Sangoma-number divination

[barge]

Hi Patrick123

Amazing progress, thanks very much!
I have a problem with the evaluation in Sangoma. I loaded a short dbase of 20 draws and analysed five of them with one filter. The program returned good results, but not just from the five analysed. I had 184 to 289 hits in Tfilt. It obviously applies the success rate across an assumed dbase of the maximun 1176 draws and 18424 trips. At the bottom of the screen it shows Filtered Rows=1179 and Hits on Filtered=434.The % success is 36% which I like(!) but these figures are projections, and it would be good to know the direct effect of the filter on the most recent draws, as filter success is clearly going to vary as the ball frequencies change.
If the above has been discussed before, my apologies, I am going crosseyed and haven't the energy at the moment to go back over the thread!

 

[Patrick123]

Hi Barge, Thanks.
I agree the Filtered rows cannot be higher than the maximum of 49, 1176, or 18424 for the standard 49 Lotto. I'll try and track down that little bug, most probably a counter that I'm not resetting to zero.

I see what you saying now, no bug. The grid will display the values for each of the past 5 draws. I then combine the values of the number of hits vs number of filtered rows for the balls, pairs & trips and use this to calculate an overall percentage confidence for the filter.

 

[Springbok]

Sawubona Patrick

I cannot keep up with the updates

I think the span of drawings when evaluating filters could be important as this can effect the success of the filter. Experiment with N x 1, N X 2 and N x 3 where N is the span ie in the 6/49 game the span is 49. Different spans can pick up different pairs and trips.

I was wondering if a control could be added to the drawing analysis screen to speed up manual checking. For example I have 9 draadsitter filters(draadsitter9, draadsitter8 ...... draadsitter2), Each examines a certain skip range and no other. Each time I want to brink up a certain draadsitter filter I have to flick from one screen to load and back to the drawing analysis screen, Would it be possible to select the draadsitter from the control on the drawing analysis screen? This would apply to the inbetweener and opening batsmen.

 

[Patrick123]

Dumela Springbok,

Thanks, I'll have a look at the span. Funny I was thinking the same as it's becoming quite a schlep to jump from one screen to the next just to load a new column filter. Once again, watch this space

 

[barge]

Hi Patrick123

I think we were a little at cross purposes, my point was that the grid isn't showing the results of the past five draws. I have loaded a total of 20 draws into Sangoma, evaluated 5, but my T1 column for example, is showing: 307,272,303,289,314. Obviously my T hit column has a total hits figure which is higher.
What I don't have is the filter results for the 5 draws I am evaluating, just the overall percentage success result.

I used Rebal and Doublehit with a dbase of 20draws, and evaluation of 5.

 

[Patrick123]

Hi barge,

These are the values that no filter should give:

	B Rows	B Filt	B Hit	B Perc	P Rows	P Filt	P 1	P 2	P Hit	P Perc	T Rows	T Filt	T 1	T 2	T 3	T Hit	T Perc
Past 1	49	49	7	14	1176	1176	294	21	315	26	18424	18424	6027	882	35	6944	37

The figures you gave for the T1 column are possible as the trip couples {1,2,3} {1,2,4} .. {1,48,49} would all add a hit if ball 1 was drawn, of course if 4 was drawn as well, then the trip couple {1,2,4} would increment the T2 column instead. There are about 880+ couplings that would contain each number.

Also check that you have scrolled right, it might just be off your screen.

I hope I'm understanding your question correctly

Regards
Patrick

 

[Springbok]

Howzit Patrick

I got a gut feeling there are two types of medians. You have the stable median and the turbulent median. Pairs with stable medians have the same median no matter the span whereas the turbulent medians jump from one side to the other. Pairs that fail may have turbulent medians apart from the chance of not hitting. Any pair has a 70% to 90% chance of hitting within the criteria so you can have up to 30% of failure even with a stable median. I will look into this and see whatsup.

 

[barge]

Hi Patrick123

Thanks for the swift response
I understand how the Pairs and Trips hits are being presented, and maybe I'm asking too much in a sense. I would like to be able to say that eg. this filter hit (Pairs or Trips) say, 3 times in the past 5 draws. At the moment we know that overall (over month or years)the filter hit X number of times, but we don't know how this relates directly to the sample we are evaluating.
If I am missing the point here feel free to let me know

 

[Springbok]

Here is an example of how medians can change according to the span used. The pair {5;31} was caught by draadsitter7 on a span of 100 draws. It hit next draw. I then tested with a span of 50 draws and it did not appear. Notice below how the median changed according to the span used. Also notice the number of times the median was recorded on each span used. It appeared twice on the 50 span drawing and seven times on the 100 span drawing. So we can assume a 100 span drawing gives a more stable median whereas a 50 span drawing gives a more turbulent median.

