Dealer's Signature. Does it Exist?

[Firefox]

In this topic, I'm going to be looking at the evidence for dealer's signature.

Not just any old conjecture though. Real live spins on my own casino quality wheel.

Sample spin:

From zero area,  4 sec rotor, fall off 24, winning number 17. Bounce and scatter approx 24 pockets.

The ball will be spun from around the zero area, 8-10 clockwise revolutions round the wheel, with an anticlockwise rotor of 4 secs per rev, 9.25 pockets per sec. The wheel is slightly tilted to produce a 2 pin game, 2 diamond dominant dropzone.

This is a very high scatter modern wheel, and very difficult to beat. Nevertheless the data will be examined to see if any trends exist. Visual ballistics were undertaken to predict 6 or 7 spins out and confirmed a drop off diamond prediction in the 10 to 24 to 33 zone 80% of the time.

12/03/2019. Cammegh High Scatter Low Profile Scallop. 4 Sec/Rev Rotor acw.  Cw B Release At 0.  18mm Heavy Teflon. 8 to 10 Ball Revolutions. 11 to 14 second spin. Slightly tilted wheel. 2 Pin game. 2 dominant diamond dropzone. Raw Data. 200 spins.

CSV Data 200 spins. 10 x lines of 20. 5 columns of 37 plus 15 spins.

31,27,11,0,17,18,10,36,26,10,2,17,25,23,1,23,1,36,1,3,
30,9,9,6,14,30,19,16,22,27,9,6,30,4,15,25,22,8,34,27,
7,30,35,3,26,35,32,32,27,19,12,19,1,3,21,8,9,5,19,15,
24,26,16,20,0,34,17,4,9,34,0,17,0,15,14,19,24,16,34,26,
19,11,9,12,23,2,25,18,3,0,30,31,4,32,36,7,12,11,33,17,
23,2,13,15,36,25,23,18,24,31,19,28,20,23,8,26,27,22,8,21,
21,6,23,16,12,34,8,24,32,34,16,0,0,5,23,20,9,34,21,7,
22,29,24,36,35,19,0,6,32,30,26,4,6,33,31,15,12,24,35,15,
21,8,1,21,6,2,36,9,9,12,3,25,11,30,15,35,23,12,32,35,
24,0,30,30,4,11,12,11,15,16,32,15,11,17,20,29,1,29,29,24

The graph of this data is attached.

The graph starts at 10, the beginning of the falloff number zone.

Although this appears relatively even, it will be noted that 75 of the 200 results occur in the first 18 pockets after the drop off while 125 of the 200 results occur in the second half of the wheel. There is a concentration of results in the 19 to 24 pockets area after the drop off.

The expected value of hits on a number would be 200/37 = 5.4

Is the just a freak result or is it a trend? More 200 spin sets to follow!

[Firefox]

More data.

13/03/2019. Cammegh High Scatter Low Profile Scallop. 4 Sec/Rev Rotor acw.  Cw B Release At 0.  18mm Heavy Teflon. 8 to 10 Ball Revolutions.11 to 14 second spin. Slightly tilted wheel. 2 Pin game. 2 dominant diamond dropzone. Raw Data.200 spins.

CSV Data 200 spins. 10 x lines of 20.

4,1,2,10,18,14,22,28,33,29,22,17,25,17,12,23,15,13,15,26,
21,12,5,25,17,20,7,34,32,0,11,32,5,16,9,7,22,21,4,19,
5,16,26,36,4,35,12,33,20,34,23,33,25,1,15,31,34,35,35,20,
4,17,3,3,2,24,15,3,6,10,10,28,14,36,18,3,10,0,23,22,
16,31,21,18,2,22,17,32,2,14,34,10,3,10,4,23,33,4,29,20,
1,24,17,1,10,36,25,13,9,4,30,35,7,10,2,29,13,32,29,14,
5,11,5,16,34,3,25,22,30,2,26,36,17,5,19,18,9,25,32,2,
31,35,16,18,31,6,31,36,24,2,23,35,0,25,11,29,26,30,9,21,
28,13,8,17,6,36,6,19,25,20,18,25,21,20,21,34,28,15,25,17,
5,19,1,7,26,19,19,29,4,21,34,35,31,36,34,5,34,22,21,19

Here more even in the first half following 10, but again a clump in the 15 to 34 sector following zero. Drop off around 8 to 33.

Too early to say anything for definite, but looks like 20-27 pockets after DD zone could be a scatter peak.

[Firefox]

Tonight's data.

