Winnalotto-Sangoma 2

[black prince]

Patrick123 and Springbok

Hi Patrick and SpringBok,

One or two numbers return from prior winning draws on average 6 out of
10 drawings. I created a filter 'Prior Draw' which assumes a number
from a prior draws will only repeat twice and before it stops. I've
learned that some numbers like 4 historically can repeat up to 3 or more
times in a row before stopping. Is there a way to check repeats historically
using Winnalotto for numbers with a history of consecutive repeats. When
they appear I could may an execption for them in the filter.

Thanks

-BP

 

[Springbok]

Helllo Black Price

The problem with finding a number from the previous draw is that there is only one previous draw to look at. You need a number of simultaneous previous draws. This can be easily achieved. Generate 10 or more random draws for your game. Now look at your real draws and take one random combination. See if a number from the last real draw occurred in that combination. If not look at your second real draw and see if a number occurred in that random combination. Carry on doing this until a number hits and write down next to the simulated combo when it hit in the real draw. Do this until you have 2 or 3 simulated combos that have not hit in the last 2 or 3 real draws. These combos should contain the winning number in the next real draw. Test this on paper first and you will see what I mean.
This is based on the way Saliu's ALL filter works in the lottery with the various layers.

 

[black prince]

Springbok

Originally Posted by Springbok
Helllo Black Price
The problem with finding a number from the previous draw is that there is only one previous draw to look at. You need a number of simultaneous previous draws. This can be easily achieved. Generate 10 or more random draws for your game. Now look at your real draws and take one random combination. See if a number from the last real draw occurred in that combination. If not look at your second real draw and see if a number occurred in that random combination. Carry on doing this until a number hits and write down next to the simulated combo when it hit in the real draw. Do this until you have 2 or 3 simulated combos that have not hit in the last 2 or 3 real draws. These combos should contain the winning number in the next real draw. Test this on paper first and you will see what I mean.
This is based on the way Saliu's ALL filter works in the lottery with the various layers.

Springbok,

I tested this with two prior winning real draws and the results were GREAT!!!
I reduced the simulated draws to 2 or 3 and the results not only showed the
prior number, but also had all winning numbers for the future draw in 10 to 12
numbers. I've programmed this process to make testing easier. I will backtest some
more to get a feel of how this works. Sometimes I only get 1 or 2 prior numbers and
other times all 5 prior numbers appear. I removed the numbers that don't appear in
'BPMedianDue' except for prior numbers reported.

Thanks

-BP

 

[black prince]

Springbok

Springbok,

...Real Draws............Random Draws....

05,13,23,26,28 ....... 10,14,24,26,30....
04,06,15,21,26 ....... 03,17,18,20,27.... <-- no matching numbers
06,16,22,26,30 ....... 04,15,16,25,29....
............................ 06,07,09,11,28....
............................ 08,13,21,22,23....
............................ 01,02,05,12,19....
............................ 01,10,20,25,29.... <-- no matching numbers
............................ 11,13,16,18,23....
............................ 03,04,09,15,20....
............................ 02,12,24,27,28....

03,17,18,20,27 and 01,10,20,12,19 have no matching numbers and
should have numbers coming in Monday's Treasure Hunt drawing 6/23/08.
Let's see what happens.

-BP

 

[Springbok]

G'day Patrick

How is it going with filterology? What I find with these filters is that they work fine for a while then fail and then work again. Something like a irregular sine wave. The one I suggested to you a couple of posts ago is brilliant for a few draws then peters out and starts working again. Here is another filter(7/49 game)-
skip(0) = 0
and
skip(1) >= 5
and
skip(2) >= 5
and
skip(3) >= 5

Here is the logic behind it. A pair has around a 70% chance of hitting in 4 draws and a 30% chance of not hitting. So we find a situation where the 30% probability has hit in the previous 3 skip cycles thus in the next it should hit within 4 draws according to probability. When it works it works a treat. Anyone wanting to use it will have to adjust the parameters for their particular game.

 

[struxo]

to black prince

Hi, you can try this too:

1 6 15 16 25 30
2 7 14 17 24 29
3 8 13 18 23 28
4 9 12 19 22 27
5 10 11 20 21 26

then check real draws with these horizontal and vertical. Same numbers from horizontal and vertical lines tend to be drawn in next draw.

regards
struxo

 

[struxo]

hi all

Another test i made for lotto 7/35:

load every 1 line
load every 2 line
load every 3 line etc.

after every load i take pair with highest last drawn. I made this till i get 10 numbers.
Interesting results

regards
struxo

 

[black prince]

Springbok

Originally Posted by Black Prince.
Springbok,
...Real Draws............Random Draws....

