@Forester regarding prediction value of ball rev time

[Kelly]

Well a strip search if the casino suspects somthing, is done without hesitation. They did it to Laslo Kowacs after he took a couple of 100.000s of an Australian Casino. They found a so called  speaking stop watch.

They did it to Christian Kaisan after they found a foot operated device on him in Austria.  He  met up in my home town (Vejle, Denmark) casino and they made him take of his shoes before they let him in. He then took 600.000 Dkr (100.000 $) of them and that was the last time they wanted to see him again. Now he can`t play anywhere in the Casino Austria group. The last times was without device.

The Ritz called in Barnett because he is THE one to ask among Surveillance and Security companys in table games, regarding electronic equipment. There was a lot of money at stake, so his fees is not important.

He also spoke to Scotland Yard and the gaming board.

They played large sums on a very large sector, so with an edge and maybe a positive fluctation and maximum bet size, it obviously payed off.

Mikes background is electronics and he has consructed Roulette computers, Baccarat shuffle tracking computer device, Black Jack computers. The works.

If you check different gaming gatherings and seminars, you will most likely see Barnett as one of the speakers along with Forbe etc.

[forester]

Same wheel 1.
kind of ball, system learns wheel /ball and predicts according to it.
Same wheel 2.
kind of ball, system learns wheel /ball and predicts according to I,t but predicted number does not have to match final number in same distance as with example with ball 1.

It can be done but there is no need for it, because it can cause more problems than benefits. I refer to level wheel prediction with different deceleration. On your next question I do not have answer, I do have smart solution but it is not artificial intelligence. Maybe it could help more, or maybe it can just complicate it more. I think Rollo is right with his explanation, I would say it is to complicated way and small change on the wheel would disturb data.

With new system of mine i need only few spins to start playing- keep it simple.  Yes there are many more things with Scott's system. All that I said, is it isn't for me. Even to apply basics it is too much.

However if there was no Mark and Scott I maybe never would put attention on tilted wheel. To be able to take maximum advantage on tilted wheel, position where ball hits diamond must be predicted. My system splits positions in 3 parts. Very top part where there I snot much advantage, or there is none. Middle of top part, where scatter is very predictable, and about middle of diamond which may be also very predictable. So if 60%of spin will hit diamond then only 2/3 of that can produce high advantage. Barnett's computer uses very selected spins where is the highest chance. If wheel is strongly tilted good timer can do the job. I think one of my friends is using 1.2 sec pulses that my device produces on the start, in his prediction.

I did not experiment with short times. He uses some combination of Scott's prediction. My opinion is that it will never produce result as measured ball by accurate device. Of course if we do it manually that we may load system with incorrect data. It is not easy to create correction where few parameters are changing at same time, but I think I did it reasonably well. Easier solution would be something as anam suggesting but it would take many spins to set it up, then we may lose it all if conditions change. I think Mark is doing something that way, at least that is what I could see from his DVD. I use very different way.

[Kelly]

Just one question:

How many seconds from ball drop/kollision do you make your prediction ?

[forester]

More then 9 sec.
Under that time deceleration loses I would call it linearity.
DIFFERNEMCE IN BETWEEN SPINS
X
XX
XXX
XXXX
XXXXX
XXX         / START OF CHANGE
XXXX
XXXXX
XXXXXX
XXXXXXX
I AM USING TOP PART, IN YOUR PREDICTION YOU ARE USING BOTTOM PART
Part in between is not very good for prediction.
Even my E2 system must exit before that part or prediction start shifting.
It does slowly shift even if played properly, how much it depends on which wheel is applied and how much system is balanced for that wheel.
Prediction is still within 10 numbers.
My E5 system can do it within 3 numbers and wheel speed does not have any change. However E5 it is designed for particular wheel. If applied on slightly different wheel prediction may start shifting. To limit it the system uses only particular ball timings where even if there is shifting it is limited and if time would cause to much of shifting spin would be invalid. Same if system thinks that timing of ball is incorrect.
IQE5 Is same as E5 but it has some intelligence, it takes few spins to tune itself for wheel where it is applied. It will not predict only if detects serious problem in ball timing based on few ball times and defined deceleration for particular times.
In computer prediction it is not only important what to do with recorded data but how do you take data.
I program in assembler and I know exactly how long each instruction takes.
If you time ball in milliseconds that process is done in subroutines and each line in program takes time to be executed. Now if you time 1 sec. the time of each instruction is multiplied by 1000. So the real time may be longer then 1 sec.
If switch is pressed to start clocking after 1 ms it will still be pressed so we need delay, while counting is still done. Part with delay and part where closed switch would mean end of clocked time, must be same number of  software lines. Many more things that need to be taken care how they are calculated. It is takes months of work.
I have received additional DVD for leveled wheel from Bob Gordon. http://www.roulettespins.com . He is selling 8 DVDs for price of one bet on table.
And if you are nice to him he sends you even some extra spins based on your needs.