Up to 600 spins now. When we get to 1000, I will amalgamate the data in one graph and do some brief statistical analysis.

NB These are all actual unchanged spin data based on a live random wheel in good condition so feel free to use for your own testing. In many cases you can just copy and paste and import the CSV data directly into another platform.

14/03/2019. Cammegh High Scatter Low Profile Scallop. 4 Sec/Rev Rotor acw.  Cw B Release At 0.  18mm Heavy Teflon. 8 to 10 Ball Revolutions.11 to 14 second spin. Slightly tilted wheel. 2 Pin game. 2 dominant diamond dropzone. Raw Data.200 spins.

CSV Data 200 spins. 10 x lines of 20.

27,19,8,29,12,17,0,26,18,23,17,5,14,13,9,15,35,12,7,33,
31,29,11,0,9,3,12,23,25,9,24,14,0,1,14,19,19,31,27,17,
19,15,19,22,21,3,2,11,27,14,30,35,23,13,32,26,23,1,13,4,
19,10,35,6,23,32,6,24,5,34,13,22,0,2,19,7,23,12,9,10,
34,13,0,25,25,29,4,35,35,29,7,34,11,34,2,36,21,25,6,9,
13,26,29,13,20,0,28,11,10,4,29,1,30,0,13,0,4,25,29,30,
6,15,21,13,35,31,16,24,29,26,28,30,31,32,29,11,25,5,35,5,
21,9,15,35,18,25,12,2,25,11,12,19,8,4,2,6,6,26,16,32,
22,11,18,21,32,13,4,18,15,15,36,36,9,9,33,13,15,4,25,27
15,4,15,21,10,24,3,4,3,7,6,27,16,4,31,26,15,11,5,23

Again the fall of diamond was predicted by VB and verified to be in the 8 to 33 zone about 80% of the time.

The data was 78 results in the first half and 122 in the second half with a possible scatter peak in the 0 to 4 area

[The General]

With this type of test it's very important to limit the degrees of freedom.   Simply plotting the numbers that come out when releasing the ball from zero will work, but it will take a ton of data to find significance. 

So here are some suggestions.

1. Let's look at the correlation values between spin samples.  I'll post the correlation values.
2. Since it's virtually impossible to spin the same distance for hundreds of spins in a row, what you really want to look at is the change in yardage from one spin to the next.  That's called the one frequency.  I've pasted the one frequency on this page for you to view.  The distance thrown will change throughout your spin period, but the one frequency, two frequency, three frequency, etc... is what you want to measure.
3. The wheel could be biased.  You should consider alternating the release numbers at random rather than always using the one.


Above is the distance in pockets for all trials when releasing the ball from the number zero area.  The graph is a standard deviation graph.
The chi is 32.9 or a chance of randomness of 1 in 1.62.  Since the best area wasn't picked in advance of the test the true standard deviation result isn't really quite as high as it appears on this graph.





(Above)Here's the one frequency.  With this chart we've limited the degrees of freedom down to just one area.  The section around 0 yardage. 



(Above)Here's a section graph of the one frequency.  Again, a measure of the consistency of throws from one spin to the next.




(Above)Here's a section graph of the two frequency.  With this chart we've limited the degrees of freedom down to just one area as well.  This is comparing the consistency of the yardage of a spin to the consistency of yardage two spins later.



Here's a section graph of the one and two frequencies combined.  Again, it's a measure of consistency  comparing the yardage on spin one to the yardage on spin two and the yardage on spin three. 

[Firefox]

I'll probably sum all the data for 1000 spins and then do curve smoothing, moving averages or some other technique and look at the standard deviations then. I'm not going to guess what the result might be to keep everything as fair as possible. I think 1000 spins should be a fair sample, but we'll see!

[Firefox]

More data today.

15/03/2019. Cammegh High Scatter Low Profile Scallop. 4 Sec/Rev Rotor acw.  Cw B Release At 0.  18mm

Heavy Teflon. 8 to 10 Ball Revolutions.11 to 14 second spin. Slightly tilted wheel. 2 Pin game. 2

dominant diamond dropzone. Raw Data. 200 spins.

CSV Data 200 spins. 10 x lines of 20.

