Sangoma-number divination

[Patrick123]

This is looking more & more promising. I'm skeptical about trusting the SA draws, especially when the numbers are announced before the draw took place(they said there was a problem, and the video feed had a delay) or just before SA lottery stopped, there was an statistically unusually high amount of jackpot winners the same draw. Would Sangoma be smart enough to do the predictions if the draws are not quite as random as what they are meant to be?

 

[Springbok]

Hi Patrick
Interesting stuff on the SA draws. Numbers announced before the draw took place This could only happen in Africa. President Mugabe actually won the Zimbo lottery. A president buys tickets in the lottery?!!!?? Could crookery be detected by sangoma? I dunno. Depends which numbers were fiddled. I think you would have to be careful in fixing the numbers. But knowing these people, they are greedy but not very intelligent. Will have a look. Can you tell me which draws are suspect and the dates they were held? Was it a one off event or did it happen frequently. Are there witnesses at the draws? Are there real accountants present or just "accountants"(nod, nod,wink, wink)?

 

[Excellence]

Good day all. Quite funny. I had Mdied installed in my computer for a few weeks now but I did not learned how to use it. I decided yesterday to learn and stumble upon the same problem: do random generated draws could replace or at least be added to real draws?
In the tutorial for the program Mr. Saliu says that the filters work for lotto games whit at least 200 000 draws. I was wondering how do you use it? Generate random draws or base the analysis on the real draws?
In my opinion, if we would use random draws, I do not see any problem in using other lotteries past draws.

 

[Patrick123]

Hi Springbok,
Surprise, surprise, it was 31 Dec 2003.
Here is a link for you, you will also find other 'interesting' items in the press release link.
http://www.nlb.org.za/pressarticle.asp?id=53
and
http://www.nlb.org.za/pressarticle.asp?id=65
Pretty indignant this lot

With regards to the the unusually high winners, I'll try & track that down, I know it was about a month before the SA Lottery stopped, but will check.

 

[Patrick123]

Got it, here is the other link relating to SA lottery.
http://www.int.iol.co.za/index.php?set_id=1&click_id=196&art_id=nw20070401150446672C835147

"Accidentally, the draw of the 3rd March was also the last draw before Uthingo's High Court application, against the awarding of the contract by Minister Mpahlwa to a company with various high-ranking ANC members as owners, was to be heard.

Another link:
http://www.freakonomics.com/blog/2007/06/29/is-the-south-african-lottey-rigged-a-hands-on-exercise-for-bored-blog-readers/

 

[Springbok]

Hi Patrick
Thanks for the links. Had a look at 3 March results. Above average winners for all the divisions. Also ticket sales well above the norm. What was so special about those numbers. Were the sangomas on the ball(no pun intended)?. What was so special with the 3 March draw? I think 9 winners in division 1 could have happened. In the UK lottery there was an occasion when 133 people won the jackpot. Link below will show you all winners and prizes.The average is 3 winners per jackpot in UK. I assume it is for the SA lottery. If this is so then that is strange. Average tickets sold for the SA lottery is around 13 million. Uk average is around 73 million. So there should be 16.5 winners on average in the UK if you base it on the SA stats. Are the SA stats on an average of 3 correct?


http://lottery.merseyworld.com/cgi-bin/lottery?days=2&Prizes=1&Sort=0&year=0&display=NoTables

 

[Springbok]

Anyway Patrick, I am now going to try and break sangoma by random testing to try and prove it is rubbish. Attempts must be made to smash any theory to pieces. I am pretty sure it will, like Castle lager, stand the test of time. Will post results later this afternoon. The testing is for one number pair based on n=H. Of course n will be the highest skipping pair. Will use the 4 real draws on top of the simulation and the next consecutive four for testing. I will dive in and out of the 49s at random. your software will speed this up. If I used my own files it would take many times longer.

 

[Patrick123]

Yep, on closer analysis, they had a guaranteed jackpot of 35M, coupled with the first draw of the month after payday, made for the greater sales. As for the numbers, must have been dem bones
It seems more like a storm in a teacup from the Freedom Front political party, as we've had up to 33 winners (15 March 2003). The average number of JP winners is 1, removing roll-overs it sits at 1.9.
Spreadsheet at http://www.nationallottery.co.za/lotto/clientfiles/lotto/lottoresults.csv

 

[Springbok]

Created another 7 simulated files with up to 1500 combinations per file. Copied 7 8 combination chunks from the real drawing. Bottom 4 added to top of the sim file and top four used for testing.
Here are the results:-
77 pairs @ 100%
6 failed in the next 4 drawings.(8%)
92% hit within 4 drawings.
Top rated pairs hit 3 times in the 7 files(43% strike rate)
36% hit in draw 1
8% in draw 2
38% in draw 3
10% in draw 4
8% failure

The real draw chunks were taken randomly from the real draw history.

 

[Springbok]

Problems with your software

Hi Patrick. I downloaded the latest incarnation of your software and ran some tests using the last 4 real draws as a test(included todays teatime) using the real draw-no simulations. Got some strange results. Ran the same test on some previous software which gave different results. Have not checked yet with my own software.