05,13,23,26,28 ....... 10,14,24,26,30....
04,06,15,21,26 ....... 03,17,18,20,27.... <-- no matching numbers
06,16,22,26,30 ....... 04,15,16,25,29....
............................ 06,07,09,11,28....
............................ 08,13,21,22,23....
............................ 01,02,05,12,19....
............................ 01,10,20,25,29.... <-- no matching numbers
............................ 11,13,16,18,23....
............................ 03,04,09,15,20....
............................ 02,12,24,27,28....

03,17,18,20,27 and 01,10,20,12,19 have no matching numbers and
should have numbers coming in Monday's Treasure Hunt drawing 6/23/08.
Let's see what happens.

-BP

Winning Numbers for Treasure Hunt 6/23/08: 4, 14, 15, 19, 29

Well only 29 matched the numbers in 03,17,18,20,27 and 01,10,20,25,29.

------------------------------------------------------------------------

With 'Springbok2', 'N-x>=H', 'Median Due' and 'Rebal Perc' filters the
numbers picked were:

'Springbok2' --> 19,25
'N-x>=H', 'Median Due' and 'Rebal Perc' --> 14,25,25 (Balls)
.................................................. --> 3:25,18:25,18:29 (Pairs)
.................................................. --> 7:10:25,4:10:18,10:14:20 (Trips)

4,14,19,29 matched 4 of the winning numbers. Well back to the drawing
board for more backtesting.

-BP

 

[Springbok]

Bonjour Taaroa

I think the best way and the simplest way for consistency in the 7/49 game of the 49s is using trips and then pairs.It also has the advantage of being mechanical. I use omega trips(1,2,3;4,5,6;7,8,9 ... 46,47,48). Load them and run over the probability scores. Pick the highest score, delete the file and type in the trip. Hit pairs and pick the one with the highest rating. Bet for the expected 4 draw margin. It is not infallible but in the long run it appears to be consistent. I will run a test over the last 100 draws and give the results in due course. I will run it in parallel with another trip for interest.

 

[Patrick123]

Hi Guys,

At this stage there is no Linux version, but I'll look at the possibility of porting WinnaLotto to a FreePascal where I'd then be able to cross-compile it for Linux.

Springbok, this new filtering method of yours, has either opened up a can of worms but more like a can of butterflies still busy testing, but this could possibility give us those elusive in-betweeners and Opening Batsmen. More on this to follow.

I have noticed this waxing and waning of filters, and if we can predict that we would be in the pound seats. At the moment I try and rely on an alternate filter to back up the main filter in one way or another.

Hi Struxo,

When you load these lines,

load every 1 line
load every 2 line
load every 3 line etc.

When you load 'every 2 line' - Skip First 1 Line.
for 3 lines - Skip first 2 lines, etc.

The reason I say this, is so that everything remains synchronised for the next draw. I have managed to produce some excellent results working with these multiple levels

Regards
Patrick

 

[Springbok]

Kunjani Patrick

There are definite waves. Some time this week I will tansfer some of the data to the RPG programme where they will orbit around an attractor and see whatsup. You know anything about Fourier cycles? Maybe the line skip function might help detect waves?

The lottery is a can of worms. All these trips 'n pairs all babbling away at us. Getting a Phd is a cake walk compared to our little excursions into the realm of randomness.

 

[Patrick123]

Ntate Springbok,

Unfortunately I'm not clued up with Fourier cycles. I'm more under the impression that you would get a bunch of waves interfering with each other analogous to a handful of pebbles been thrown simultaneously into a pond.

 

[black prince]

Struxo

Originally Posted by Struxo
hi all
Another test i made for lotto 7/35:

load every 1 line
load every 2 line
load every 3 line etc.

after every load i take pair with highest last drawn. I made this till i get 10 numbers.
Interesting results

regards
struxo

I forgot about test drawings before the real lottery drawing. I'll check this
out and see what occurs from prior winning draws.

-BP

 

[Excellence]

Hi Springbok

How is it going with filterology? What I find with these filters is that they work fine for a while then fail and then work again. Something like a irregular sine wave. The one I suggested to you a couple of posts ago is brilliant for a few draws then peters out and starts working again. Here is another filter(7/49 game)-
skip(0) = 0
and
skip(1) >= 5
and
skip(2) >= 5
and
skip(3) >= 5

Here is the logic behind it. A pair has around a 70% chance of hitting in 4 draws and a 30% chance of not hitting. So we find a situation where the 30% probability has hit in the previous 3 skip cycles thus in the next it should hit within 4 draws according to probability. When it works it works a treat. Anyone wanting to use it will have to adjust the parameters for their particular game.

