BP skipper

[HalfBee]

I'm starting this thread to discuss any feedback, problems, suggestions for the BP Skipper program derived from the idea BP posted in the WinnaLotto Sangoma2 thread.

I'll only repost the url if requested. Haven't decided how long to leave the file posted for download, will see how it goes.

 

[ganbatte]

Hi HalfBee

Good idea for the new thread as I'm sure that Patrick and Springbok will have lots more ideas for Winnalotto when they return to the board.

I have downloaded your file and will check it out later.

 

[ganbatte]

Hi BP

... This method finds patterns as a whole for all numbers to eliminate, but now I'm trying to determine patterns for each individual number instead. This means each number forms it's own unique patterns and if I can get patterns least used for each number, I can eliminate even more numbers based on their own unique patterns. This seems to be more difficult to program, but I want to find if it works.
-BP

I suddenly realised this morning that some of the lowest patterns (eg. number 19) were ready to be drawn according to Springbok's Sniper 1 technique from Winnalotto so I am looking at incorporating that or at least the ball median into the calculations to see when it's due.

Could you tell me what you mean by "patterns for each individual number" as the existing programme works out skip patterns for each individual number in each draw, unless you mean tracking each number across many draws to determine a pattern?

 

[black prince]

ganbatte

Originally Posted by ganbatte
Could you tell me what you mean by "patterns for each individual number" as the existing programme works out skip patterns for each individual number in each draw, unless you mean tracking each number across many draws to determine a pattern?

The method you are currently using takes all patterns prior to winning and
determines which patterns are least used for the GROUP. That seems to work
75% and greater each time. I am experminting with the patterns for each
individual number to see if it has favoriate patterns it likes to use for short
periods of draws. Of course patterns due change, but will individual numbers
repeat certain groups of patterns? What can we do with this idea? Suppose
the number 1 has these patterns (55655,56556,65555,65655,66565,56655
and 65656) in the last 30 draws and was a winning number. Was there the
same group of patterns in draws 31 to 60, and so on. If there is any
corrollation of patterns this would allow me to look at 1's current pattern to
see if there is one of the patterns it usually wins with. I hope this makes
some sense. This is a backtest project that has to be programmed. I'm just
finishing the first program that gathers patterns for each number.

-BP

 

[HalfBee]

I think I'm following what your after. But I think 30 draws is a bit limiting as a number might only have 1 or 2 patterns in that span, whereas other numbers may have 10 or more. Will have to think about it some more...

 

[black prince]

HalfBee

Originally Posted by HalfBee
I think I'm following what your after. But I think 30 draws is a bit limiting as a number might only have 1 or 2 patterns in that span, whereas other numbers may have 10 or more. Will have to think about it some more...

Your right about the range of 30. I will expand this to see what happens. My
fear is that when you go back too far in history, I lose the current trends
by mixing current with past. The only way to get the sweet spot is comparing
various ranges that produce patterns that are winners consistantly.

-BP

 

[ganbatte]

Hi BP

The method you are currently using takes all patterns prior to winning and
determines which patterns are least used for the GROUP. That seems to work
75% and greater each time. I am experminting with the patterns for each
individual number to see if it has favoriate patterns it likes to use for short
periods of draws. Of course patterns due change, but will individual numbers
repeat certain groups of patterns? What can we do with this idea? Suppose
the number 1 has these patterns (55655,56556,65555,65655,66565,56655
and 65656) in the last 30 draws and was a winning number. Was there the
same group of patterns in draws 31 to 60, and so on. If there is any
corrollation of patterns this would allow me to look at 1's current pattern to
see if there is one of the patterns it usually wins with. I hope this makes
some sense. This is a backtest project that has to be programmed. I'm just
finishing the first program that gathers patterns for each number.
-BP

Now I understand why I was confused.

When I build my Excel version I was automatically calculating the skips and patterns of each number and for numbers falling into the same pattern group .

The following example from my spreadsheet for number 13 (5 when skip<6 and 6 when skip>6):

Draw...........Skips..................Pattern............Number of other winning balls with same pattern
01...............1,2,9,1,7..............5,5,6,5,6...........12 wins in 40 draws (30%)
03...............2,9,1,7,7..............5,6,5,6,6...........14 wins in 40 draws (35%)
16...............1,7,7,8,5..............5,6,6,6,5............5 wins in 40 draws (12.5%)
18...............7,7,8,5,14............6,6,6,5,6............6 wins in 40 draws (15%)
26...............7,8,5,14,38...........6,6,5,6,6............6 wins in 40 draws (15%)
34...............8,5,14,38,12.........6,5,6,6,6............6 wins in 40 draws (15%)
43...............5,14,38,12,8.........5,6,6,6,6............2 wins in 40 draws (5%)
49...............14,38,12,8,2.........6,6,6,6,5............2 wins in 40 draws (5%)
64...............38,12,8,2,2...........6,6,6,5,5............3 wins in 40 draws (7.5%)

(Draw 01 is the most recent on 9 Aug 2008).