 

[Springbok]

The pairing 35, 42 is given different results according to which version of the software is being used. In one it is awarded 100% and the other no awards at all.

 

[Patrick123]

Hi Springbok,
I was not available yesterday, and will only have a chance to check the source tomorrow morning (Monday). I'll give you feedback as soon as possible.

 

[Springbok]

Hi Patrick
Don't worry about it. If it happens again I will tell you precisely what was analysed and where the "error" was. I lost the stats and have no idea what group I was looking at. Thinking about Sangoma, you could have a career change and become one yourself. Have a stall in a market place and crouch behind a laptop disguised as a large bone
When I transferred the program to my old laptop I realised how memory intensive it is. It has to create virtual memory after the analysis. The desktop handles it fine with its huge memory although the fan has to kick in. So far the program has performed outstandingly. It does the same thing as my home grown stuff but so much quicker.I still think my stuff has an edge in some respects but not in speed. The thing now is to extend the forecast to more pairings in the same draw. I will be thinking about this in the forthcoming week. Also I will be looking at the trips in the program and how to use them in conjunction with the pairings. As they say" a luta continua"(with the lottery that is)

 

[Patrick123]

Hi Springbok,
I'm back again, I'll give the program a once over, just to be sure.. Yep, the program's very memory intensive. It keeps stats on all the drawings for the balls, pairs & trips, that's why I thought I should use a DB, but most computers nowadays should be able to handle it. I can envisage myself in Zulu adobe, chanting away there, with the camouflaged laptop strategically placed as I predict the numbers

 

[Springbok]

Hi Patrick
Had a good Sunday 2 bets that day. Both on highest rated and both hit first draw. I am still back testing and I came across a sequence with 2 back to back failures on highest rates. They were a 4, 5 pair which hit after 35 skips. It was rated highest at 23 skips. So there are occasionaly rogue pairs. I am thinking of stretching the range of analysis from 1000 to 1500. Surely this will catch any rogues out there.

 

[Patrick123]

Hi Springbok,
Yes, the larger the sample, the higher the possibility that we pick up if the pair had erratic behavior in the past, thus it should reduce the % rating for that pair.

 

[Springbok]

Just been running an analysis for the lunchtime draw. End of last week we had a glut of 100% candidates. Now we have a lack, Running two tests ,one with 1500 history and one with 1000 history reveals zero 100% candidates. There is {2,44} rated at 80% with a high of 18 and current skip of 15. Then we have {2, 41} rated at 50% with a high 20 and current skip 19. Both the stats are the same whether it is the 1500 or 1000 history. Based on the above knowledge I would prefer the 50% chance as it has never exceeded 20 skips in 1500 draws over tthe 80% chance. We can cut the history until we start getting 100% candidates. if we got 100% candidates on a reduced history, would they be more valid than any of the above. I ran another test with a history of 700. {2,44} now goes to 100% with a historic high of 15 over 700 draws. {2, 41} ramains at 50%. So we have two different perspectives on these two pairs Based on the 1000+ history we would expect {2,41} to be the safer of the two. But in the 700 history, {2, 44} is the safer. So the wise choice would be to bet on all 3 numbers{2,41,44}. Again we could rephrase the question will the next draw follow the 1000+ history or the 700 history or will it follow its own history

 

[Springbok]

To confuse the issue a liitle bit further we can take another perspective. Three numbers appear to be the ones we might need(hopefully). Is it possible to individually rate each of these numbers. This can be done on Patrick's software. I did it on MDIED. I used a 200 draw hstory. For real number 2 its sorted skips were 19, 25. Twice in the last 200 draws it hit a skip of 19. It turned once out of the 2 times. Chance of hitting lunchtime 50%. Real number 41 is at the moment sitting on skip 27. This is its historical high(200 draws). 100% of hitting lunchtime. Real number 44 sitting on skip 15. Sorted skips 15,17,28. It has a 33% probability. The actual forecasts depend on the draw history. The 2 strongest numbers according to this are 2 and 41. Perhaps if you were betting on all 3 numbers you would put equal bets on the 2 strongest and the least on the weakest. I would go for 2, 41,

 

[Springbok]

Well none of them came in despite all the analysis. Had I cast my mind back to yesterday where we had the top rated hitting twice in a row, we have to realise that they do not always do this especially three times in a row. Hey Patrick, I have been looking at the single ball predictions on the program. It can deliver spectacular results but there are also periods of drought when a ball at 100% does not hit draw after draw(here a draw = 4 draws). AN excellent example of this was no 5 from 20 0r 30 draws ago when I analysed a certain section , This damned ball was top rated at 23 skips and went on for another 12 or so before revealing itself. Single ball betting can be dodgy because of these rogues.

 

[barge]

Springbok

I have been following your and Patricks excellent efforts over the past few weeks. I would like to test a few databases to help the search along, I have used Ion Salius software, so am familiar with MDIE etc. I have a couple of minor queries regarding Winnalotto. Early in the thread Patrick mentioned the progs ability to generate up to fifty pairs files..I can't see how this is done (if it is still part of the software)?
I'm also not sure about the function of the mode, or rebalance. Any info from yourself or Patrick would be appreciated, if you can find a minute!