* pair{5;31} Span 50 draws
------------------------
* Skips >>>>> 1 7 6 8 7 4 2 6 * Sorted Skips: 1 2 4 6 6 7 7 8 ** Skip Median: 6

* pair {5;31}* Span 100 draws
------------------------
* Skips >>>>> 1 7 6 8 7 4 2 6 1 8 1 0 7 1 0 0 0 0 0 3 1 0 1 1 10 * Sorted Skips: 0 0 0 0 0 0 0 1 1 1 1 1 1 1 2 3 4 6 6 7 7 7 8 8 10 ** Skip Median: 1

 

[Springbok]

Below you will see an extract from the UK 6/49. The target was the bottom line. Skipsitter6(Draadsitter6) identified {27,37} as the intended hit next draw. It's median was 2. Now a number hit in the 2nd draw later. This would indicate that the actual median on the 100 draw span on this occasion should have been 3 and was actually an inbetweener. So when testing look at this factors if the intended filter failed, maybe it should have been something else. A skipsitter could morph into an inbetweener and an inbetweener could morph into a skipsitter depending on the accuracy of the median. If none of the numbers fail to hit within any of the criteria then you are a victim of bad luck.The bad luck zone should be in the 10 to 30% probability range.

Anybody wanting to become familiar with these filters must spend much time testing them and asking questions. Why did it hit, if it didn't hit, did it hit within the range of another filter or was I a victim of just plain bad luck? Be aware some pairs have a sense of humour. A pair in the UK lottery(a skipsitter) failed on the skipsitter and inbetweener criteria. In the fourth draw later, the pair showed me the middle finger and both number hit together

12 23 25 26 34 49
3 6 15 23 28 42
4 18 22 27 44 48
3 18 21 24 36 38

 

[Springbok]

Here is another example from the UK lottery. Skipsitter5 identified {16,26} with a median of 1. The target was the bottom line. It hit the draw after. So the median of 1 was incorrect and should have been 2, so it was actually an inbetweener.


15 27 35 37 38 39
3 9 10 12 46 47
16 22 28 30 33 35
3 13 22 32 44 46

 

[Patrick123]

Hi Springbok,
In many ways, this indirectly relates to an earlier conversation that we had on this thread, namely the optimum sample size. Too small, and we run the risk of a turbulent median. Too high?? are we then only wasting a few cpu cycles, waiting for the program to analyse the information, or are we smoothing out any potential ripples?

 

[Springbok]

hi Patrick

At the moment I test with sample of 100. It seems to be on the border of the stable and turbulent zone and seems to give good results.

Here is an example of a filter morphing and then remorphing itself
My old friend {16, 26} median 2, 100 draw span(as usual) pops up again this time 5 draw down testing area in the uk game.

My trusty Draadsitter5 identifies the pair. Now the median should have been 3 and the pair should have been an inbetweener. It fails the next draw and would sit on the median3. So it then morphed into a draadsitter and hit in the third draw.

A span of 200 draws plus seems to give stable medians. I will look at these more closely in the future to see whatsup. It would seem so far that the lottery revolves around the Battle of the Medians. Whoever wins the battle, will be like William the Conqueror at the Battle of Hastings.

5 9 16 25 34 38 42
5 6 14 26 30 32 37
9 12 15 25 37 39 48
2 19 22 24 28 29 49

 

[Springbok]

Patrick, I downloaded the UK Daily Play as a test bed. It is a 7/27 game played once a day. A quick look at it through the program and fiddling with the filters showed me how easily and quickly you can park 4 pairs in 4 of the 7 slots. Link below if you want to play with it.


http://www.national-lottery.co.uk/player/files/DailyPlay.csv

 

[Excellence]

Good day gentlemen

Very good I may say. Thanks Patrick for the great add-ins.

I would like to propose something relating the medians and the span. I think that, thanks to theory of probability we can cheat. Indeed span is a sensible subject. A very small number of draws analyzed will, in most of the cases indicate a false value of the median. When I say small I mean about 20, 30. If we increase the number of draws analyzed to, let’s say 50 the medians might start to indicate something.

For example. I analyzed whit Winnalotto 100 draws. First time I skipped the last 50 draws and analyzed the remaining of 50. I obtained the next result: 27 numbers (not pairs or trips) had a median above the theoretical value (5 for draws, 4 for skips). I also say that all values apply for 7 / 49, but it is obvious they can be calculated and adjusted for any other game. My question was: What happened to this numbers in the next 50 draws? Well here it is: in the next 50 draws we had 26 numbers that exceeded their theoretical median, 59,26% of them were also exceeding their median in the previous 50 draws set. 10 numbers that in the first set where under the median transformed into numbers above median. I will name them from now on Above Group and Under Group.