As you can see there is no discernible "pattern to the pattern" for number 13 in the past 64 draws when it hit 9 times.

 

[bloubul]

Now I understand why I was confused.

When I build my Excel version I was automatically calculating the skips and patterns of each number and for numbers falling into the same pattern group .

The following example from my spreadsheet for number 13 (5 when skip<6 and 6 when skip>6):

(Draw 01 is the most recent on 9 Aug 2008).

As you can see there is no discernible "pattern to the pattern" for number 13 in the past 64 draws when it hit 9 times

The following example from my spreadsheet for number 13 (5 when skip<6 and 6 when skip>6):

Quote:
Draw...........Skips..................Pattern............Number of other winning balls with same pattern
01...............1,2,9,1,7..............5,5,6,5,6...........12 wins in 40 draws (30%)
03...............2,9,1,7,7..............5,6,5,6,6...........14 wins in 40 draws (35%)
16...............1,7,7,8,5..............5,6,6,6,5............5 wins in 40 draws (12.5%)
18...............7,7,8,5,14............6,6,6,5,6............6 wins in 40 draws (15%)
26...............7,8,5,14,38...........6,6,5,6,6............6 wins in 40 draws (15%)
34...............8,5,14,38,12.........6,5,6,6,6............6 wins in 40 draws (15%)
43...............5,14,38,12,8.........5,6,6,6,6............2 wins in 40 draws (5%)
49...............14,38,12,8,2.........6,6,6,6,5............2 wins in 40 draws (5%)
64...............38,12,8,2,2...........6,6,6,5,5............3 wins in 40 draws (7.5%)


(Draw 01 is the most recent on 9 Aug 2008).


ganbatte

Verduidelik dit als in Afrikaans vir my asb.

BlouBul

 

[ganbatte]

Bloubul

It's difficult enough explaining some things in English without having to try in Afrikaans. I also don't know if I'm right so am waiting on BP & HalfBee.

 

[black prince]

ganbatte

Originally Posted by ganbatte
It's difficult enough explaining some things in English without having to try in Afrikaans. I also don't know if I'm right so am waiting on BP & HalfBee.

I hope I can help clear what I'm doing. The first programs collects the
skip patterns for all winning draws before it wins. Then the second program
creates patterns for all numbers as a group, not each single number. So,
if a pattern only has 1, 2, 3, or 4 hits, this means any number. In other
words, the chances for any number for the low pattern hits should be
eliminated.

What I'm curently experminating with is what patterns are individual numbers
using most often as a winners. It's just expermental, but if there is a use
for this approach I will let you know.

I believe winning draw history can be extremely helpful to predict future
winning draws. It takes time understand draw history. When I find that
something works 75% or more, It's added to my filter list until I reach the
lowest remaining numbers keeping all winning numbers.

-BP

 

[HalfBee]

I believe winning draw history can be extremely helpful to predict future winning draws. It takes time understand draw history. When I find that something works 75% or more, It's added to my filter list until I reach the lowest remaining numbers keeping all winning numbers.

-BP

I also believe in draw history and patterns, but am looking at it a bit differently. What I notice is that the Hit Frequency (how often a number hits over other numbers) has a cycle or pattern. I see groupings every 10-20 draws that the numbers all seem to gather within about 10-15 of each other as ranked by frequency. Sometimes they are at the upper end, sometimes lower, and sometimes near the middle... but they all fall within a given range.

Most of the time they are fairly evenly distributed over the entire list, but I'm trying to find a pointer indicating they are about to group. This method might not work for 'regular' lottos like 6/49 and such, or games that have actual balls drawn from a machine, but for the 5/39 game it keeps showing up.

Perhaps skip values can help indicate when they are about to group, or help show where the upper/lower bounds of the right range is located. Then again, I just like programs that logically slice and dice yet have the flexability to let me experiment with different methods toward the same goal.

 

[HalfBee]

The following example from my spreadsheet for number 13 (5 when skip<6 and 6 when skip>6):

My patterns are exactly opposite (6 if skip<06 and 5 if skip>06) but that's just the way I read BP's code listing. I was using 1 & 8 to indicate under/over (1 as skip<06 and 8 for skips>06) the skip values to evaluate them and found it a bit easier to know what was low and high. i.e. 11818 or such.

The patterns are tricky, once a number hits it remains Lxxxx until it passes the skip value then it becomes Hxxxx. If it hits twice within that skip range it stays LLxxx for just as long. But once it passes the skip value it won't change no matter how many draws pass until it hits again, so a pattern HLLLL might hit in one draw or in 30, without an indication that's predictable. It's the more short range patterns which probably only need 3 skips toward the low end.

 

[black prince]

HalfBee

Originally Posted by HalfBee
I also believe in draw history and patterns, but am looking at it a bit differently. What I notice is that the Hit Frequency (how often a number hits over other numbers) has a cycle or pattern. I see groupings every 10-20 draws that the numbers all seem to gather within about 10-15 of each other as ranked by frequency. Sometimes they are at the upper end, sometimes lower, and sometimes near the middle... but they all fall within a given range.

