Rated: +90
Re: Roulette Computer | How they work
« Reply #1 on: March 03, 2013, 09:46:53 PM »
This post will clarify the difference in timing accuracy between my image recognition hybrid roulette computer, and any other computer that uses clicks of a button to get ball and rotor timings.

Summary of each method to get timings:

1. Hybrid roulette computer ( uses image recognition technology to track the ball and green zero as it passes. Timing accuracy is within 5ms, meaning there is about 5ms error (5/1000 of a second)

2. All other computers: the player clicks a hidden button each time the ball and green zero passes. Timing accuracy is within 50ms, meaning the timings will have about 50ms error (50/1000)

So the Hybrid's timings are about 10 times more accurate than any computer.

Myths Explained:

One particular seller of roulette computers doesn't have any automated timing equipment like the hybrid, so instead of creating one himself, he spreads various myths which are addressed below:

False claim 1: Image recognition timings are limited to 30 frames per second, which is a limit of 33ms accuracy.

Truth: Even if this were true, 33ms error is still much better than 50ms error. But the fact is even my basic automatic timing computers do much better. For example, see the video below. This is my earliest version Hybrid. As explained in the video, there are 3 black boxes and each is capable of running at 60 frames per second. Combined the equipment is capable of 60 + 60 + 60 = 180 frames per second, which gives us less than 5ms error. Is that even needed? Only for wheels that are difficult to beat and require specific features such as diamond targeting.

Hybrid Roulette Computer V1

But this is just my basic Hybrid model.

Let's compare it to the latest model:

Roulette Computer with Auto Predictions

This model also gets timings accurate to within 5ms, but it uses different technology. I've censored parts of the software, but you can clearly see the tracking box that follows a specific area on the wheel. When the ball is in the box, you see another windows highlight the ball as it passes. Now the ball isnt just detected once on each pass, so it isnt limited by the video frame rate. Specifically the ball can be detected 3-4 or more times on each pass when the ball is at a fast speed. The same thing is done for the rotor/green zero.

To see this in action, watch towards the end of the spin when the vertical scroll bar appears on the left. That scroll bar is part of a box that contains all the timings obtained for each pass. As the box fills up with timings, it eventually needs a scroll bar so they can all be seen. The box isnt filled until the end of the spin, so only at the end do you see the scroll bar appear. You will especially notice the scroll bar moves multiple lines very quickly. And towards the end of the spin, there are about 10 detections of the ball for each pass. Earlier on, there are about 5 detections for each pass.

What does this mean? Well for starters, we have about 30 frames per second accuracy (33ms error), which is already much better than the 50ms for normal computers. But basically when there are 5 detections, this is the equivalent of 5 people "manually" clicking buttons with very high accuracy, all at the same time.

Additionally, we could broaden the detection area to half the wheel and get 50 timings in one pass and have timings accurate to 1ms. Big overkill though.

Additionally, the hybrid can take into account the POSITION at which the ball is detected. Then it can adjust the timing. After all, the ball will not always be detected when the ball is right AT the the diamond. But we dont use this feature as it is overkill and not needed. We already get 5ms accuracy with the current method.

Normal computers can never compete with this. The closest they can get is to have multiple people taking timings of the same spin. And my uber model computer is the only one that can do this as it can use two phones to take timings at the same time. Thats two players and half the error, plus much earlier predictions without the errors. See - unfortunately it really is much easier for 'competitors' to spread lies about my technology than it is for them to do the work and create better technology.

Anyway to summarize this point, if you see anyone saying image recognition computers can only be 30ms accurate because of the frame rate, you can be sure they have no experience in the area, or are just trying to make their devices seem more 'competitive'.

False claim 2: The other roulette computer seller claims his devices have 0.0000001ms accuracy.

Truth: Misleading nonsense designed to trick people that don't know better. Firstly, that is a theoretical number for his hardware, and the hardware in reality never achieves even close to that accuracy. Recently he found the accuracy is actually about 1ms, which is the same as my standard computers. Even that is overkill for hardware, but hey lets assume it really is 0.000001ms accuracy. Still  it makes no difference between the manual clicking of buttons from a human is only 50ms accurate. Simple fact that he prefers not to tell you about. Put in layman terms, the hardware itself makes no significant difference because manual clocking (button clicking) already begins with enormous errors. The difference between 1ms and 0.0000001ms is rather small. But it sounds really cool and high tech to use a lot of digits.

Perhaps you should expect that roulette computer sellers are going to tell everyone what they have is best, and that all others are scams. But when you use your head and actually investigate their claims, you find out who's full of it. The false claims come from a guy that said my roulette computer would give random predictions because my equipment cant process accurate timings. Anyone that actually investigates properly knows he's just trying to sell his product. It is just much easier for a dishonest seller to publish rubbish about other computer developers than it is to create better technology.

Share via delicious Share via digg Share via facebook Share via furl Share via linkedin Share via myspace Share via reddit Share via stumble Share via technorati Share via twitter

Please help. Playtech live games don't work on my computer anymore

Started by iggiv

8 Replies
Last post November 19, 2013, 12:50:55 PM
by iggiv
Roulette computers compared: FAA Vs. my Free Roulette Computer (basic version)

Started by Steve

8 Replies
Last post October 14, 2014, 12:33:32 PM
by Steve
Do you have a roulette computer?

Started by nkmagic

0 Replies
Last post September 06, 2013, 10:50:42 PM
by nkmagic
Roulette Computer Partnerships

Started by Steve

0 Replies
Last post January 09, 2015, 06:00:03 AM
by Steve