Coding request in exchange of 1 month unlimited use of the tool!

[Bayes]

Hi Richard,

Here are the results of another test using the first 20,000 spins in the file you linked to in your above post. The frequencies of the gaps are listed at the end of the file. In order to make a true comparison, and to find out whether your method of picking dozens has any merit, I will generate some results using a RANDOM bet selection for the same number of bets. I'll post this later today.

[Richard]

Hi Bayes,

Thanks for the further test. I found the 17 gap and examined it (no n/b), but somehow
I missed finding the 18 gap, which shows at the end of the results?

The gap of course was bound to increase and just hoping 18 holds

I will have to try and find the 18 gap a bit later on.

[Bayes]

Richard, in the test I uploaded the maximum gap was 17:

Gap   Frequency
-----------------
0         1561
1         1070
2          715
3          486
4          319
5          207
6          132
7           86
8           57
9           40
10           23
11           14
12            8
13            5
14            2
15            1
16            1
17            1

Tip: if you want to find a gap length, just use the "Find" or "Search" feature in a text editor. e.g to find the gap of 17, just enter "+17" in the search field.

Running the same number of spins using a random bet selection, I got the following frequencies:

Gap     Frequency
------------------
0            1565
1            1047
2             698
3             463
4             318
5             221
6             144
7             101
8              60
9              37
10              23
11              19
12              14
13               6
14               5
15               3
16               2
17               1
18               1

As you can see, the longest gap in this case was 18, but more significantly, compare the total number of occurrences beyond 10. Using your bet selection the total from 11 to 17 was 32, but using the random bet selection it was 51 -  an increase of 59%. This suggests that there is some merit in your method, at least if you look at the number of longer losing streaks, but I'll run the full 700,000 spins to get the bigger picture.

[Richard]

Bayes,

It definitely shows 18 gaps at the end of the result?

It's certainly interesting to compare the two results and I'm eagerly looking forward
to the results of the 700,000 spin test, with trepidation naturally

[Bayes]

Bad news I'm afraid Richard. 

Here are the results from the 700,000 spins. First, here are the frequencies for your bet selection:

Gap   Frequency
------------------
0        54472
1        36837
2        24877
3        16851
4        11334
5         7580
6         5206
7         3521
8         2410
9         1662
10         1152
11          757
12          518
13          363
14          240
15          160
16           98
17           74
18           56
19           44
20           25
21           15
22            9
23            6
24            6
25            3
26            2
27            1
28            1

And here are those for random bet selection:

Gap    Frequency
---------------
0       54497
1       36813
2       24780
3       16898
4       11403
5        7693
6        5183
7        3510
8        2402
9        1622
10        1108
11         751
12         496
13         340
14         240
15         170
16         109
17          77
18          56
19          37
20          26
21          19
22          14
23          11
24          10
25           6
26           4
27           1
28           1
29           1
30           1
31           1

As you can see, there is very little difference, certainly nothing of statistical significance. Doing the same calculations as before for those gaps > 10, the total for your BS is 2,378 and for random it's 2,371 - a very close match. The results from the previous much shorter trial just reflected the fact that "anything can happen in the short term", and with a much bigger sample any apparent advantage was "ironed out". 

[Richard]

Well Bayes, as you say certainly bad news .

You may recall in a previous post I mentioned I will have some questions concerning the
big test.

What I need if possible is to examine in detail all those gaps from 18 and above. Obviously
I have missed something when working this out! I don't give up at the first hurdle. It took
me years to work out Signum.

It's a pity a 18 gap or above didn't show in the early tests.

[Bayes]

No problem Richard, I've uploaded the results file.

[Richard]

Bayes, thanks I've downloaded the results file.

I must admit I was surprised it went as far 28 gaps,so examined this first of all.

I thought the zero problem was nailed, (please see replies 28,29 & 30) the following
occurred :

[table=,]
spins,col1,col2,col3,result
35,,,,
12,,,,
1,2D,1D*,3D^,+5
---,,,,
31,,,,
19,,,,
0,3D,1D,2D,+6
---,,,,
28,
7,
4,2D,1D*^,3D,+0
---,,,,
[/table]

The program recorded +7 ?

I'll take a look at gap 27 next.

[Richard]

Used the text editor to find 27 and it could not find it.

It found 26, so will examine it next.

[Richard]

It's ok I found gap 27

Found a gap 22, with the zero problem :

[table=,]
spins,col1,col2,col3,result
20,,,,
33,,,,
6,3D^,1D*,2D,+6
---,,,,
8,,,,
4,,,,
32,1D,1D,-,n/b
---,,,,
32,,,,
22,,,,
4,,,,
35,,,,
0,1D,2D,3D,+7
---,,,,
35,,,,
8,,,,
16,,,,
5,,,,
27,1D,3D*^,2D,+0
---,,,,
[/table]

[Richard]

Hi Bayes,

Ok, was tired last night looking at this and had a fresh look this morning.

Gap 28 as mentioned above has a zero fault.

I thought I found gap 27, but it turned out to be part of gap 28. The editor highlights
the +27 and I missed the fact that it's part of +28. I searched up and down, but could
not find any other +27?

Found only one gap 26 and this appears to check out ok. I have an idea to solve the
gaps above 18, but of course it will take a fair bit of work, as I need to apply it all of
the results above +18 to know if is effective.

[Bayes]

Those pesky zeros! 

Ok, I can see the problem, I'll take a look at the code. There is only 1 gap of 28 and 1 of 27, 2 of 26...

[Bayes]

Ok, I fixed the problem and re-ran the program. The longest gap is now 26, file attached.

[Richard]

Ok, thanks, I've downloaded the amended results

There still appears to be a problem with the totals at the end of the results. It is still
showing 1 27 and I can only find 1 26 and not 2 26 as shown on the results.

I don't know if any of the other totals are affected. It seems strange the program should
count a total that does not exist?

Anyway I shall press on with finding a solution to the gaps above +18.

Thanks again for the excellent program, much appreciated.

[Richard]

Hi Bayes,

I'm having difficulty if reconciling the discrepancy in the total gaps.

I put in the editor find +22 and get the following result:

22    3
23    2
24    2
25    1
26    1


At the end of the results it's showing the following:

22   14
23     8
24     6
25     4
26     2
27     1

I would be grateful if you could look at this, as I presume the earlier gaps will be
affected as well.