When I find more time I will publish results of scatter analyzes on Bobs wheel and results of my prediction applied on his wheel.

[Kelly]

I guess my question was kind of faulty, because i meant with your visual method.  Which "E" is which i have kind of lost track of.

9 secs out is roughly 5,5 ball revs before ball drop.

I will keep my opinion to myself because i won`t start an argument, but i will say that you will get no known AP players to belive that you can predict on a level wheel 5-6 revoloutions before drop. If you are very accurate you will start to see something between 3 and 4th last. Even there, you have no really safe indication on the kollision diamond.

I know that in theory you may be able to make a linear line from prediction to kollision.  Problem is, that one speed at the 5th last round can lead to a drop on the "North" diamond but also at the "South" diamond.  Or 2 different speeds at 5th last leads to same kollision diamond.  So the problem is to step at the correct point in the linear line knowing that it actually IS the correct point.  The
predictability increases very fast from the 4th and forward but is still extremely diffycult.

I don`t fancy the training dvd`s much anymore because  they rarely resembles the wheels i meet in the casino and im past the so called training stage where you still can improve your tecknik.

[forester]

[anam]

Forester said:

More then 9 sec.
Under that time deceleration loses I would call it linearity.
DIFFERNEMCE IN BETWEEN SPINS
X
XX
XXX
XXXX
XXXXX
XXX / START OF CHANGE
XXXX
XXXXX
XXXXXX
XXXXXXX

So, you are saying that the change in ball velocity has two distinct phases in a roulette spin and that the first phase (farthest out from the ball drop) is the phase you use to make your prediction.  Therefore, your prediction is based on change in ball velocity.  Am I correct?  

Also, you are measuring ball velocity with respect to a fixed point on the wheel rim and not a specific wheel slot (eg the 0).  Am I correct here too?

regards,

anam[/b]

[forester]

Anam that is change on differences in between 2 spins.
Wheel 1. One ball rotation 1 sec next one 1.2 sec.
Wheel 2 one ball rotation 1 sec next one 1.160 sec.
Now the ball clocked time of 1.1 sec would not be the same on wheel 1 as on wheel 2.
Wheel 1. Ball clocked time of 800 ms is not one full rotation from 1 sec rotation. But it is more then that. How much it is defined by change of time changes from rotation to rotation on particular wheel. Defining difference of differences is complicated process considering possibilities of errors created by our change in clocking response time.
Process of defining accurate time of ball rotation is linked with that results, same as interpretation what does it mean to final result.
Yes that is how I do measure it, with E5 wheel is calculated separately.
E2 observes in known time, change of the ball in relationship to the any number on the wheel. Time needs to be adjusted to length where the ball would slow down by 1/3 of speed at start of observation. Therefore when change on the end is multiplied by 3 it gives result how far the ball was from reference point when we started. Observation is in relationship to number therefore wheel speed change is included up to the limits.

[Kelly]

Well the most precise opportunity of clocking i have ever had is on a 600 spin dvd where a stop clock is running in the front.

If you run it real slow and freeze the picture at the exact spot, you can simply read the stop watch at that time. Move it 3 slots and freeze again and read the stop watch and you can calculate the speed.

In real life i wouldn`t be able to anything like that.  Its purely for experimenting.  There are already known prediction tecknikes that works visually and the lack of edge, compared to a laser device, is fully compensated for by using split bets (no forced tip as on hit on single numbers in europe) and Kelly money management to catapult the actual money winnings upwards.

 You can easyli swap a 25% edge with a 10% edge and make more money with the 10% edge than the 25 if you use a full Kelly compared to a flat bet 25.

My point is, go out and get some dough if you have an edge instead of trying to squeeze the edge to ridicoulous hights, where the winning amount of units is crawling sky high in Standard Deviation and is raising eyebrows in the pit immeadeately.

You may also win a large amount with Kelly, but your hit rate is still smaller (with a smaller edge) and may be in line with deviations although they are a result of an edge.

[forester]

That is very correct. Betting only 1 number in pick area should be the most profitable, but it is not the most practical way. No matter what is our advantage we can still lose?