The problem with this filter is that it’s not a closed cycle. Let me show how your strategy doesn’t beat the random expectations, but in fact, matches them.

In a 49 numbers game there are 1176 pairs. The probability that a pair will have a skip above 3 (4 draws) is 0,3 (0,2873 in fact but let’s say 0,3). 0,3 ^ 4 is 0,0081. How many pairs we should expect to have 4 skip cycles above 3? 1176 * 0,0081 which is 9,5256. So around 9 or 10 pairs. BUT! How many pairs should we expect to have only 3 skip cycles above 3? (0,3 ^ 3) * 1176 which is 31,752. So how many pairs will end up having just 3 skip cycles above 3? 31,752 – 9,5256 = 22,2264. AND 22,2264 / 31,752 = 0,7. If I choose a random pair and play it for 4 draws should win in 70% of the cases (71% to be precise).

Closed cycles. Let me take the next example, pair Y has the next skips: (first is the most recent)
0, 9, 5, 2, 6, 8, 5, 3

As we can see 5, 8, 6 are consecutive skips above 3. The probability of that to happen’ is 0,3 ^ 3 = 0,027. The skip cycle is ended because we have a skip of 2 after 6! Our last 2 ended skips are also above 3. What is the probability that we will have another 3 consecutive skips above 3? 0,027 ^ 2. So we should expect around 0,857304 pairs to hit the third consecutive skip above 3. I still have serious doubts that even this strategy could beat random odds, I don know how many pairs should have a skip cycle of 3 skips above 3, and the next skip cycle only 2, but the result could be the same as in the previous strategy. Even so, maybe random play can bring u ahead.

I also read another strategy presented by you that was referring to skips also, the one when u were choosing the pairs that had more skips of 5, 6 then 7, 8. It is absolutely normal that a pair has more skips of 5, 6 rather then 7, 8. The probability of a pair to have a skip of 5 or 6 is 9,76% while the one that a pair will have a skip of 7 or 8 is 5,23%. I would analyze a smaller sample of draws and play only those pairs that have the current skip of 5 and that have less skips of 5, 6 then 7, 8. The number of draws should be relatively low for that to happen.

All the best!

 

[Springbok]

Hello Excellemce

I will study your post carefully and test your ideas. In the meantime look at the filter below. I have 3 of them going from 20, 25 ,30 skip cycles. They start at current skip of 2 and seek out x skip cycles of 6 or less than 6. This filter can perform flawlessly. In other words the highest rated pair hits within 4 draws and normally in the next 2 draws. Not only that all the pairs fall within the 4 draw parameter. This totally thrashes random play.

skip(0) = 2
and
skip(1) <= 6
and
skip(2) <= 6
and
skip(3) <= 6
and
skip(4) <= 6
and
skip(5) <= 6
and
skip(6) <= 6
and
skip(7) <= 6
and
skip(8) <= 6
and
skip(9) <= 6
and
skip(10) <= 6
and
skip(11) <= 6
and
skip(12) <= 6
and
skip(13) <= 6
and
skip(14) <= 6
and
skip(15) <= 6
and
skip(16) <= 6
and
skip(17) <= 6
and
skip(18) <= 6
and
skip(19) <= 6
and
skip(20) <= 6
and
skip(21) <= 6
and
skip(22) <= 6
and
skip(23) <= 6
and
skip(24) <= 6
and
skip(25) <= 6
and
skip(26) <= 6
and
skip(27) <= 6
and
skip(28) <= 6
and
skip(29) <= 6
and
skip(30) <= 6
and
skip(31) <= 6

 

[Excellence]

Hi Springbok

I will study your post carefully and test your ideas. In the meantime look at the filter below. I have 3 of them going from 20, 25 ,30 skip cycles. They start at current skip of 2 and seek out x skip cycles of 6 or less than 6. This filter can perform flawlessly. In other words the highest rated pair hits within 4 draws and normally in the next 2 draws. Not only that all the pairs fall within the 4 draw parameter. This totally thrashes random play.

Well, I have ended up with the next conclusion: ANY strategy that is based only on numbers/pairs skips will not beat random play, nevah!

