FF versus Mark Howe computer

[Rollo]

Yes. I've seen a casino which have them on a couple tables. It's called "race track" or Rennbahn.

I don't know why some casino choose to offer this service. Maybe there are some popular "betting systems" which it faciilitates.

[Fatgambler]

Abc have you tested repeated the same spin severel times? So you can understand if it is just a random numbergenerator or a roulette comp that have wrong formula/algorithms?

[Myrulet-admin]

He did all kind of tests but it was as RNG even if repeats same spin and do clocking at same place.
I do not know where ABC is. For long time I did not hear from him.
Probably he is to busy counting money or to busy working to make up for loses. 

[Fatgambler]

Ok, but have any one tried using ff on his demowheel? Does any one know if it is tilted or leveled.

[Myrulet-admin]

I have some of his videos including one that he was selling for 100 pounds (tilt).
I can't remember well on which video but I know that some of his videos I couldn't get predictions because the ball decelerating to fast, picture is bad quality and on some the prediction was ok. 
At his last video he is clocking the ball when it is only 0.5 sec per rotation. It is imposable to get any accuracy on such fast ball, especially not if the wheel is leveled.
However his prediction on that video looks ok but it is assembled from few parts, so Mark Howe probably did video recording and select only spins where results were matching his prediction.

[Kelly]

I did my own test of Marks commputer based on the Nokia phone video a couple of years ago, here are the raw data:



Now, there are several ways to "Acid test" these results. One of them is to compare each strike that he had with the actual bounce lenght of the ball. Imagine you have an average bounce on a wheel of 12 pockets. Now imagine that the bounce lenght of 5 hits was 2, 11, 17, 27, 33 pockets (no relation to the actual test). You can then make your own opinion by actually watch those 5 hits. For example, the strike with a bounce of 11 pockets appears to be perfect because the average scatter is 12 so the ball came down exactly as it should. Now look at the bounce of 27 pockets that also gave a hit, maybe even with the same rotor speed. It was a hit, but did the computer actually predict such a large and almost non appearing bounce lenght ? Or was it a lucky strike.

You can then proceed to test the remaining spins with the same AND different relation combinations and slowly you get a picture of the actual prediction power.
I HAVE done these tests but i keep the results to myself as i don`t wanna participate in the ongoing war between the 3 of you.

What i can say about the tilt degree of the wheel is, that it appears level if you look at the strike rates of each diamond except for 1 diamond that appears 15 times but  the strike pattern looks like a level wheel. The dominant strike diamond is probably due to a worn track in combination with "dealers signature" kind of movement.

Strike diamonds:



Rotor speeds



Scatter pattern



The scatter pattern include all wheel speeds. But if you go into each spin and look at the bounce in faster and slower speeds, there are really not much difference in this short 50 spin sample. Im sure one would be able to get a more correct pattern over 2 - 300 spins.

[Myrulet-admin]

[minerox]

hi guys

good work kelly,

is this the format that you would use in looking at a wheel?

do you have software to enter your recording into or maybe i shouldn't ask
no one cares what howe and co do
i believe that we have already decieded who we believe!!!!!


minerox
mtmatssoy

[Kelly]

Im not gonna tell wether it works or not. The data is there and its pretty simple to add up. But i can tell you that i haven`t bought it.
You can actually gather a lot of info when analysing the spins that didn`t hit too. In case where his prediction was for example 9 pockets in front of the strike number, you can say that it was bad luck that it didn`t hit because the ball came down as it should, but the bounce was out of the ordinary, maybe 20 pockets.

Minerox, im not sure what you mean by format. I have collected the raw data and categorized it. The scatter pattern and the tilt/strike diamond pattern is basic for establishing wether i can beat that particular wheel at all. In Mark Howys case, i can say that using a thumper i get a very nice hit rate. There are more components in my analysing of a possible wheel opportunity but they are not nessecary to analyse his wheel and computer prediction power.

PS: Yes i have 2 kind of software that i use for the tracking. 1 from snowman and one from laurance, additionally i run a third excel sheet that laurance helped me set up with my specifications and a few input from snow and laurance.

[Myrulet-admin]

I did my own test of Marks computer based on the Nokia phone video a couple of years ago, here are the raw data:

Kelly, I am not clear with what are you trying to say.
When I look the chart it is showing good result.
Are you saying that you've tested Mark's Nokia on leveled wheel and that it works? It couldn't be years ago since he converted it from PDA only a year ago.

[Kelly]

Its from the video you can get on his homepage, and you are right, it was probably in last spring i did the test. So maybe only 13 - 14 months ago. I just thought it was longer ago.

Obviously there are some more hits than expectation, otherwise it probably would not have been used to promotion, but i was not even looking at the amount of hits, i wanted to see wether his computer was able to predict the correct drop zone. Pretty much the same thing Barnett did with your computer i think.

 In retrospect, when you know the average bounce and his prediction, its easy to count out where the ball is going to drop to come down correct.  One of Marks hits was a result of a "roller", you can ask yourself wether his computer predicted a roller and on top of that, the correct number. The 2 acid tests were:

1. Strike number versus prediction number versus average scatter.
2. Strike number versus outcome number versus average scatter.