Most of the time they are fairly evenly distributed over the entire list, but I'm trying to find a pointer indicating they are about to group. This method might not work for 'regular' lottos like 6/49 and such, or games that have actual balls drawn from a machine, but for the 5/39 game it keeps showing up.

Perhaps skip values can help indicate when they are about to group, or help show where the upper/lower bounds of the right range is located. Then again, I just like programs that logically slice and dice yet have the flexability to let me experiment with different methods toward the same goal.

Sounds like BushHappy's method. I'm also interested in your method and
look forward to trying your ideas. I hope everyone realizes that
any improvements to my ideas (patterns) and others a welcomed. We
learn from each other.

-BP

 

[bloubul]

Hi HalfBee

If I upload my Lotto database for you, will you help me with 12 numbers for Saterday August 16 2008. "My patterns are exactly opposite"

BlouBul

 

[HalfBee]

Hi HalfBee

If I upload my Lotto database for you, will you help me with 12 numbers for Saterday August 16 2008. "My patterns are exactly opposite"

BlouBul

Hmm... what size is the lotto? Is it the 6/49?
My number pool for 5/39 is usually 18 numbers and even then I only get 3 or 4 in the pool about every 2nd or 3rd draw. I do get all 5 in the 18 pool about once a week, but there are some weird draws at times. Like today, there were two historical highs that both hit. At 30+ skips each they weren't expected (but were in the pool because a couple filters had them).

Your patterns being opposite isn't a problem if you read them that way. The pattern is just there to visually show you what patterns are popular with the winning numbers. I like larger pools with key numbers, it's those that I'm working hard to predict. If you look at the current winning patterns there's usually one that will match a 3 or 4 number group and will show up in the next draw. It's guessing which pattern to use. Most of the time it's the under skip limits in positions 1 and 2. But then these are usually thrown up by Hot & Very Hot filters in WinnaLotto anyways...

 

[ganbatte]

Hi BP & HalfBee

I don't know if you guys have heard of the EuroMillions lottery but I have been trying to work out good patterns for the £42,000,000 or $78,000,000 tax free draw this Friday evening.

If you guys feel like it, could you possibly have a go at predicting using BP's patterns? I have also been following HalfBee's results for the California Fantasy 5 and hitting 3 out of 5 balls for 8/10, 8/11 and 8/13 is great.

The data for the lottery can be downloaded at: http://www.national-lottery.co.uk/player/files/EuroMillions.csv. To win you need 5 balls from 1 to 50 and 2 "lucky star" balls from 1 to 9.

I tried the patterns for the draw last week but wasn't too successful. Maybe you guys will do much better.

 

[black prince]

HalfBee

Originally Posted by HalfBee
I also believe in draw history and patterns, but am looking at it a bit differently. What I notice is that the Hit Frequency (how often a number hits over other numbers) has a cycle or pattern. I see groupings every 10-20 draws that the numbers all seem to gather within about 10-15 of each other as ranked by frequency. Sometimes they are at the upper end, sometimes lower, and sometimes near the middle... but they all fall within a given range.

I noticed a certain pattern which may have the chararistics of what your
referring to. There are number(s) which have all hits less than 6 in the last
5 skips in which the skips added up are <= 30. For example, the # 28 -
has skips 0,2,5,1,4. I noticed very few numbers won with more than 5 or 6
low skips and the skips total <= 30. They seemed to have been
exhausted or appeared beyond their max? This is true for numbers with skips
where one or two skips are greater than 6, but the skips totaled are still
<=30. They reappeared when the skips were greater than 30. I looking
at this further with backtesting as usual.

-BP

 

[HalfBee]

ganbatte,

Do the 'Lucky Stars' have to be in the proper order too? And can they be doubled i.e. 99 22 etc or are they just 2 balls from the same set of 9?
These 'extra' numbers are just a popular way the lotto folks found to raise the odds to get higher jackpots. Like 5/50 isn't high enough already.

I did download that file last week when testing my program, it's a little out of date now but only by a couple draws. Those postings for Fantasy 5 are based on WinnaLotto and a bit of tweaking based on last 5 draws and draws 5 to 10 draw frequency. Haven't put any of the pattern stuff to use, is still in the 'testing' phase. I'll probably release a new version of the program I posted that adds skips 1-5 for an <= 30 => indicator to see if it has any bearing on possible combinations. I'll probably make it a threshold setting since a 5/39 game and a 5/50 game would have different numbers (like median skip value x 5 or such).

 

[ganbatte]

HalfBee

The lucky Stars balls have to be two balls from the same set of nine eg. 1&3 or 2&8.

The extra numbers definitely make it much harder than a normal lotto.

 

[ganbatte]

HalfBee

Most people run analysis for the Lucky Star balls separately as they are drawn from a separate set of balls.

The trouble with EuroMillions is that you definitely need all 5 main balls to even start with a chance for the big bucks.