Your new filter is not much different then the other one. Will gonna follow the same logic. What is the probability that a pair has a skip bellow or equal to 6? 88,72%. How many pairs will have a skip cycle of 20 skips under 7? (88,72% ^ 20) * 1176 = 107,446. How many pairs will have the next skip under or equal to 2? (26,78% + 19,61% + 14,35%) * 107,446 = 65,278. So we should obtain, based to your filter (20 skips taken into consideration) around 42,167 pairs. In the next 4 draws we should expect that 71,266% of this pairs will be drawn, which is 30,051. I believe anyone knows how those probabilities are obtained so I won’t stress anyone with that.

If u can prove to me, live, that this filter is better then random play over a sufficient amount of draws (71,266% of success IN OVERRAL, just based on this filter) I will eat my winter coat, and hat.

BTW, the closed skip cycles don’t have results that beat random play. In my previous example, we should expect 3 pairs and 1 of them to fail. I am not sure about that, I want someone to prove me wrong.

Again, if someone will make a strategy based ONLY on numbers/pairs skips will obtain results as good as random play.

1 more thing. If we substitute real draws with random generated ones and obtain the same results (or better), doesn’t this means that past draws do not influence current draws and that your strategy is just in fact another criterion of choosing numbers randomly ? Theory of probabilities is the law of every choice that is based on a certain criterion.

All the best!

 

[taaroa]

I think the best way and the simplest way for consistency in the 7/49 game of the 49s is using trips and then pairs.It also has the advantage of being mechanical. I use omega trips(1,2,3;4,5,6;7,8,9 ... 46,47,48). Load them and run over the probability scores. Pick the highest score, delete the file and type in the trip. Hit pairs and pick the one with the highest rating. Bet for the expected 4 draw margin. It is not infallible but in the long run it appears to be consistent. I will run a test over the last 100 draws and give the results in due course. I will run it in parallel with another trip for interest.

Bonjour Springbok,

A stop over from time to time is a good thing. I think more and more to stop playing bookie lottery as special conditions due to the place where I live give me bad odds. People who can play for Lunchtime then for Teatime have better odds. If they win at Lunchtime drawing, they can stop playing for teatime. I must play Lunchtime and teatime at once, so, if I win at Lunchtime, I cannot stop playing for Teatime.

I'm studying 649 and Euromillions. I'm also looking at Thunderball.

About cycles. It's clear that lotteries have a cyclic behaviour. Whatever method you use, it will give good results then go erratic and later give good results again. It depends on what lotto you are playing. I think you are aware of that.

If you look at MDIE or Bright5 or 6, the Any6 filter shows these cyclic behaviour. There are regular intervals between drawings with small standard deviation when Any6 is low. Ion made us aware of that. So, instead of playing every draw, I think that a better method would be to wait for the favorable time to play. We can save a lot of money.

Now, we can make these conditions happen more often. I use your method with layers. Take the first 12 to 14 layers, eliminate doubles and create triples and test them in the filter window. Then take the better numbers and create lines for lotto 5 or this. Run MDIE or Pick632. Bright6 have bizarre reports. If you set the good filters, you can get good lines to purge and play.

By the way, when you want to load a file to purge, what must me the file format? I tried with Excel plain or CSV. It doesn't work.

Thank you.

taaroa

 

[Springbok]

Kunjani Excellence

I hear what you are saying. We must always evaluate filters against random expectations. We start off with a priori assumptions and then test these assumptions empirically so we don't make fools of ourselves by living in a fantasy land. Early this morning I tested the method I suggested to Taaroa on the last 50 draws of the 7 number game of the 49s. There were only 4 failures and none of the failures were back to back. What was being tested were 2 a priori assumptions. First was the 4 draw barrier, secondly were the probability evaluations of Sniper1 . Firstly we determined the safety of the required numbers to hit in the next 4 draws using a trip. Once we had the highest rated trip we split the trip into 3 pairs and the pair with the highest evaluation was used . This soundly thrashed random expectations.I will test this when I have time over the last 200 draws for further confirmation.

 

[Springbok]

Bonsoir Taaroa

With the 49s you got to surf it for a max of 4 draws. You cannot get off your surfboard to have a sleep and then rejoin your surfboard. On the other hand have you thought about the line skip feature when you load the data file and only bet on every other draw. Maybe you should test this. I don't use the purge feature of the programme as yet. Maybe Patrick can help you here.

 

[Patrick123]

Bonjour Taaroa,

I'd suggest you select about 7 balls, generate the rows, then save to a file and look at the file in notepad.

Essentially for a 6/49 game, it will expect:
Row#, Ball1, Ball2, Ball3, Ball4, Ball5, Ball6

basically the row number and the 6 balls seperated by a space or a comma.

Regards
Patrick