22,21,11,8,13,22,8,26,21,11,20,4,21,9,4,2,29,31,12,26,
28,23,27,18,15,7,4,4,34,2,16,36,6,29,29,2,28,27,9,22,
17,4,8,21,30,18,36,36,32,15,20,31,21,31,9,34,0,26,24,4,
21,26,28,24,30,31,0,30,29,16,9,1,0,34,17,23,7,25,11,15,
18,17,36,2,29,7,8,26,8,17,18,30,11,23,35,5,27,2,25,14,
8,25,19,13,2,3,27,24,0,33,3,35,34,32,27,32,1,32,3,35,
10,29,36,29,3,20,15,20,20,23,12,1,1,12,9,33,18,35,12,25,
24,11,28,19,15,35,16,26,7,1,28,2,19,7,12,9,3,0,14,25,
33,15,19,29,28,27,34,12,30,12,18,3,3,27,15,26,7,1,0,22
30,30,28,5,3,20,19,32,8,29,21,19,18,9,21,9,8,21,8,2

Look at the graph today. Can you see any trends?

[Firefox]

The last data set making 1000 spins this week.

That's 27 columns of 37 and 1 for luck. Thus in the total data set, the expected value of each number is 1000/37 = 27.03

16/03/2019. Cammegh High Scatter Low Profile Scallop. 4 Sec/Rev Rotor acw.  Cw B Release At 0.  18mm Heavy Teflon. 8 to 10 Ball Revolutions.11 to 14 second spin. Slightly tilted wheel. 2 Pin game. 2 dominant diamond dropzone. Raw Data. 200 spins.

CSV Data 200 spins. 10 x lines of 20.

20,19,30,17,35,29,19,9,34,19,15,29,7,6,12,4,10,17,21,25,
26,31,13,33,35,31,30,32,25,30,7,23,24,21,4,30,36,17,20,23,
34,12,3,27,34,21,32,33,31,30,12,32,25,6,7,34,11,29,18,28,
4,26,25,5,15,22,18,16,9,4,14,27,28,29,31,22,30,29,7,6,
28,35,0,26,0,28,4,19,26,31,4,1,24,35,1,3,36,15,34,4,
32,20,27,27,34,2,22,16,21,32,2,2,26,20,19,28,18,7,19,12,
19,29,27,13,22,13,0,31,16,3,34,12,24,7,28,18,16,15,2,22,
26,24,21,15,33,4,12,9,19,1,17,9,15,7,15,17,1,2,0,32,
4,17,3,10,19,29,23,15,12,12,21,3,4,2,34,19,13,31,18,35
14,23,3,18,9,13,19,18,3,1,27,6,9,33,24,32,35,25,14,4

2 clear winners this set of 200, see graph.

[Firefox]

Attached is graph amalgamated for all 1000 spins.

The mean is 1000/37 = 27.03

The standard deviation is SQRT(npq)  = (1/37 x 36/37 x1000)^1/2 = 5.13

The top numbers occur about 40 times which is over 2.5 standard deviations off the mean. This means on their own there is an 80% chance these frequencies would not occur by chance alone. However, as they form part of a group with increased frequency on both sides, we can be almost 100% certain there is a definite bias to these numbers.

If we played this dealer when he spins normally from zero clockwise, when he throws a 4 second anticlockwise rotor then betting on numbers 15,19,4 we will be betting on 1 show in 25 numbers. That's a massive 44% advantage over the house.

When he spins another rotor like 3.5 or 3 seconds then we will need to have examined what happens for this rotor speed. Or else we don't bet. We have a chance to count the rotor speed before we bet.

When he spins clockwise from elsewhere it's just an offset. For example if he spins from 5 red then the best numbers will be 16,33,1 and the numbers around them.

When he spins the ball anticlockwise we will have to get some data, but hopefully it will be a mirror eg 3,35, 12 would be the best spinning from zero or 23,8,30 spinning from red 5 etc. But you must check. Anti and Clockwise spins must be treated independently. The ball can hit different diamonds or come off in a different place so it must be assessed.

If he does a fast or a slow or a strange spin (chatter on the ball or backspin/sidespin) then you don't need to bet. Only  bet his normal speed spin and a rotor speed you like or know what will happen. If you are really observant, you can time the ball as well as the rotor to make sure they are in range, but you will have to do both and get the bets down if they are OK. In practice with a consistent dealer, I don't bother timing the ball. I just watch to see where he releases, and how, as well as time the rotor. If it's obviously an unusual fast/slow or tricky spin, then I don't even bother timing the rotor, I just don't bet that time.

In this case, I was dealing, but also predicting the fall off number using VB. It was generally in the range 8-33 (most common 10) which explains why there is a peak at 15 to 4. We are basically looking at a scatter plot clockwise from a 10 strike which on this wheel at this rotor speed gives a scatter peak at 24 pockets 2/3 of the wheel. There are lots of other scatter values too, as it is a very difficult wheel to read, but 24 pockets is a typical value at this rotor speed, and enough to give a scatter peak.