What is the relation between median and numbers? I think the frequency of the number. As bigger the frequency, as normally the median tends to the value of the theoretical value. Now we will go a litter deeper in the theory.

What should be the frequency of a number in 50 draws? 7. But in 100 draws? 14, obviously. How many times that happens? The Gauss Bell (? Don’t know if it correct in english) could provide an answer. In 95% of the cases the frequency will be between 3 and 11 for 50 draws. We also know that in 68% the frequency will be between 5 and 9. For the 100 draws margin we have the next values: 11-17 for 68% and 8-20 for 95%. Hope the calculations are the correct ones.

How can we use the this? Well, if a number has a frequency for the first set (50 draws) of 10 and is an Under will become an Above in the next 50? If it is an Above, then the skips will be larger. I said in the beginning that we can cheat. Indeed we can. What if we would select the last 90 draws, skip the last 40 and perform the above analyze for the last 50. After we analyze the next 40 and maybe we can make some estimations for the last 10 draws remaining.

Patrick, I know you have a lot things to do but can I ask for another column that would present the frequency of a number? It will fasten the analyze a lot. It will also be nice that in the criteria definition we could use statistical indicators like max, min and median?

Hope that I made myself clear whit the things presented above. Right now is a bit confusing to me to but I hope and think that I can find a relation between the frequency and the median. I would appreciate your opinions gentlemen about this.

 

[black prince]

Using the Wheel tab?

Hi Everyone, Great work!!:finger: I'm reading many posts and becoming familiar
with N+X>=H, Detailed, Rebalance, and SkipSitter filters. I have a quick question
about the Wheel tab. After setting the numbers of balls, numbers per ticket,
4 out of 5 picks as is my case, and number of tickets, what should I see
with percentages to feel the iteration should stop. I have 16 filtered numbers,
for pick5 (5/30),4 out of 5, with 10 tickets. What is the optimal percentages
I should look for. So far, only 4% and 5% show with some numbers for draw 1,
2, 3, 4, 5 and random.

Thanks

BP

 

[barge]

Hi All

A small tip for those who may not be aware of it. (I have no connection with it) The lottery support site Merseyworld.com has the best results setup for the UK and Euro lottos. It also has the results for all the lottos downloadable in text,comma or pre-formatted form,and links to results for most lotteries around the world.
There is also a wheeling (Perms) section which has a few smart features, and there are lot of possibly useful stats, percentages etc.

 

[Springbok]

Hi Barge

Had a look at the Merseyworld site. I see the Euromillions are offering £88 million prize 28 Septmber. Now that is serious moola. It is equivalent to R 1, 232,000,000 which translates into over a billion rands. Now that buys a hell of a lot of krugerrands. It is better than owning a goldmine on the Witwatersrand. I downloaded the previous 166 results to see if it was amenable to our techniques. Well, good old draaditter6, I think, hit in the very next draw in the last 4 draw test. So that's a good start. There is a lack of draws so I will generate a thousand or so synthetic draws and add the real draws on top-this could help the historical highs. But I have got to find a reasonable way to forecast the wretched star numbers. To hit the five numbers and miss the star numbers would have me sectioned under the Mental Health Act

 

[blitzed]

hiya Springbok,

I whipped up probabilities for 2of9 star numbers over 28 draws...perhaps you could use this distro chart as a basis for a star filter:

0x 0.09%
1x 0.70%
2x 2.71%
3x 6.72%
4x 11.99%
5x 16.45%
6x 18.01%
7x 16.18%
8x 12.13%
9x 7.70%
10x 4.18%
11x 1.95%
12x 0.79%
13x 0.28%
14x 0.09%

Alternatively you could track each star slot with the 28draw 1of9 star chart:
0x 3.70%
1x 12.94%
2x 21.83%
3x 23.65%
4x 18.48%
5x 11.09%
6x 5.31%
7x 2.09%
8x 0.68%
9x 0.19%

cheers!
blitzed

 

[barge]

Hi Springbok and Blitzed

Serious moollah indeed, personally, I would be willing to share
I have been thinking about the 2 or 3 hit guaranteed wheels on Mersyworld. I used the 19 ticket wheel in a small syndicate for a year or so. Not too successfully, as I don't think these wheels are very useful for more than their guarantee, we got about half of our money back. The interesting part is that in most draws, there were at least three lines with 2 numbers amongst the 19,presumably the 163 ticket wheel works the same way. Merseyworld has a neat wheel populator, which allows you to include up to 3 selections into a randomly generated guaranteed wheel. if we could use sangoma forecasts to target some of the lines in eg a 3 hit guaranteed wheel, we might just improve the odds a bit.................?