3. Result: Transparent ground zero chart. (not shown)

I asked him how much his computer added in bounce on a faster rotor, the reply was "just add a few pockets". which for the video is okay, well, you can actually not see any correlation between bounce and a faster rotor, but i asume that over a larger spin sample you could see a difference.

[Myrulet-admin]

Kelly, are you talking about analyzing his spins or about that you've tested his computer.
His spins do look ok and there is enough consistency to rotor hit the problem is that he assembles it from few video files and on the end after longer part he lost all predictions, claiming that computer needs retuning. He was the one who recommended lava lamp on promotional video but he never used it.

Making adjustment for different rotor speed is very hard because with each ball and on every wheel it may be different. You've asked me that question and I told you only few pockets. From 5 sec to 2.5s it is only 4 pockets. If you have recommendation please tell me. The problem is that on fast rotor relationship of ball roll and ball instant stop may change in favor of instant stop. And that i do not play 2.5 because and ball traveling time may chnge.
The FF isn't a miracle and cant do it all.

[Kelly]

No all i got was the video.Like i said, he is not getting a recomendation but he is not getting a red light either, simply because i don`t wanna be a part of the circus. I got an opinion based on the  "transparent chart".

Im not much into the technical problems but i know that other european computers builds a data base where the rotor speed is saved  with the bounce lenght and that it is a part of the calculation as the learning spin data base builds up in size. You say that  from 5 sec to 2.5 sec its only 4 pockets. You mean bounce lenght ?  If its bounce lenght, all i can say is that it is extremely dependent on ball material. Teflon don`t bounce as far, but a nylon or ivorene ball on a 2.5 sec rotor with low frets you have something between 27 and 40 pockets bounce depending on the individual wheel.

[PJ]

Thanks for the chart Kelly.. at the risk of stating the obvious.. there is absolutely no consistency in the distance between 'the strike number' and the 'predicted number'.. my rabbits foot would do just as well..!! and only about 30% hit the strike number short enough of the predicted number to stand any sort of a chance of being successful..

Cheers
PJ

[forester]

Am I looking it wrongly?
25,4
22,22
7,31
12,29
35,28
32,15

[PJ]

Am I looking it wrongly?
25,4
22,22
7,31
12,29
35,28
32,15

I think Kelly means:

Ball impacts with rotor at #4.. and then bounces 10 numbers before stopping at #31.. and MH prediction was #25.

Ball impacts with rotor at #22.. and then bounces 1 number before stopping at #18.. and MH prediction was #22.

Ball impacts with rotor at #31.. and then bounces 11 numbers before stopping at #0.. and MH prediction was #7.

Ball impacts with rotor at #29.. and then bounces 8 numbers before stopping at #32.. and MH prediction was #12.  etc etc etc.

If I'm wrong Kelly please advise!

Cheers
PJ

[Kelly]

You are right PJ.  The only thing is, i think i noticed the number below the strike diamond at the moment of ball impact with the diamond.
There are several opinions on wether to call the strike number "the number below the strike diamond" or "the strike number on the rotor". Basieux goes for  the number below the diamond, because he divides the computer prediction in 2 parts: The Deterministic Part and the Chaotic Part. The deterministic part is the part where the computer determines where the ball will strike the first obstacle, (thats the part where it is actually possible to calculate what is gonna happen), which is the strike diamond and the chaotic part is the scatter.

Laurance and snow goes for calling the strike number, the number that hits the rotor. In Snows case it makes sense because if you have a pocket defect, a lose insert or a lose fret, this particular pocket will absorbe much more of the ball momentum and the bounce will be slightly shorter than the bounce from other pockets. So in case you suspect a pocket bias, you obviously look for the average bounce from the pocket and not from the diamond.

Anyway, i went for the Basieux logic because it makes most sense in this particular case.

Now, there are many ways to combine the data, and you only get a good picture by combining them in several ways. For example, take all the spins wich had the average bounce. Where there any hits in them ? Or how far were they from the actual outcome number ? Did the drop zone come close to where it should have been to make an exact strike ?

What should always be kept in mind is, if you don`t know the rotor position when the ball strikes, a prediction is simply not possible. So having that in mind and looking at the drop zones and strike numbers combined with the average scatter gives you a clear window of wether the computer actually predicts the correct drop zone in relationship with the rotor position in that exact moment.

Thers a few loose ends though. I only have the average scatter for 58 spins so the main area might not be exact to what the computer calculates, but as the scatter builds up in what we got, i think we have an allright picture.

[forester]

What should always be kept in mind is, if you don`t
know the rotor position when the ball strikes, a prediction is simply not possible.

The FFZ doesn't know rotor position at all and prediction is very possible.
FFA must know it.

[Kelly]

Are  you saying that the rotor position doesn`t matter when the ball drops ?

[forester]

It doesn't matter as long as it is related to prediction.

The FFZ doesn't know which number is there at the moment of click but knows how much rotor will shift from that number until the ball drops.
That is why with FFZ we can use any clocking points.

[Kelly]

It doesn`t know the exact number, but knows a point on the rotor.
Same thing. It knows where this point is, related to when the ball drops.

[forester]

Yes,
For the moment you made me worried :-)