You don't have to use VB with dealer's signature, but it is useful to combine methods to confirm what is happening. At the very least, look at the fall off diamonds and scatter, even if you don't predict the fall off number.

In my view, dealer's signature and VB are quite closely related. It's the same spin but one is just an earlier prediction than the other. So many parameters are the same eg Rotor speed.

What you need to look for.

1. Consistent dealer ball speeds. Best are slower and shorter spins.

2. Consistent rotor speeds that you can work out what happens for each rotor speed. 0.1 or 0.2 second intervals between each speed.

3. Fall off diamonds which are consistent really help. A one pin game, two pin with overlapping scatter peaks, or 3 pin with 2 dominant ones. Watch what happens to the ball and where it starts and ends to see what consistency there may be. You may have to watch for 20 to 30 spins or more before playing.  Just make the minimum odd cover bet on an even chance every few spins.

4. If the wheel is all over the place, but the dealer is good, wait till that dealer goes to a better wheel later. If the wheel is good but the dealer random, wait for the dealer to change. Know the dealers and the wheels. You need the right dealer and the right wheel.

5. Treat C and anti-C spins separately. Watch the starting number, watch the rotor speed, watch the throw. You don't have to bet that spin if it isn't right.

[The General]

Firefox's data on a standard deviation graph






Ten pocket section graph of Firefox's data.  Again, a standard deviation graph.

[Firefox]

^ Many thanks for those!

Here it is obvious (I hope) we should be betting in the 0 to 17 sector for these conditions. There are various ways to smooth the graph including curve fitting, cubics and quadratics. But non of that sophistication is really needed for something like this.

A running mean of 3, n becomes {(n-1) + n + (n+1)}/3  and a couple of other tweaks also makes the scatter plot easier to visualise.

See attachment.

[Herby]

The wheel is slightly tilted to produce a 2 pin game

Hi,  short questions: what is a 2 pin game ?

tilted to produce or tilted, therefore it produces a 2 pin game ?
tnx

[Herby]

each rotor speed. 0.1 or 0.2 second intervals between each speed.

How you measure 0.1 second time intervals without beeing thrown out ?

[Herby]

2 diamond dominant dropzone.

You have a 2 diamond dominant dropzone, why don't you distinguish the results ?

[Herby]

Data 200 spins

200 spins from the same dealer ? Same day ?

Thanks in advance for the answers. 

[Firefox]

Hi,  short questions: what is a 2 pin game ?

tilted to produce or tilted, therefore it produces a 2 pin game ?
tnx

A 2 pin game is one with 2 dominant diamonds. The word pin coming from bagatelle, pinball, or ten pin bowling denoting a vertical object a ball strikes.

Tilted to produce I believe is correct usage, but certainly "tilted so as to produce", or "tilted in order to produce", are clearer.

All groups of 200 spins were from the same dealer (me) on the same day over 5 days on my wheel (see video). In releasing the ball, I was aiming for the same steady pace, 8 to 10 revolutions round the wheel.

I have a 2 diamond dropzone, but one doesn't know which diamond will be hit at the time of betting. Therefore for practical play purpose, we are concerned with the combined results from the drop zone.  I could have split the results but this does not have practical betting use, and complicates the data collection and presentation.

The scatter hits from each diamond will overlap. For example a ball which just clips the first diamond will have a longer scatter, while a similar ball which just clears the first diamond may drop to have a more solid hit on the second diamond which results in a shorter scatter.

I chose 2 diamonds because it represents something which will produce clearer results, while at the same time a practical thing you can find in the real world. 2 and 3 pin games are most common while 1 and 4 pin games, much rarer.

As for timing the rotor speed, I can time to 0.2s quite easily by counting ABCD1 ABCD2  etc. I believe more advanced visual players can time to 0.1s without using any device. You would have to contact The General to see how that might be done!

The impact between speeds depends on the length of the spin, and the rotor speed. For example the difference between a 4.0 and 4.2 second rotor is 0.44 pockets per second (0.2 pockets per second for 4.0 to 4.1 second rotors). Over a 12 second spin, these would be  2-5 pockets error respectively. That's not much when betting a 1/4  or 1/3 wheel sector. You are still going to be in the frame. But clearly as spin lengths and wheel speeds increase, the error is greater.

However, it is up to the player when he bets and what dealers or speeds he wants to play. One doesn't have to bet. This is one advantage the player has over the casino. He can decide when he wants to play, whereas the casino has to honour each legal spin it throws out there.