The highest ever hit rate that I have had was in 33 spins that 30 spins ball end up in predicted sector of 6 numbers.

But that was the edge created by prediction and luck.
Sometimes it can be opposite. Every spin spot on predicted, ball can bounce unpredictably and every spin slightly off predicted ball stops as predicted.
On top of that we can add wrongly placed bets therefore it is sure that final result is not only based on system advantage. Recently from 20 spins 4 times I have had wrong chip placement. 6 units on numbers 7 and 6 on 12 and only 1 spit 28-29. The ball stopped at 28 right in middle of 7 and 12. The difference in winning something as 1:12. Few spins later missed to place chips on number 22 even it was centre of prediction…etc.

[Rollo]

My point is, go out and get some dough if you have an edge instead of trying to squeeze the edge to ridicoulous hights,

It all depends on how the scatter distrubution looks like, which I have little or no knowledge about. But as an example, if the peak is at 4%, closest surrounding numbers fall to 3.5% and then to 3%, then to bet on all of them reduces the edge to an average of about 3.5% (1/28.5), compared to betting only the peak of 4% (1/25). I suspect that one could reduce the profit per game with half by betting say 8 numbers instead of ony the most likely one.

There is an optimum out there somewhere. If the most likely number has the probability of 1/30 to come up, one makes on average 20% profit on the bets in the (very) long run. But there is about 3.5% probability that all the first 100 bets will loose! THe price for betting only the most likely number is that one must bet a lower fraction of ones budget at a time. Betting only 1% of the budget at a time, still will go bankrupt once a month on average (if you play 30 days a week, 100 times per day)! [the true solution is to get better odds than 1/30).

where the winning amount of units is crawling sky high in Standard Deviation and is raising eyebrows in the pit immeadeately.

But, isn't it more suspicious to bet numbers which lie next to each other on the wheel?

[forester]

xx
Xxxx
Xxxxx
Xxxxxxx
Xxxxx
Xxx
Xx

Same spins could be in this format

Xxxxx
Xxxxx
Xxx
X
Xxxxx
Xxxxxxxx
Xxx

as you said if we play 1 number, we can badly lose.
In usual I play 7 numbers trying to cover them in this format
X
Xx
Xxx
Xxxx
Xxx
Xx
X
It can reduce advantage by half but produces stability in winnings.
That is in average 15 units.
To place 15 units at only 1 number is very risky even with good prediction on 50 spins we maybe do not get single hit. I think 5-7 numbers would be optimized play.
Kelly is one that made me think that way, before I use to cover 12-15 umbers

[Rollo]

In usual I play 7 numbers trying to cover them in this format
X
Xx
Xxx
Xxxx
Xxx
Xx
X

So you bet more on the number in the center of the prediction and less on the numbers next to it? Yeah, that's a good hedge. High probability to at least double the money there.

But isn't it obvious to everyone that you are not betting to be lucky, but that you are trying to exploit a prognosis, when you bet on numbers which lie next to each other on the wheel? Doesn't that attract the attention of security personel, expelling you as soon as you are winning?

[anam]

Its been my experience that any time you are winning big you are likely to draw some heat from the casino staff, even if you are winning through blind luck alone.  With that in mind, if you do have an effective prediction system, its best to optimize  your chances with a bet spread and leave before you draw too much scrutiny due to your winning.  Just my thoughts.

anam

[forester]

Rollo, I do not make thousands every time on roulette, I am not professional player.
Most of dealers and pit bosses think that I lose more then what I win. Maybe I do   .
Sometimes dealer stops me by calling NMB earlier.
In usual I stay cool but sometimes I tease dealer and encourage other players to place late bets pointing to ball that still is spinning and spinning. If on table there are people that start placing bets when dealer spins the ball, I am lucky specially if they doing it until I start placing.
It is really interesting to watch how many kind of people work as dealers.
Some do not care and they just spin the ball.
Some acting as you taking money from them trying to stop you.
If because of the others they really cannot call NMB too early they think that by changing wheel / ball speed they can achieve something.
Some are very smart and they know what they doing, I think they even know what I am doing. Not once that I was asked why do I play as I play.
Of course I do it because I think casino can switch on magnets but if I place late bets they cannot do it :shock: . Or when wheel makes one rotation where is the ball that is the winning number. Do i care, if someone thinks that I am stupid but lucky.
If you go in casino sometimes you will see some area loaded with chips and everybody winning. Many people are playing dealers signature, and sometimes they do well. This is first hand information, casino do have specially trained dealers to break that pattern or to even set up people encourage them to play that way, and then manipulate game. After that players lose fast, because they highly increased bids.
Dealers signature is my first way of playing 20 years ago, and I still place my earlier bets based on that. I cannot say that I win by playing it. Actually on one tilted wheel it was very successful.

[Kelly]

There is no problem in defining  the sector size if the prediction scheme is telescopic:
x
xx
xxx
xx
x

You simply play the sector size that has a higher hit rate than the expectation. Once you get outside the telescopic range you get into negative expectation such as 1/40  etc. Pyramid betting is perfect or the telescopic hit rate.

Usually it is never like that in the beginning, but is getting there once you get past 4-500 actual placed bets.

If you have something like this with the exact same wheel speed:

x
xxxx
xxxx
x
xx
x
x
xxxx
xxx
xxx

You are actually better of, just having 1-2 chips on each peak.

There is not much point in playing a 6 number arc with 3 numbers with expectation 1/30 and 3 numbers with 1/44 just to get a smoother run with smaller fluctations. You will lose long term doing that.

The Huxleys more or less all produces a second main area placed 180 degrees from the absolute main area. You will see it, not only in the scatter histogram but also in the prediction histogram. Have a few chips in ALL main areas.  For the VB players, it also save most of the spins which is predicted at a revoloution too early or too soon. Because the main area in those samples, depending of wheel speed of course will be at 180 degree from original prediction. (The ball has taken a full extra revoloution and the wheel 1/2 revoloution)

Late sector bets always gives suspecion if they are large and they are winning. But in the end, as at the Ritz, they will have a hard time suing you for a hit rate which is within deviations no matter it may be a fortune.

Thats why you should learn to use the clever split bets.

Covering zero sector 12, 35, 3, 26, 0, 32, 15, 19 AND
5/10 sector 180 degre: 33, 16, 24, 5, 10, 23, 8, 30 11, 36
26/23
0/3
5/8
------------You can stop here where you got 6 numbers covered or
32/35
12/15
32/33
35/36
16/19
10/11
----------- If you pick them all, you have covered 18 numbers

But none of the bets will imeadeately be recognized as sector bets. You can do the same all the way round on the wheel.

[gkd]

I think you are worrying too much about the heat created by the spread betting. Set a betting strategy and then monitor heat (and potential heat) level to decide when to exit.

Heat is a function of winning time and winning amount.  They have seen all the startegies and systems so will take little notice. All the empirical evidence suggest that the best way to avoid "significant"  heat is win quickly and then move on.

All the statistical evidence suggests that a betting startegy based on normal distribution around the predicted number (assuming a number not sector system) is optimal. Remember the betting pattern is really bankroll driven not return as the best long term odds for a 'true' number/sector predictive system will always be a single number but your practical bankroll is unliklely to be robust (safe) under that approach.  

On that basis simply empoly a 'normal' spread as discussed by Forester above (5 or 7 makes little difference under most conditions), stake the amount supported by the bankroll calc (again good stats available) and play hit and run - get out as quickly as your winnings dictate for the game/environment you are playing in.

[mikemcbain]

gkd & co

Heat is a function of winning time and winning amount.

[/color]

The size of the Casino and Roulette table activity also have a bearing, for example at my local two table casino with a dozen regular players at any one time plus half a dozen visitors I never have a problem if I walk away with three times my buy-in amount but I get "heat" if I walk away with five times my buy-in.

At a major casino interstate with say 20 - 50 crowded tables I've never seen heat even when other players have walked away with ten times their buy-in.

Mike.

[forester]

x
xxxx
xxxx
x
xx
x
x
xxxx
xxx
xxx

You are actually better of, just having 1-2 chips on each peak.

Playing 1-2 chips can take forever, that scatter is always coincidence, soon can position 4, 5, 6 start getting more hits.
2 close pick points are big problem, placing bets on one and ball goes for the other one, after 5-6 spins shift prediction and ball starts hitting the other pick. Not once happened that 80% of spins are within 18 numbers and I cannot profit...

[Kelly]

Well that is your opinion. Obviously you have played where there weren`t 2 peaks.

Of course you don`t anticipate 2 peaks, you KNOW  that there are 2 peaks, otherwise you don`t play that way.

  Ask Kaisan what happens when you play outside the peaks. His name is "sachse" on the paroli board.  There are wheels where ALL your edge is gone if you start placing bets outside the peaks which might not be bigger than 1-3 numbers. Kaisan doesn`t play wheels he hasn`t tracked for at least 300 spins.

Here are another scatter gram that might illustrate that the scatter ALONE is enough to create more peaks:

[Rollo]

The two peaks in both scatter diagrams above (and in other scatter diagrams I have seen) are 18 numbers apart. That is half a wheel. (A pair of smaller peaks are visable in the right diagram at 15 and 33, +9 and -9 numbers apart from the main peak, or a quarter of a wheel).

Is it because the ball sometimes jumps across the center of the wheel? Or has it to do with the vertical diamonds (which are distributed 90 degrees apart)? I don't get it.

By the way, the scatter looks good to me. In the diagram to the left, one doubles the money either of the two peak you go for. (The wheel of the left diagram looks even more fantastic). They are made from 111 observations, which certainly looks enough to be conclusive. 300 observations would be excessive, taking something like 10 hours. Peaks like those in the right diagram should be identifiable after as little as 40 spins, I would say. Enough for being worthwhile to play on, I mean.

(Again, I want to remind you all that my experience from actually playing roulette is quite limited. My reasoning is theoretical.)

[Kelly]

At this particular Huxley, the reason was that either ran the ball forwards between 6 and 24 slots, but also very often the ball dropped, hit one of the seperators and jumped backwards.

When you in your "prediction locator scheme", (same as the scattergrams, you just have your prediction point in the middle and from +1+2...+18 one way and -1-2-3...-18 the other way) is getting 2 peaks roughly 18 numbers apart, it is very often because there might be a tilt that occasionally makes the ball take 1 extra revoloution more than expected  and in that time the rotor has turned 1/2 a wheel size and your main area will be 1/2 a wheel away from the original point when the ball drops.

Same as Barnetts computer will show, if it frequently is predicting an early key revoloution.

[mikemcbain]

Kelly said...

makes the ball take 1 extra revoloution more than expected and in that time the rotor has turned 1/2 a wheel size and your main area will be 1/2 a wheel away from the original point when the ball drops.

Kelly, could you elaborate on this a little more please?

If the wheel is running at 3 secs per rev then doesn't it require the ball to be travelling at exactly half of that or 1.5 secs per rev to finish up 18 pockets from where it would have if it hit the same diamond on the previous rotation?

How much leeway or error margin is there in these times before the ball misses that same diamond I wonder?

Mike.

[forester]

If pick point is only 3 numbers and if we play numbers next to it and losing because of that we cannot win. Who can have prediction within 3 numbers accurate?
Pick point developed by looking distance from diamond does not match pick developed from prediction. Of course that Barnett will have 2 pick points if he uses 1200-1400ms.
If braking point in between 2 revolutions is 1300ms, all detected speeds from 1200 -1300ms will be shifted by about 1.3s x wheel speed.

But that is not all of error. Even speeds detected for same ball revolution will not have same result. If it is detected 1200ms for remaining amount of rotations will take less time  if compared to detected 1300ms, That shifting is reversed to previous one (1.3s x wheel speed) so the error increases. I think after observing wheel that he tunes computer so that 1 ball rotation fits better in time frame.

[Kelly]

@Mike
It doesn`t fit all wheel/ball ensembles.  If you have the last round taking 2.0 secs it fits 4 sec. rotors.  You will also have wheel/ball ensembles where the the last ball rev is slightly faster and that might fit a a 3.8 or 3.9 rotor better.

On top of that there are other factors. If the ball takes 1 more revoloution than it was supposed to, it usually has a lot LESS energy in this final round that causes it to drop like a rock almost without scatter instead of bouncing forwards. So in the end, a 1/2 wheel rotation might not be ideal even if the velocities speaks for it.

Some velocities actually gives you 3 peaks instead of 2 mainly because the rotor is so much slower that it only turns 12 slots instead of 18 and on top of that has a frequent bounce causing it to even 12 slots furteher or maybe just 2 peaks but where the second is placed 12 slots after the main.

Exactly where the peak in prediction is located, you will only be able to see in your prediction locator scheme.

@forester
If i get a visual tracking that shows  a peak of 1-2 numbers breaking through 3 or 4 SD, i will play that offset  only with 1-2 numbers in exactly  the same way i got the data for the tracking. Usually, i will check wether i get a better  edge or higher SD if i extend  the sector to either 5, 7 or 9 numbers and i will then pick the sector size that yields the better opportunity.  I have had wheels where the edge is gone playing 5 instead
of 2 numbers but was heavy (higher than 3 SD) again playing 9 numbers.


I don`t know any details on Barnetts computer, you will have to ask him.